BR122021003390A2 - AUDIO CODER, AUDIO DECODER, METHOD OF CODING AND AUDIO INFORMATION, METHOD OF DECODING AUDIO INFORMATION AND COMPUTER PROGRAM USING AN OPTIMIZED HASH TABLE - Google Patents

Abstract

AUDIO CODER, AUDIO DECODER, METHOD OF CODING AND AUDIO INFORMATION, METHOD OF DECODING AUDIO INFORMATION AND COMPUTER PROGRAM USING AN OPTIMIZED HASH TABLE An audio decoder for providing decoded audio information based on information encoded audio source comprises an arithmetic decoder for providing a plurality of decoded spectral values based on an arithmetically encoded representation of the spectral values, and a frequency domain to time domain converter for providing a time domain audio representation using the decoded spectral values. , to obtain the decoded audio information.

Description

[0001] By transforming a windowed portion of the input audio signal from the time domain to the time frequency domain, an energy compression is achieved in many cases such that some of the spectral values comprise a magnitude significantly greater than several other spectral values. Thus, there are, in many cases, a comparably small number of spectral values having a magnitude that is significantly above an average magnitude of the spectral values. A typical example of a time domain to frequency domain transformation resulting in an energy compression is called modified discrete cosine transformation (MDCT).

[0002] Spectral values are generally scaled and quantized according to a psychoacoustic model, so that quantization errors are comparably smaller for psychoacoustically more important spectral values, and are comparably larger for psychoacoustically less important spectral values. The scaled and quantized spectral values are encoded to provide an efficient representation of their bitrate.

[0003] For example, the use of so-called Huffman coding of quantized spectral coefficients is described in International Standard ISO/IEC 14496-3:2005(E), part 3, subpart 4.

[0004] However, it has been observed that the quality of encoding the spectral values has a significant impact on the required bit rate. Furthermore, it was observed that the complexity of an audio decoder, which is generally implemented in a portable consumer device, and which must then be created cheaply and with low power consumption, is dependent on the encoding used in encoding the spectral values. .

[0005] In view of this situation, there is a need for a concept for encoding and decoding audio content, which provides an improved trade-off between bitrate efficiency and resource efficiency.

[0006] SUMMARY OF THE INVENTION

[0007] An embodiment according to the invention creates an audio decoder to provide several decoded spectral values based on an arithmetically encoded representation of the spectral values. The audio decoder also comprises a frequency domain to time domain converter to provide a time domain audio representation using the decoded spectral values to obtain the decoded audio information. The arithmetic decoder is configured to select a mapping rule that describes a mapping of a code value representing a spectral value, or a most significant bit plane of a spectral value, in encoded form, into a symbol code representing a value spectral, or a most significant bit plane of a spectral value, in a decoded form, depending on a context state described by a current numeric context value. The arithmetic decoder is configured to determine the current numeric context value depending on the various previously decoded spectral values. The arithmetic decoder is configured to evaluate the hash table, whose entries define both the significant state values between the numeric context values and the range boundaries of the numeric context values, to select the mapping rule. The arithmetic decoder is configured to evaluate the hash table to find an index value of hash table i in which the value ari_hash_m[i]>>8 is equal to or greater than c, whereas if the index value of hash table i found is greater than 0, the value ari_hash_m[i-1]>>8 is less than c. Additionally, the arithmetic decoder is configured to select the mapping rule that is determined by a probability model index (pki) that is equal to ari_hash_m[i]&&0xFF when ari_hash_m[i]>>8 is equal to c, or, otherwise, equal to ari_lookup_m[i]. In the present embodiment, the hash table ari_hash_m is defined as shown in figures 22(1), 22(2), 22(3) and 22(4). Furthermore, the ari lookup m mapping table is defined as shown in figure 21.

[0008] It has been observed that the combination of the aforementioned algorithm with the hash table of figures 22(1) to 22(4) allows a particularly efficient selection of a mapping rule, such as the hash table according to figure 22(1) 22(4) defines, particularly well, both significant values of the numeric context value and state intervals. Furthermore, the interaction between said algorithm and the hash table according to figures 22(1) to 22(4) has been shown to bring particularly good results while keeping the computational complexity reasonably small. Furthermore, the mapping table defined in figure 21 is also particularly well adapted to said algorithm when in combination with the hash table mentioned above. To summarize, the use of the hash table as shown in figures 22(1) to 22(4) and the mapping table as defined in figure 22 in connection with the algorithm as described above brings along good encoding/decoding efficiency. and low computational complexity.

[0009] In a preferred embodiment, the arithmetic decoder is configured to evaluate the hash table using the algorithm as defined in Figure 5e, wherein c designates a variable representing the current numeric context value or a scaled version thereof, wherein i is a variable describing a current hash table index value, where in i_min is a variable initialed to designate a hash table index value of a first hash table entry and selectively updated depending on a comparison between c and (j >> 8). In the algorithm mentioned above, a condition “c<(j>>8)” defines that a state value described by the variable c is smaller than a state value described by the ari_hash_m[i] table entry. Furthermore, in the algorithm mentioned above, “j&0xFF” describes a mapping rule index value described by the ari_hash_m[i] table entry. Still i_max is a variable initialized to designate a hash table index value of a last hash table entry an selectively updated depending on a comparison between c and (j>>8). A condition “c>(j>>8)” defines that a state value described by the variable c is greater than a state value described by the ari_hash_m[i] table entry. The return value of said algorithm designates an index pki of a probability model and is an index value of the mapping rule. “ari_hash_m” designates the hash table, and “ari_hash_m[i]” designates a hash table entry ari_hash-m having hash table index value i. “ari_lookup_m” designates a mapping table, and “ari_lookup_m[i_max]” designates an entry of the ari_lookup_m mapping table having the mapping index value i_max.

[00010] It was observed that the combination of the algorithm mentioned above, as shown in Figure 5e, with the hash table of Figures 22(1) to 22(4) allows a particularly efficient selection of a mapping rule, such as the hash according to with figure 22(1) to 22(4) define, in a particularly well-suited way, both significant values of the numerical context value and state intervals. Furthermore, the interaction between said algorithm according to Figure 5e and the hash table according to Figures 22(1) to 22(4) has been shown to bring particularly good results in combination with a fast algorithm for searching the table. . Furthermore, the mapping table defined in figure 21 is also particularly well adapted to said algorithm when in combination with the hash table mentioned above. To summarize, the use of the hash table as shown in figures 22(1) to 22(4) and the mapping table as defined in figure 22 in connection with the algorithm as defined in figure 5e brings good encoding/decoding efficiency and low computational complexity. In other words, it was observed that the two-section algorithm in Figure 5e is well suited to operate with the ari_hash_m and ari_lookup_m tables, as defined above.

[00011] However, it should be noted that small changes (which are easily practicable) or even significant changes to the search algorithm can occur without changing the concept according to the present invention.

[00012] In other words, the research method is not restricted to the mentioned methods. Although the use of the two-section method (for example, according to Figure 5e) still improves performance, it would also be possible to perform a simple complete search, with, however, bringing a certain increase in complexity.

[00013] In a preferred embodiment, the arithmetic decoder is configured to select the mapping rule that describes a mapping of a code value into a symbol code based on the mapping rule index value pki, which is e.g. , provided as a return value from the algorithm shown in Figure 5e. The use of said pki mapping rule index value is very efficient, as the interaction of the above-mentioned tables and the above-mentioned algorithm is optimized to provide a meaningful mapping rule index value.

[00014] In a preferred embodiment, the arithmetic decoder is configured to use the index value of the mapping rule as a table index value to select the mapping rule that describes a mapping of a code value into a symbol code . Using the mapping rule index value as a table index value enables computational efficiency in memory-efficient selection of the mapping rule.

[00015] In a preferred embodiment, the arithmetic decoder is configured to select one of the subtables of the ari_cf_m[64][17] table, as defined in figure 23(1), 23(2), 23(3), as the rule mapping selected. This concept is based on the fact that the mapping rules defined by the subtables of the ari_cf_m[64][17] table, as defined in figure 23(1),(2), (3), are well adapted to the results that can be achieved by executing the above-mentioned algorithm according to figure 5e in combination with the table according to figures 21 and 22(1) to 22(4).

[00016] In a preferred embodiment, the arithmetic decoder is configured to obtain the numeric context value based on a prior numeric context value using an algorithm according to Figure 5c, wherein the algorithm receives, as input values, a value of a variable c representing a numeric prior context value, a value or a variable i representing an index of double tuples of the spectral values to decode into a vector of the spectral values. A value or variable N represents a window length of a reconstruction window of the frequency domain to time domain converter. The algorithm provides, as an output value, an updated value or variable c representing the current numeric context value. In the algorithm, a “c>>4” operation describes a right shift by 4 bits of the value or variable c. Furthermore, q[0][i+1] designates a context subregion value associated with a past audio structure and associated with a higher frequency index i+1, higher than 1, than a present frequency index of one of the two total spectral values currently being decoded. Similarly q[1][i-1] designates a context subregion value associated with a present audio structure and associated with a minor frequency index i-1, less than 1, then a present frequency index of double tuples of the spectral values currently being decoded. q[1][i-2] designates a context subregion value associated with a present audio structure and associated with the lower frequency index i-2, less than 2, than a tuple present frequency index doubles of the spectral values currently being decoded. q[1][i-3] designates a context subregion value associated with the present audio structure and associated with a lower frequency index i-3, less than 3, than a present frequency index of double tuples of the spectral values currently being decoded. It was observed that the algorithm according to figure 5e when in combination with the tables of figure 21 and 22(1) to 22(4) is well adapted to provide the index value of the mapping rule based on a context value current numeric c obtained using the algorithm of figure 5c, where obtaining the current numeric context value using the algorithm of figure 5c is particularly computationally efficient, as the algorithm according to figure 5c requires only a very simple calculation.

[00017] In a preferred embodiment, the arithmetic decoder is configured to update a value of the context subregion q[1][i] associated with a present audio structure and associated with the present frequency index of the double tuples of the spectral values currently decoded using an algorithm according to Figure 5l, wherein a designates an absolute value of a first spectral value of the double tuples of the currently decoded spectral values, and where b designates a second spectral value of the double tuples of the currently decoded spectral values . It can be seen that the preferred algorithm is very well suited for a simple update of the context subregion value.

[00018] In a preferred embodiment, the arithmetic decoder is configured to provide a decoded value m representing double tuples of the decoded spectral values using the arithmetic decoding algorithm according to Figure 5g. It has been observed that said arithmetic decoding algorithm is very well suited for cooperation with the algorithms described above.

[00019] Another embodiment according to the invention creates the decoder to provide decoded audio information based on encoded audio information. The audio decoder comprises an arithmetic decoder for providing a plurality of decoded spectral values based on an arithmetically encoded representation of the spectral values. The audio decoder also comprises a frequency domain to time domain converter to provide a time domain audio representation using the decoded spectral values to obtain the decoded audio information. The arithmetic decoder is configured to select a mapping rule that describes a mapping of a code value representing a spectral value, or a most significant bit plane of a spectral value, in encoded form, into a symbol code representing a value spectral, or a most significant bit plane of a spectral value, in a decoded form, depending on a context state described by a current numeric context value. The arithmetic decoder is configured to determine the current numeric context value depending on the various previously decoded spectral values. The arithmetic decoder is configured to evaluate the hash table, whose entries define both the significant state values between the numeric context values and the range boundaries of the numeric context values, to select the mapping rule. The hash table ari hash m is defined as shown in figure 22(1), 22(2), 22(3) and 22(4). The arithmetic decoder is configured to evaluate the hash table to determine whether the current numeric context value is identical to a table context value described by a hash table entry or to determine a range described by the hash table entries in which the value current numeric context is found, and to derive a mapping rule index value that describes a selected mapping rule depending on a result of an evaluation. It has been observed that the hash table ari hash m given in figures 22(1) to 22(4) is well suited for analyzing the table context values described by the hash table entries and ranges described by the hash table entries, for thus derive the value of the mapping index. It was observed that defining both table context values and ranges by the hash table according to Figure 22(1) to 22(4) provides an efficient mechanism for mapping rule selection when in combination with a simple concept for evaluation of the hash table that uses entries of said hash table to both verify the context values of the table and to determine in which range defined by the hash table entries of the context values of the non-table it lies.

[00020] In a preferred embodiment, the arithmetic decoder is configured to compare the current numeric context value, or a scaled version of the current numeric context value, with a series of numerically ordered entries or subentries of the hash table, to iteratively obtain a hash table index value of a table entry, such that the current numeric context value lies within a range defined by the obtained hash table entry designated by the obtained hash table index value and an adjacent hash table entry . In this case, the arithmetic decoder is configured to determine a next entry in the series of hash table entries depending on the result of a comparison between the current numeric context value, or a scaled version of the current numeric context value, and an input or current sub-input. It was recognized that this mechanism allows a particularly efficient hash table evaluation according to figure 22(1) to 22(4).

[00021] In a preferred embodiment, the arithmetic decoder is configured to select a mapping rule defined by a second sub-entry of the hash table designated by the current index value of the hash table if it is observed that the current numeric context value or a scaled version of this is equal to the first sub-entry of the hash table designated by the current index value of the hash table. Of course, hash table entries as defined according to Figures 22(1) to 22(4) take on a dual function. A first sub-entry (i.e., a first part of an entry) of the hash table is used to identify particularly significant states of the numeric (current) context value, while a second sub-entry of the hash table (i.e., a second part of such an entry) defines a mapping rule, for example, defining an index value of the mapping rule. Thus, the hash table entries are used very efficiently. Furthermore, the mechanism is particularly efficient in providing the index value of the mapping rules for the particularly important states of the current numeric context values, which are described by the hash table entries, or, more precisely, by the hash table sub-entries. . Thus, a complete hash table entry, as defined in Figures 22(1) to 22(4), defines rules for a mapping of a particularly important state of the numeric (current) context value to a mapping rule and range boundaries. of the regions (or ranges) of less important states of the current numeric context value.

[00022] In a preferred embodiment, the arithmetic decoder is configured to select a mapping rule defined by an entry or sub-entry of an ari_lookup_m mapping table if it is not observed that the current numeric context value is equal to a sub- hash table entry. In this case, the arithmetic decoder is configured to choose the entry or sub-entry of the mapping table depending on the iteratively obtained hash table index value. Thus, a particularly efficient two-table mechanism is created, allowing to efficiently provide a mapping rule index value for both particularly important states of the numeric current context value and for less important states of the numeric current context value (where the less important states of the current numeric context value are not explicitly, i.e. individually, described by the entries or sub-entries of the hash table).

[00023] In a preferred embodiment, the arithmetic decoder is configured to selectively provide a mapping rule index value defined by the hash table entry designated by the index value obtained from the hash table if it is observed that the current numeric context value is equal to the value defined by the hash table entry designated by the current index value of the hash table. Thus, there is an efficient mechanism that allows double use of hash table entries.

[00024] Other embodiments of the invention create methods for providing decoded audio information based on encoded audio information. Said methods fulfill the functionality of the audio decoders discussed above. Of course, the methods are based on the same ideas and thoughts as audio decoders, so a discussion is omitted here for brevity. It should be noted that the methods can be supplemented by any of the features and functionalities of the audio decoders.

[00025] Another embodiment according to the invention creates an audio decoder to provide encoded audio information based on input audio information. The audio encoder comprises an energy-compressed time-domain-to-frequency-domain converter for providing a frequency-domain audio representation based on a time-domain representation of the input audio information, such that the representation frequency domain audio signal comprises a set of spectral values. The audio encoder also comprises an arithmetic encoder configured to encode a spectral value or a preprocessed version thereof using a variable length code. The arithmetic encoder is configured to map a spectral value, or a value from a most significant bit plane of a spectral value, into a code value. The arithmetic encoder is also configured to select a mapping rule that describes a mapping of a spectral value, or a most significant bit plane of a spectral value, into a code value, depending on a context state described by a value of numeric current context. The arithmetic encoder is configured to determine the current numeric context value depending on the various previously encoded spectral values. The arithmetic encoder is also configured to evaluate a hash table, whose entries define both the significant state values among the numeric context values and the range boundaries of the numeric context values, to select the mapping rule. The hash table ari_hash_m is defined as shown in figures 22(1) to 22(4). The arithmetic encoder is configured to evaluate the hash table to determine whether the current numeric context value is identical to a table context value described by a hash table entry or to determine a range described by the hash table entries in which the value current numeric context is found, and to derive a mapping rule index value that describes a selected mapping rule depending on a result of said evaluation. It should be noted that the audio encoder functionality is in parallel with the audio decoder functionality discussed above. Of course, reference is made to the above discussion of the main ideas of the audio decoder for the sake of brevity.

[00026] Furthermore, it should be noted that the audio encoder can be supplemented by any of the features and functionalities of the audio decoder. In particular, any of the features relating to mapping rule selection can be implemented in the audio encoder as well, whereby the encoded spectral values take the place of the decoded spectral values, and so on.

[00027] Another embodiment according to the invention creates a method for providing encoded audio information based on input audio information. The method realizes the functionality of the audio encoder described before based on the same ideas.

[00028] Another embodiment according to the invention creates a computer program to perform at least one of the methods described above.

[00029] BRIEF DESCRIPTION OF THE FIGURES

[00030] The embodiments according to the present invention will subsequently be described with reference to the attached figures, in which:

[00031] Figure 1 shows a schematic block diagram of an audio encoder, according to an embodiment of the invention;

[00032] Figure 2 shows a schematic block diagram of an audio decoder, according to an embodiment of the invention:

[00033] Figure 3 shows a code representation of the pseudo-program of a “values_decode()” algorithm for decoding spectral values;

[00034] Figure 4 shows a schematic representation of a context for a state calculation;

[00035] Figure 5a shows a code representation of the pseudo-program of an “arith_map_context()” algorithm for mapping a context;

[00036] Figure 5b shows a code representation of the pseudo-program of another algorithm “arith_map_context()” for mapping a context;

[00037] Figure 5c shows a pseudo-program code representation of an “arith_get_context()” algorithm to obtain a context state value;

[00038] Figure 5d shows a pseudo-program code representation of another algorithm “arith_get_context()” to obtain a context state value;

[00039] Figure 5e shows a code representation of the pseudo-program of an algorithm “arith_get_pk()” for deriving an index value of the cumulative frequency table “pki” from a state value (or a state variable);

[00040] Figure 5f shows a code representation of the pseudo-program of another algorithm “arith_get_pk()” for deriving an index value of the cumulative frequency table “pki” from a state value (or a state variable);

[00041] Figure 5g shows a pseudo-program code representation of an “arith_decode()” algorithm for arithmetically decoding a symbol from a variable length code;

[00042] Figure 5h shows a first part of a pseudo-program code representation of another algorithm “arith_decode()” for arithmetically decoding a symbol from a variable length code;

[00043] Figure 5i shows a second part of a pseudo-program code representation of the other algorithm “arith_decode()” for arithmetically decoding a symbol from a variable length code;

[00044] Figure 5j shows a code representation of the pseudo-program of an algorithm for deriving the absolute values a, b from the spectral values of a common value m;

[00045] Figure 5k shows a code representation of the pseudo-program of an algorithm for inserting the decoded values a, b into a matrix of decoded spectral values;

[00046] Figure 5l shows a code representation of the pseudo-program of an algorithm “arith_update_context()” to obtain a value of the context subregion based on the absolute values a, b of the decoded spectral values;

[00047] Figure 5m shows a pseudo-program code representation of an “arith_finish()” algorithm for filling the entries of a matrix of decoded spectral values and a matrix of context subregion values;

[00048] Figure 5n shows a code representation of the pseudo-program of another algorithm for deriving absolute values a,b from spectral values decoded from a common value m;

[00049] Figure 5o shows a pseudo-program code representation of an “arith_update_context()” algorithm for updating an array of decoded spectral values and an array of context subregion values;

[00050] Figure 5p shows a pseudo-program code representation of an “arith_save_context()” algorithm for filling entries of a matrix of decoded spectral values and entries of a matrix of context subregion values;

[00051] Figure 5q shows a legend of the definitions;

[00052] Figure 5r shows another legend of the definitions;

[00053] Figure 6a shows a syntax representation of a Unified Speech and Audio Coding (USAC) raw data block;

[00054] Figure 6b shows a syntax representation of a single channel element;

[00055] Figure 6c shows a syntax representation of an element of the channel pair;

[00056] Figure 6d shows a syntax representation of an “ICS” control information;

[00057] Figure 6e shows a syntax representation of a frequency domain channel flow;

[00058] Figure 6f shows a syntax representation of arithmetically encoded spectral data;

[00059] Figure 6g shows a syntax representation for decoding a set of spectral values;

[00060] Figure 6h shows another syntax representation for decoding a set of spectral values;

[00061] Figure 6i shows a legend of the data elements and variables;

[00062] Figure 6j shows another legend of data elements and variables;

[00063] Figure 6k shows a syntax representation of a single USAC channel element “UsacSingleChannelElement()”;

[00064] Figure 6l shows a syntax representation of a USAC channel pair Element “UsacChannelPairElement()”;

[00065] Figure 6m shows a syntax representation of an “ICS” control information;

[00066] Figure 6n shows a syntax representation of the USAC core coder data “UsacCoreCoderData”;

[00067] Figure 6o shows a syntax representation of a frequency domain channel stream “fd_channel_stream()”;

[00068] Figure 6p shows a syntax representation of the arithmetically encoded spectral data “ac_spectral_data()”;

[00069] Figure 7 shows a schematic block diagram of an audio encoder, according to the first aspect of the invention;

[00070] Figure 8 shows a schematic block diagram of an audio decoder, according to the first aspect of the invention;

[00071] Figure 9 shows a graphical representation of a mapping of a numeric current context value into a mapping rule index value, in accordance with the first aspect of the invention;

[00072] Figure 10 shows a schematic block diagram of an audio encoder, according to a second aspect of the invention;

[00073] Figure 11 shows a schematic block diagram of an audio decoder, according to the second aspect of the invention;

[00074] Figure 12 shows a schematic block diagram of an audio encoder, according to a third aspect of the invention;

[00075] Figure 13 shows a schematic block diagram of an audio decoder, according to the third aspect of the invention;

[00076] Figure 14a shows a schematic representation of a context for a state calculation, as used in accordance with working design 4 of the USAC Design Standard;

[00077] Figure 14b shows an overview of the tables as used in the arithmetic coding scheme in accordance with working design 4 of the USAC Design Standard;

[00078] Figure 15a shows a schematic representation of a context for a state calculation, as used in embodiments according to the invention;

[00079] Figure 15b shows an overview of the tables as used in the arithmetic coding scheme according to an example comparison;

[00080] Figure 16a shows a graphical representation of a read-only memory demand for the silent encoding scheme in accordance with a comparison example, and in accordance with working design 5 of the USAC Design Standard, and in accordance with with Huffman AAC Coding (advanced audio coding);

[00081] Figure 16b shows a graphical representation of a memory demand for read-only total USAC decoder data in accordance with a comparison example and in accordance with the concept in accordance with working design 5 of the USAC Design Standard ;

[00082] Figure 17 shows a schematic representation of an arrangement for a comparison of a silent coding in accordance with working design 3 or working design 5 of the USAC Design Standard with a coding scheme in accordance with the comparison example ;

[00083] Figure 18 shows a tabular representation of the average bit rates produced by a USAC arithmetic encoder in accordance with working design 3 of the USAC Design Standard and in accordance with a comparison example;

[00084] Figure 19 shows a table representation of minimum and maximum levels of the bit reservoir for an arithmetic decoder in accordance with working design 3 of the USAC Design Standard and for an arithmetic decoder in accordance with a comparison example;

[00085] Figure 20 shows a tabular representation of the average complexity numbers for decoding a 32-kbit bit stream in accordance with working design 3 of the USAC Design Standard for different versions of the arithmetic encoder;

[00086] Figure 21 shows a table representation of a content of a table “ari_lookup_m[742]”, according to an embodiment of the invention;

[00087] Figures 22(1) to 22(4) show a table representation of a content of a table “ari_hash_m[742]”, according to an embodiment of the invention;

[00088] Figures 23(1) to 23(3) show a table representation of a content of a table “ari_cf_m[64][17]”, according to an embodiment of the invention; It is

[00089] Figure 24 shows a table representation of a content of a table “ari_cf_r[]”;

[00090] Figure 25 shows a schematic representation of a context for a state calculation;

[00091] Figure 26 shows a tabular representation of an average coding performance for transcoding the WD6 reference quality bit streams for a comparison example (“M17558”) and for an embodiment according to the invention (“New Proposal");

[00092] Figure 27 shows a table representation of a coding performance for transcoding the WD6 reference quality bit streams per operating point for a comparison example (“M17558”) and for an embodiment according to the invention ( “Retrained tables”)

[00093] Figure 28 shows a table representation of a comparison of the Silent Encoder Memory Demand for WD6, for a comparison example (“M17588”) and for an embodiment according to the invention (“New Proposal”);

[00094] Figure 29 shows a tabular representation of the characteristics of tables as used in an embodiment according to the invention (“Retrained coding scheme”);

[00095] Figure 30 shows a table representation of numbers of the average complexity for decoding the WD6 reference quality 32kbit bit streams for the different versions of the arithmetic encoder;

[00096] Figure 31 shows a table representation of numbers of the average complexity for decoding the WD6 reference quality 12 kbit bit streams for the different versions of the arithmetic encoder;

[00097] Figure 32 shows a tabular representation of the average bit rates produced by the arithmetic encoder in an embodiment according to the invention and in WD6;

[00098] Figure 33 shows a table representation of the minimum, maximum and average USAC bit rates on a basis using the proposed scheme;

[00099] Figure 34 shows a tabular representation of the average bit rates produced by a USAC encoder using WD6 arithmetic encoder and an encoder according to an embodiment according to the invention (“New Proposal”);

[000100] Figure 35 shows a table representation of the best and worst cases for an embodiment according to the invention;

[000101] Figure 36 shows a table representation of the bit reservoir limit for an embodiment according to the invention;

[000102] Figure 37 shows a syntax representation of the arithmetically encoded data “arith_data”, according to an embodiment of the invention;

[000103] Figure 38 shows a legend of the help element definitions;

[000104] Figure 39 shows another legend of the definitions;

[000105] Figure 40a shows a code representation of the pseudo-program of a function or algorithm “arith_map_context”, according to an embodiment of the invention;

[000106] Figure 40b shows a code representation of the pseudo-program of a function or algorithm “arith_get_context”, according to an embodiment of the invention;

[000107] Figure 40c shows a code representation of the pseudo-program of a function or algorithm “arith_map_pk”, according to an embodiment of the invention;

[000108] Figure 40d shows a code representation of the pseudo-program of a first part of an “arith_decode” function or algorithm, according to an embodiment of the invention;

[000109] Figure 40e shows a code representation of the pseudo-program of a second part of an “arith_decode” function or algorithm, according to an embodiment of the invention;

[000110] Figure 40f shows a pseudo-program code representation of a function or algorithm for decoding one or more less significant bits, according to an embodiment of the invention;

[000111] Figure 40g shows a code representation of the pseudo-program of an “arith_update_context” function or algorithm, according to an embodiment of the invention;

[000112] Figure 40h shows a code representation of the pseudo-program of an “arith_save_context” function or algorithm, according to an embodiment of the invention;

[000113] Figures 41(1) and 41(2) show a table representation of a content of a table “ari_lookup_m[742]”, according to an embodiment of the invention;

[000114] Figures 42 (1), (2), (3), (4) show a table representation of a content of a table “ari_hash_m[742]”, according to an embodiment of the invention;

[000115] Figures 43 (1),(2),(3),(4),(5),(6) show a table representation of a content of a table “ari_cf_m[96][17]”, according to an embodiment of the invention; It is

[000116] Figure 44 shows a table representation of a table “ari_cf_r[4]”, according to an embodiment of the invention.

[000117] DETAILED DESCRIPTION OF ACHIEVEMENTS

[000118] Audio encoder according to figure 7

[000119] Figure 7 shows a schematic block diagram of an audio encoder, according to an embodiment of the invention. The audio encoder 700 is configured to receive an input audio information 710 and to provide, based on this, a coded audio information 712.

[000120] The audio encoder comprises an energy-compressed time-domain to frequency-domain converter 720 that is configured to provide a frequency-domain audio representation 722 based on a time-domain representation of the audio information input 710, such that the frequency domain audio representation 722 comprises a set of spectral values.

[000121] The audio encoder 700 also comprises an arithmetic encoder 730 configured to encode a spectral value (outside the set of spectral values that form the frequency domain audio representation 722), or a pre-processed version thereof, using a variable length password to obtain the encoded audio information 712 (which may comprise, for example, several variable length passwords).

[000122] The arithmetic encoder 730 is configured to map a spectral value, or a value of a more significant bit plane of a spectral value, into a code value (i.e., into a variable-length password) depending on a state of context.

[000123] The arithmetic encoder is configured to select a mapping rule that describes a mapping of a spectral value, or a most significant bit plane of a spectral value, into a code value, depending on a context state (current ). The arithmetic encoder is configured to determine the current context state, or a numerical current context value that describes the current context state, depending on several previously encoded (preferably, but not necessarily, adjacent) spectral values.

[000124] For this purpose, the arithmetic encoder is configured to evaluate a hash table, the entries of which define both the significant state values among the numeric context values and the limits of the ranges of the numeric context values.

[000125] The hash_table (also referred to as “ari_hash_m” below) is preferably defined as given in the table representation of figures 22(1), 22(2), 22(3) and 22(4).

[000126] Furthermore, the arithmetic encoder is preferably configured to evaluate the hash table (ari_hash_m), to determine whether the current numeric context value is identical to a table context value described by the hash table entries (ari_hash_m) and/or or to determine a range described by the hash table entries (ari_hash_m) in which the current numeric context value lies, and to derive an index value from the mapping rule (e.g., designated with “pki” here) that describes a mapping rule selected depending on an evaluation result.

[000127] In some cases, a mapping rule index value may be individually associated with a numeric context (current) value being a meaningful state value. Furthermore, a common mapping rule index value may be associated with different numeric context (current) values that lie within a range linked by range boundaries (where the range boundaries are preferably defined by hash table entries). ).

[000128] As can be seen, mapping a spectral value (from the frequency domain audio representation 722), or a more significant bit plane of a spectral value, into a code value (from the encoded audio information 712), can be accomplished by encoding the spectral value 740 using a mapping rule 742. A state tracker 750 may be configured to track context state. State tracker 750 provides information 754 that describes the current context state. Information 754 describing the current context state may preferably be in the form of a numeric current context value. A mapping rule selector 760 is configured to select a mapping rule, e.g., a cumulative frequency table, that describes a mapping of a spectral value, or a most significant bit plane of a spectral value, onto a value of code. Of course, the mapping rule selector 760 provides the mapping rule information 742 for encoding the spectral value 740. The mapping rule information 742 may be in the form of a mapping rule index value or a mapping rule table. cumulative frequencies selected depending on a mapping rule index value. The mapping rule selector 760 comprises (or at least evaluates) a hash table 752, the entries of which define both the significant state values between the numeric context values and thresholds and the ranges of the numeric context values. Preferably, the hash table entries 762 (ari_hash_m[742]) are defined as given in the table representation of figures 22(1) to 22(4). The hash table 762 is evaluated to select the mapping rule, that is, to provide the mapping rule information 742.

[000129] Preferably, but not necessarily, a mapping rule index value can be individually associated with a numeric context value being a meaningful state value, and a common mapping rule index value can be associated with different values numeric context that lies within a range connected by range boundaries.

[000130] To summarize the above, the audio encoder 700 performs an arithmetic encoding of a frequency domain audio representation provided by the time domain to frequency domain converter. Arithmetic coding is context-dependent, such that a mapping rule (e.g., a cumulative frequency table) is selected depending on the previously encoded spectral values. Of course, spectral values adjacent in time and/or frequency (or, at least, within a predetermined environment) to each other and/or to the currently encoded spectral value (i.e., spectral values within a predetermined environment of the currently encoded spectral value coded) are considered in arithmetic coding to adjust the probability distribution evaluated by arithmetic coding. By selecting an appropriate mapping rule, the numeric current context values 754 provided by a state tracker 750 are evaluated. Since typically, the number of different mapping rules is significantly smaller than the number of possible values of the current numeric context values 754, the mapping rule selector 760 allocates the same mapping rules (described, for example, by a value mapping rule index) to a comparably large number of different numeric context values. However, there are typically specific spectral configurations (represented by specific numerical context values) to which a particular mapping rule must be associated in order to obtain good coding efficiency.

[000131] It has been observed that the selection of a mapping rule depending on a numeric current context value can be performed with particularly high computational efficiency if the entries of a single hash table define both significant state values and range limits of the values. numeric (current) context. Furthermore, it was observed that the use of the hash table as defined in figures 22(1), 22(2), 22(3), 22(4) brings particularly high coding efficiency. It has been observed that this mechanism, in combination with said hash table, is well adapted to the requirements of mapping rule selection, as there are many cases where a single significant state value (or significant numeric context value) is embedded between a range left range of multiple non-significant state values (to which a common mapping rule is associated) and a right range of multiple non-significant state values (to which a common mapping rule is associated). Furthermore, the mechanism of using a single hash table, whose entries are defined in the tables in figures 22(1), 22(2), 22(3), 22(4) and define both the significant state values and the limits of ranges of numeric (current) context values can efficiently deal with different cases, where, for example, there are two adjacent non-significant state value ranges (also referred to as numeric non-significant context values) without a significant state value between they. Particularly high computational efficiency is achieved due to the number of table accesses being kept small. For example, a single iterative table lookup is sufficient in most implementations to discover whether the current numeric context value is equal to any of the significant state values, or in which of the non-significant state value ranges the value numeric current context remains. Consequently, the number of time- and energy-consuming table accesses can be kept small. Thus, the mapping rule selector 760, which uses the hash table 762, can be considered as a particularly efficient mapping rule selector in terms of computational complexity, while still allowing good coding efficiency (in terms of bit rate) to be obtained. bits).

[000132] Other details regarding the derivation of mapping rule information 742 from the current numeric context value 754 will be described below.

[000133] Audio decoder according to figure 8

[000134] Figure 8 shows a schematic block diagram of an audio decoder 800. The audio decoder 800 is configured to receive encoded audio information 810 and to provide, based on this, decoded audio information 812.

[000135] The audio decoder 800 comprises an arithmetic decoder 820 that is configured to provide various spectral values 822 based on an arithmetically encoded representation 821 of the spectral values.

[000136] The audio decoder 800 also comprises a frequency domain to time domain converter 830 that is configured to receive the decoded spectral values 822 and to provide the time domain audio representation 812, which may constitute the audio information. decoded audio using the decoded spectral values 822 to obtain decoded audio information 812.

[000137] The arithmetic decoder 820 comprises a determining spectral value 824, which is configured to map a code value of the arithmetically encoded representation 821 of the spectral values into a symbol code representing one or more of the decoded spectral values, or at least a portion (e.g., a most significant bit plane) of one or more of the decoded spectral values. The determining spectral value 824 can be configured to perform a mapping depending on a mapping rule, which can be described by information from the mapping rule 828a. The mapping rule information 828a may, for example, be in the form of a mapping rule index value, or a selected cumulative frequency table (selected, for example, depending on a mapping rule index value). .

[000138] The arithmetic decoder 820 is configured to select a mapping rule (e.g., a cumulative frequency table) that describes a mapping of the code value (described by the arithmetically coded representation 821 of the spectral values) into a symbol code ( which describes one or more spectral values, or a more significant bit plane thereof, in a decoded form) depending on a context state (which may be described by context state information 826a).

[000139] The arithmetic decoder 820 is configured to determine the current context state (described by the numerical current context value) depending on the various previously decoded spectral values. For this purpose, a state tracker 826 can be used, which receives information describing previously decoded spectral values and which provides, based on this, a numerical current context value 826a that describes the current context state.

[000140] The arithmetic decoder is also configured to evaluate a hash table 829, whose entries define both the significant state values between the numeric context values and the range boundaries of the numeric context values, to select the mapping rule. Preferably, the hash table entries 829 (ari_hash_m[742]) are defined as given in the table representation of figures 22(1) to 22(4). The hash table 829 is evaluated to select the mapping rule, i.e., in order to provide the mapping rule information 829.

[000141] Preferably, a mapping rule index value is individually associated with a numeric context value being a significant state value, and a common mapping rule index value is associated with different numeric context values that lie within a range linked by the range boundaries. Evaluation of the hash table 829 may, for example, be performed using a hash table evaluator that may be part of the mapping rule selector 828. Of course, information from the mapping rule 828a, for example, in the form of a value of the index of the mapping rule, is obtained based on the numeric current context value 826a that describes the current context state. The selector of the mapping rule 828 may, for example, determine the value of the index of the mapping rule 828a depending on a result of the evaluation of the hash table 829. Alternatively, the evaluation of the hash table 829 may directly provide the value of the index of the mapping rule.

[000142] Regarding the functionality of the audio signal decoder 800, it should be noted that the arithmetic decoder 820 is configured to select a mapping rule (e.g., the cumulative frequency table) that is, on average, well adapted to the values spectral values to be decoded, as the mapping rule is selected depending on the current context state (described, for example, by the numerical current context value), which in turn is determined depending on the various previously decoded spectral values. Thus, statistical dependencies between adjacent spectral values to be encoded can be explored. Furthermore, the arithmetic decoder 820 can be implemented efficiently, with a good trade-off between computational complexity, table size, and encoding efficiency, which utilizes the mapping rule selector 828. By evaluating a (single) hash table 829 , in entries describing both significant state values and range boundaries of non-significant state value ranges, a single iterative table lookup may be sufficient in order to derive the mapping rule 828a information from the current numeric context value 826a. Furthermore, it was observed that the use of the hash table as defined in figures 22(1), 22(2), 22(3), 22(4) brings particularly high coding efficiency. Of course, it is possible to map a number of different possible numeric context (current) values into a comparably small number of different mapping rule index values. Thus, it is possible to map a comparably large number of different possible numeric context (current) values into a comparably small number of different mapping rule index values. Using the hash table 829, as described above, and as defined in the tabular representation of Figures 22(1) to 22(4), it is possible to explain the observation that, in many cases, a single isolated significant state value (hash value meaningful context) is embedded between a left range of non-significant state values (non-significant context values) and a right range of non-significant state values (non-significant context values), where a different control rule index value mapping is associated with the meaningful state value (meaningful context value), when compared to the state values (context values) of the left bucket and the state values (context values) of the right bucket. However, the use of the hash table 829 is also well suited to situations in which the two ranges of numeric state values are immediately adjacent, without a significant state value between them.

[000143] To conclude, the mapping rule selector 828, which evaluates the hash table 829 “ari_hash_m[742], brings particularly good efficiency when selecting a mapping rule (or when providing a mapping rule index value). depending on the current context state (or depending on the numeric current context value that describes the current context state), as the hashing mechanism is well adapted to typical context scenarios in an audio decoder.

[000144] Other details will be described below.

[000145] Context value hashing mechanism according to figure 9

[000146] Next, a context hashing mechanism will be disclosed, which can be implemented in the mapping rule selector 760 and/or the mapping rule selector 828. The hash table 762 and/or the hash table 829, as defined in the table representation of figures 22(1) to 22(4), can be used to implement said value context hashing mechanism.

[000147] Now with reference to figure 9, which shows a hashing scenario of the current numeric context value, other details will be described. In the graphical representation of Figure 9, an abscissa 910 describes values of the current numeric context value (i.e., numeric context values). An ordinate 912 describes mapping rule index values. Tags 914 describe mapping rule index values to numeric non-significant context values (which describe non-significant states). Tags 916 describe mapping rule index values to “individual” (true) numeric meaningful context values that describe individual (true) meaningful states. Tags 916 describe mapping rule index values to “improper” numeric context values that describe “improper” meaningful states, where an “improper” meaningful state is a meaningful state in which the same mapping rule index value is associated with one of the adjacent context ranges of numeric non-significant contexts.

[000148] As can be seen, a hash table entry “ari_hash_m[i1]” describes an individual meaningful state (true) having a numeric context value of c1. As can be seen, the index value of the mriv1 mapping rule is associated with the individual meaningful state (true) having the numeric context value c1. Of course, both the numeric context value c1 and the mapping rule index value mriv1 can be described by the hash table entry “ari_hash_m[i1]”. A range 932 of the numeric context values is linked by the numeric context value c1, wherein the numeric context value c1 does not belong to the range 932, such that the largest numeric context value of the range 932 is equal to cl — 1. A mapping rule index value of mriv4 (which is different from mriv1) is associated with numeric context values from range 932. The mapping rule index value of mriv4 can, for example, be described by the table entry “ ari_lookup_m[i1-1]” from an additional table “ari_lookup_m”.

[000149] Additionally, an index value of the mriv2 mapping rule may be associated with numeric context values that lie within a range 934. A lower link of the range 934 is determined by the numeric context value c1, which is a meaningful numeric context value, where the numeric context value c1 does not belong to the range 932. Of course, the smallest value in the range 934 is equal to c1 + 1 (assuming numeric context integer values). Another limit of the range 934 is determined by the numeric context value c2, where the numeric context value c2 does not belong to the range 934, so the largest value in the range 934 is equal to c2 - 1. The numeric context value c2 is a so-called “improper” numeric context value, which is described by a hash table entry “ari_hash_m[i2]”. For example, the index value of the mriv2 mapping rule may be associated with the numeric context value c2, such that the numeric context value associated with the significant “improper” numeric context value c2 is equal to the index value of the mapping rule associated with the range 934 linked by the numeric context value c2. Furthermore, a range 936 of the numeric context value is also linked by the numeric context value c2, wherein the numeric context value c2 does not belong to the range 936, so that the smallest numeric context value of the range 936 is equal to c2 + 1. An mriv3 mapping rule index value, which is typically different from the mriv2 mapping rule index value, is associated with the numeric context values of range 936.

[000150] As can be seen, the index value of the mriv4 mapping rule, which is associated with the range 932 of numeric context values, can be described by an entry “ari_lookup_m[i1-1]” of a table “ari_lookup_m” , the index of the mriv2 mapping rule, which is associated with the numeric context values of the range 934, can be described by an entry of the “ari_lookup_m[i1]” table of the “ari_lookup_m” table, and the index value of the mriv2 mapping rule mriv3 mapping can be described by an entry from the “ari_lookup_m[i2]” table of the “ari_lookup_m” table. In the example given here, the hash table index value i2 may be greater, by 1, than the hash table index value i1.

[000151] As can be seen from Figure 9, the mapping rule selector 760 or the mapping rule selector 828 can receive a numeric current context value 764, 826a, and decide, by evaluating the entries of the “ari_hash_m” table, whether the numeric current context value is a meaningful state value (regardless of whether it is an “individual” meaningful state value or an “improper” meaningful state value), or whether the numeric current context value falls within one of the ranges 932 , 934, 936, which are bounded by the significant (“individual” or “improper”) state values c1, c2. Either checking whether the numeric current context value is equal to a significant state value c1, c2 or an evaluation of the intervals 932, 934, 936 in which the numeric current context value remains (in the case where the numeric current context value numeric does not equal a meaningful state value) can be performed using a single common hash table lookup.

[000152] Additionally, an evaluation of the hash table “ari_hash_m” can be used to obtain a hash table index value (e.g., i1-1, i1, or i2). Thus, the mapping rule selector 760, 828 may be configured to obtain, by evaluating a single hash table 762, 829 (e.g., the “ari_hash_m” hash table), a hash table index value (e.g., i1- 1, i1 or i2) designating a meaningful state value (e.g., c1 or c2) and/or a range (e.g., 932,934,936) and an information whether the current numeric context value is a meaningful context value (also designated as a meaningful state value) or not.

[000153] Furthermore, if it is observed in an evaluation of the hash table 762, 829, “ari_hash_m”, that the current numeric context value is not a “meaningful” context value (or “meaningful” state value), the hash table index value (e.g., i1-1, i1, or i2) obtained from a hash table evaluation (“ari_hash_m”) can be used to obtain a mapping rule index value associated with a range 932, 934 ,936 numeric context values. For example, the hash table index value (e.g., i1-1, i1, or i2) can be used to designate an additional mapping table entry (e.g., “ari_lookup_m”) that describes the hash table index values. mapping rule associated with the ranges 932, 934, 936 whose current numeric context value is found.

[000154] For further details, reference is made to the detailed discussion below of the “arith_get_pk” algorithm (where there are different options for this “arith_get_pk()” algorithm, examples of which are shown in figures 5e and 5f).

[000155] Furthermore, it should be noted that the size of the intervals may differ from one case to another. In some cases, a range of numeric context values comprises a single numeric context value. However, in many cases, a range can comprise several numeric context values.

[000156] Audio encoder according to figure 10

[000157] Figure 10 shows a schematic block diagram of an audio encoder 1000 according to an embodiment of the invention. The audio coder 1000 according to Figure 10 is similar to the audio coder 700 according to Figure 7, so that identical signals and media are designated with identical reference numerals in Figures 7 and 10.

[000158] The audio coder 1000 is configured to receive an input audio information 710 and to provide, based thereon, a coded audio information 712. The audio coder 1000 comprises a time domain to frequency domain converter with power compression 720, which is configured to provide a frequency domain representation 722 based on a time domain representation of the input audio information 710, such that the frequency domain audio representation 722 comprises a set of spectral values. The audio encoder 1000 also comprises an arithmetic encoder 1030 configured to encode a spectral value (outside the set of spectral values that form the frequency domain audio representation 722), or a preprocessed version thereof, using a password of length variable to obtain the encoded audio information 712 (which may comprise, for example, several passwords of variable length).

[000159] The arithmetic encoder 1030 is configured to map a spectral value, or multiple spectral values, or a value from a most significant bit plane of a spectral value or multiple spectral values, into a code value (i.e., into a password of variable length) depending on a context state. The arithmetic encoder 1030 is configured to select a mapping rule that describes a mapping of a spectral value, or multiple spectral values, or a most significant bit plane of a spectral value or multiple spectral values, into a code value. depending on a context state. The arithmetic encoder is configured to determine the current context state depending on several previously encoded spectral values (preferably not necessarily adjacent). For this purpose, the arithmetic encoder is configured to modify a numerical representation of a numerical prior context value, which describes a context state associated with one or more previously encoded spectral values (e.g., to select a corresponding mapping rule), depending on a context subregion value, to obtain a numeric representation of a current numeric context value that describes a context state associated with one or more spectral values to be encoded (e.g., to select a corresponding mapping rule ).

[000160] As can be seen, the mapping of a spectral value, or several spectral values, or a more significant bit plane of a spectral value or several spectral values, into a code value can be accomplished by an encoding of the spectral value 740 using a mapping rule described by a mapping rule information 742. A state tracker 750 may be configured to track the context state. State tracker 750 may be configured to modify a numerical representation of a numerical prior context value, which describes a context state associated with an encoding of one or more previously encoded spectral values, depending on a value of the context subregion. , to obtain a numerical representation of a numerical current context value that describes a context state associated with an encoding of one or more spectral values to be encoded. Modifying the numeric representation of the numeric prior context value may, for example, be performed by a numeric representation modifier 1052, which receives the numeric prior context value and one or more values from the context subregion and provides the numeric prior context value. numeric current context. Of course, state tracker 1050 provides information 754 that describes the current context state, for example, in the form of a numeric current context value. A mapping rule selector 1060 may select a mapping rule, for example, a cumulative frequency table, that describes a mapping of a spectral value, or multiple spectral values, or a most significant bit plane of a spectral value. or of several spectral values, in one code value. Of course, the mapping rule selector 1060 provides the mapping rule information 742 to the spectral coding 740.

[000161] It should be noted that, in some embodiments, the state tracker 1050 may be identical to the state tracker 750 or the state tracker 826. It should also be noted that the mapping rule selector 1060 may, in some embodiments, be identical to the mapping rule selector 760, or the mapping rule selector 828. Preferably, the mapping rule selector 828 may be configured to use a hash table “ari_hash_m[742]”, as defined in the table representation of the figures 22(1) to 22(4), for selecting the mapping rule. For example, the mapping rule selector may perform functionality as described above with reference to Figures 7 and 8.

[000162] To summarize the above, the audio encoder 1000 performs an arithmetic encoding of a frequency domain audio representation provided by the time domain to frequency domain converter. Arithmetic coding is context-dependent, such that a mapping rule (e.g., a cumulative frequency table) is selected depending on the previously encoded spectral values. Certainly, spectral values adjacent in time and/or frequency (or at least within a predetermined environment) to each other and/or to the currently encoded spectral value (i.e., spectral values within a predetermined environment of the currently encoded spectral value) are considered in arithmetic coding to adjust the probability distribution evaluated by arithmetic coding.

[000163] When determining the current numeric context value, a numeric representation of a numeric prior context value, which describes a context state associated with one or more previously encoded spectral values, is modified depending on a value of the subregion of context, to obtain a numeric representation of a current numeric context value that describes a context state associated with one or more spectral values to be encoded. This approach allows avoiding a complete recalculation of the current numeric context value, whose complete recalculation consumes a significant amount of resources in conventional approaches. A wide variety of possibilities exist for modifying the numeric representation of the numeric prior context value, including a combination of a resizing of a numeric representation of the numeric prior context value, an addition of a value from the context subregion, or a derived therefrom to the numeric representation of the numeric prior context value or to a processed numeric representation of the numeric prior context value, a substitution of a part of the numeric representation (rather than the entire numeric representation) of the numeric prior context value depending on the value of the context subregion, and so on. Thus, typically the numeric representation of the current numeric context value is obtained based on the numeric representation of the previous numeric context value and also based on at least one value of the context subregion, where typically a combination of operations is performed. to combine the numeric prior context value with a context subregion value, such as two or more operations outside of an addition operation, a subtraction operation, a multiplication operation, a division operation, a Boolean-AND operation, a Boolean-OR operation, a Boolean-NAND operation, a Boolean NOR operation, a Boolean-negation operation, a complement operation, or a shift operation. Thus, at least a portion of the numeric representation of the numeric prior context value is typically kept unchanged (except for an optional shift to a different position) when deriving the current numeric context value from the numeric prior context value. In contrast, other parts of the numeric representation of the numeric prior context value are changed depending on one or more values of the context subregion. Thus, the numerical current context value can be obtained with a comparably small computational effort, while avoiding a complete recalculation of the numerical current context value.

[000164] Thus, a meaningful numeric current context value can be obtained, which is well suited for use by the selector of mapping rule 1060, and which is particularly well suited for use in combination with the ari_hash_m hash table as defined in the representation in table of figures 22(1),22(2),22(3),22(4).

[000165] Consequently, efficient coding can be obtained while keeping the context calculation sufficiently simple.

[000166] Audio decoder according to figure 11

[000167] Figure 11 shows a schematic block diagram of an audio decoder 1100. The audio decoder 1100 is similar to the audio decoder 800 according to Figure 8, so that identical signals, means and functionalities are designated with identical reference numerals.

[000168] Audio decoder 1100 is configured to receive encoded audio information 810 and to provide, based thereon, decoded audio information 812. Audio decoder 1100 comprises an arithmetic decoder 1120 that is configured to provide various values decoded spectral values 822 based on an arithmetically encoded representation 821 of the spectral values. The audio decoder 1100 also comprises a frequency domain to time domain converter 830 that is configured to receive the decoded spectral values 822 and to provide the time domain audio representation 812, which may constitute the decoded audio information. using the decoded spectral values 822, to obtain decoded audio information 812.

[000169] The arithmetic decoder 1120 comprises a spectral value determiner 824, which is configured to map a code value of the arithmetically encoded representation 821 of the spectral values into a symbol code representing one or more of the decoded spectral values, or at least one part (e.g., a most significant bit plane) of one or more of the decoded spectral values. The spectral value determiner 824 may be configured to perform mapping depending on a mapping rule, which may be described by information from the mapping rule 828a. The mapping rule information 828a may, for example, comprise an index value of the mapping rule, or may comprise a set of entries selected from a cumulative frequency table.

[000170] The arithmetic decoder 1120 is configured to select a mapping rule (e.g., a cumulative frequency table) that describes a mapping of a code value (described by the arithmetically encoded representation 821 of the spectral values) into a symbol code (describing one or more spectral values) depending on a context state, which context state may be described by context state information 1126a. The context state information 1126a may be in the form of a numeric current context value. The arithmetic decoder 1120 is configured to determine the current context state depending on the various previously decoded spectral values 822. For this purpose, a state tracker 1126 can be used, which receives information describing the previously decoded spectral values. The arithmetic decoder is configured to modify a numeric representation of the numeric prior context value, which describes a context state associated with one or more previously decoded spectral values, depending on a value of the context subregion, to obtain a numeric representation of a numeric current context value that describes a context state associated with one or more spectral values to be decoded. A modification of the numeric representation of the numeric previous context value can, for example, be performed by a modifier of the numeric representation 1127, which is part of the state tracker 1126. Of course, the current context state information 1126a is obtained, e.g. , in the form of a numeric current context value. Mapping rule selection may be performed by a mapping rule selector 1128, which derives mapping rule information 828a from the current context state information 1126a, and which provides mapping rule information 828a to the value determiner. spectral 824. Preferably, the mapping rule selector 1128 can be configured to use a hash table “ari_hash_m[742]”, as defined in the table representation of figures 22(1) to 22(4), for rule selection mapping. For example, the mapping rule selector may perform functionality as described above with reference to Figures 7 and 8.

[000171] Regarding the functionality of the audio signal decoder 1100, it should be noted that the arithmetic decoder 1120 is configured to select a mapping rule (e.g., the cumulative frequency table) that is, on average, well adapted to the value spectral value to be decoded, as the mapping rule is selected depending on the current context state, which, in turn, is determined depending on several previously decoded spectral values. Thus, statistical dependencies between adjacent spectral values to be decoded can be explained.

[000172] Furthermore, modifying a numerical representation of a numerical prior context value that describes a context state associated with a decoding of one or more previously decoded spectral values, depending on a value of the context subregion, to obtain a numerical representation of a numerical current context value that describes a context state associated with a decoding of one or more spectral values to be decoded, it is possible to obtain meaningful information about the current context state, which is well suited for a mapping of an index value of the mapping rule, and which is particularly well suited for use in combination with the hash table ari_hash_m as defined in the tabular representation of figures 22(1),22(2),22(3),22(4 ), with comparably small computational effort. Maintaining at least one part of a numeric representation of the numeric prior context value (possibly in a bit-shifted or scaled version) while updating another portion of the numeric representation of the numeric prior context value depending on the values of the context subregion that have not been considered in the numeric previous context value, but must be considered in the numeric current context value, the number of operations to derive the numeric current context value can be kept reasonably small. Furthermore, it is possible to exploit the fact that the contexts used to decode adjacent spectral values are typically similar or correlated. For example, a context for a decoding of a first spectral value (or a first plurality of spectral values) is dependent on a first set of previously decoded spectral values. A context for decoding a second spectral value (or a second set of spectral values) that is adjacent to the first spectral value (or the first set of spectral values) may comprise a second set of previously decoded spectral values. Since the first spectral value and the second spectral value are assumed to be adjacent (e.g., with respect to associated frequencies), the first set of spectral values, which determine the context for encoding the first spectral value, may comprise some overlap with the second set of spectral values, which determines the context for decoding the second spectral value. Thus, it can be easily understood that the context state for decoding the second spectral value comprises certain correlation with the context state for decoding the first spectral value. A computational efficiency of context derivation, i.e., of deriving the current numeric context value, can be obtained by exploiting such correlations. It was observed that the correlation between context states for a decoding of adjacent spectral values (e.g., between the context state described by the numerical previous context value and the context state described by the numerical current context value) can be efficiently exploited. modifying only those parts of the numeric prior context value that are dependent on the context subregion values not considered for deriving the prior numeric context state, and deriving the current numeric context value from the numeric prior context value.

[000173] To conclude, the concepts described here allow particularly good computational efficiency when deriving the current numeric context value.

[000174] Other details will be described below.

[000175] Audio encoder according to figure 12

[000176] Figure 12 shows a schematic block diagram of an audio encoder, according to an embodiment of the invention. The audio coder 1200 according to Figure 12 is similar to the audio coder 700 according to Figure 7, such that identical means, signals and functionalities are designated with identical reference numerals.

[000177] The audio coder 1200 is configured to receive an input audio information 710 and to provide, based thereon, a coded audio information 712. The audio coder 1200 comprises a time domain to frequency domain converter with power compression 720 that is configured to provide a frequency domain audio representation 722 based on a time domain audio representation of the input audio information 710, such that the frequency domain audio representation 722 understand a set of spectral values. The audio encoder 1200 also comprises an arithmetic encoder 1230 configured to encode a spectral value (outside the set of spectral values that form the frequency domain audio representation 722), or several spectral values, or a preprocessed version thereof. using a variable length password to obtain the encoded audio information 712 (which may comprise, for example, several variable length passwords.

[000178] The arithmetic encoder 1230 is configured to map a spectral value, or multiple spectral values, or a value from a most significant bit plane of a spectral value or multiple spectral values, into a code value (i.e., into a password of variable length), depending on a context state. The arithmetic encoder 1230 is configured to select a mapping rule that describes a mapping of a spectral value, or multiple spectral values, or a most significant bit plane of a spectral value or multiple spectral values, into a code value , depending on a context state. The arithmetic encoder is configured to determine the current context state depending on several previously encoded spectral values (preferably, but not necessarily, adjacent). For this purpose, the arithmetic encoder is configured to obtain various context subregion values based on previously encoded spectral values, to store said context subregion values, and to derive a current numeric context value associated with one or more spectral values to be encoded depending on the stored context subregion values. Furthermore, the arithmetic encoder is configured to calculate the norm of a vector formed by several previously encoded spectral values, to obtain a common value of the context subregion associated with the several previously encoded spectral values.

[000179] As can be seen, the mapping of a spectral value, or several spectral values, or a more significant bit plane of a spectral value or several spectral values, into a code value can be accomplished by an encoding of the spectral value 740 using a mapping rule described by a mapping rule information 742. A state tracker 1250 may be configured to track the context state, and may understand a computer from the value of the context subregion 1252, to calculate the norm of a vector formed by several previously encoded spectral values, to obtain a common value of the context subregion associated with the several previously encoded spectral values. The state tracker 1250 is also preferably configured to determine the current context state depending on a result of said computation of a context subregion value performed by the computer of the context subregion value 1252. Of course, the state tracker 1250 state 1250 provides information 1254, which describes the current context state. A mapping rule selector 1260 may select a mapping rule, e.g., a cumulative frequency table, that describes a mapping of a spectral value, or a most significant bit plane of a spectral value, into a code value. . Of course, the mapping rule selector 1260 provides the mapping rule information 742 to the spectral encoding 740. Preferably, the mapping rule selector 1260 may be configured to use a hash table “ari_hash_m[742]” as defined in the representation in the table in figures 22(1) to 22(4), for selecting the mapping rule. For example, the mapping rule selector may perform functionality as described above with reference to Figures 7 and 8.

[000180] To summarize the above, the audio encoder 1200 performs an arithmetic encoding of a frequency domain audio representation provided by the time domain to frequency domain converter 720. The arithmetic encoding is context dependent, so that a mapping rule (e.g., a cumulative frequency table) is selected depending on the previously encoded spectral values. Of course, spectral values adjacent in time and/or frequency (or, at least, within a predetermined environment) to each other and/or to the currently encoded spectral value (i.e., spectral values within a predetermined environment of the currently encoded spectral value encoded) are considered in arithmetic coding to adjust the probability distribution evaluated by arithmetic coding.

[000181] In order to provide a numerical current context value, a value of the context subregion associated with the various previously encoded spectral values is obtained based on a calculation of a norm of a vector formed by the various previously encoded spectral values. The result of determining the numerical current context value is applied to the selection of the current context state, that is, to the selection of a mapping rule.

[000182] By calculating the norm of a vector formed by several previously encoded spectral values, meaningful information describing a part of the context of one or more spectral values to be encoded can be obtained, wherein the norm of a vector of previously encoded spectral values encoded can typically be represented with a comparably small number of bits. Thus, the amount of context information, which needs to be stored for future use in deriving a current numeric context value, can be kept sufficiently small by applying the approach discussed above for calculating context subregion values. It was observed that the norm of a vector of previously encoded spectral values typically comprises the most significant information regarding the state of the context. In contrast, it has been observed that the signal of said previously encoded spectral values typically comprises a subordinate impact on the context frame, so that it makes sense to negate the signal of the previously decoded spectral values in order to reduce the amount of information to be stored for use. future. Furthermore, it was observed that calculating a norm from a vector of previously encoded spectral values is a reasonable approach for deriving a context subregion value, as the average effect, which is typically obtained by calculating the norm, leaves the most important information about the substantially unaffected context state. To summarize, the value of the context subregion computer's calculation of the value of the context subregion 1252 allows it to provide compact context subregion information for storage and future reuse, in which the most relevant information about the state of context is preserved rather than reducing the amount of information.

[000183] Furthermore, it was observed that a numerical current context value obtained as discussed above is very well suited for a selection of a mapping rule using the hash table “ari_hash_m[742]”, as defined in the table representation of the figures. 22(1) to 22(4). For example, the mapping rule selector may perform functionality as described above with reference to Figures 7 and 8

[000184] Certainly, efficient encoding of the input audio information 710 can be obtained, while keeping the computational effort and the amount of data to be stored by the arithmetic encoder 1230 sufficiently small.

[000185] Audio decoder according to figure 13

[000186] Figure 13 shows a schematic block diagram of an audio decoder 1300. How the audio decoder 1300 is similar to the audio decoder 800 according to Figure 8, and the audio decoder 1100 according to Figure 11, identical means, signs and functionalities are designated with identical numbers.

[000187] Audio decoder 1300 is configured to receive encoded audio information 810 and to provide, based thereon, decoded audio information 812. Audio decoder 1300 comprises an arithmetic decoder 1320 that is configured to provide various values decoded spectral values 822 based on an arithmetically encoded representation 821 of the spectral values. The audio decoder 1300 also comprises a frequency domain to time domain converter 830 that is configured to receive the decoded spectral values 822 and to provide the time domain audio representation 812, which may constitute the decoded audio information. using the decoded spectral values 822, to obtain decoded audio information 812.

[000188] The arithmetic decoder 1320 comprises a spectral value determiner 824 that is configured to map a code value of the arithmetically encoded representation 821 of the spectral values into a symbol code representing one or more of the decoded spectral values, or at least a portion (e.g., a most significant bit plane) of one or more of the decoded spectral values. The spectral value determiner 824 may be configured to perform a mapping depending on a mapping rule, which is described by an information of the mapping rule 828a. The mapping rule information 828a may, for example, comprise a mapping rule index value, or a selected set of cumulative frequency table entries.

[000189] The arithmetic decoder 1320 is configured to select a mapping rule (e.g., the cumulative frequency table) that describes a mapping of a code value (described by the arithmetically encoded representation 821 of the spectral values) into a symbol code (which describes one or more spectral values) depending on a context state (which may be described by context state information 1326a). Preferably, the arithmetic decoder 1320 may be configured to use a hash table “ari_hash_m[742]”, as defined in the table representation of figures 22(1) to 22(4), for mapping rule selection. For example, the arithmetic decoder 1320 may perform functionality as described above with reference to FIGS. 7 and 8. The arithmetic decoder 1320 is configured to determine the current context state depending on the various previously decoded spectral values 822. For this purpose, a tracker state code 1326 can be used, which receives information describing the previously decoded spectral values. The arithmetic decoder is also configured to obtain various context subregion values based on previously decoded spectral values and to store said context subregion values. The arithmetic decoder is configured to derive a numeric current context value associated with one or more spectral values to be encoded depending on the stored values of the context subregion. The arithmetic decoder 1320 is configured to calculate the norm of a vector formed by the plurality of previously decoded spectral values, to obtain a common value of the context subregion associated with the plurality of previously decoded spectral values.

[000190] The calculation of the norm of a vector formed by the various previously encoded spectral values, to obtain a common value of the context sub-region associated with the various previously decoded spectral values, can, for example, be carried out by the sub-region value computer. -context region 1327, which is part of the state tracker 1326. Of course, current context state information 1326a is obtained based on the values of the context subregion, where the state tracker 1326 preferably provides a value of numeric current context associated with one or more spectral values to be decoded depending on the stored values of the context subregion. The selection of mapping rules may be performed by a mapping rule selector 1328, which derives mapping rule information 828a from the current context state information 1326a, and which provides mapping rule information 828a to the value determiner. spectral 824.

[000191] Regarding the functionality of the audio signal decoder 1300, it should be noted that the arithmetic decoder 1320 is configured to select a mapping rule (e.g., the cumulative frequency table) that is, on average, well adapted to the value spectral value to be decoded, as the mapping rule is selected depending on the current context state, which, in turn, is determined depending on several previously decoded spectral values. Thus, statistical dependencies between adjacent spectral values to be decoded can be explored.

[000192] However, it was observed that it is efficient, in terms of memory usage, to store values of the context sub-region, which are based on the calculation of a norm of a vector formed in several previously decoded spectral values, for future use in determining the numeric context value. It was also observed that such context subregion values still comprise the most relevant context information. Thus, the concept used by state tracker 1326 constitutes a good compromise between coding efficiency, computational efficiency, and storage efficiency.

[000193] Other details will be described below.

[000194] Audio encoder according to figure 1

[000195] Next, an audio coder according to an embodiment of the present invention will be described. Figure 1 shows a schematic block diagram of such an audio encoder 100.

[000196] The audio coder 100 is configured to receive an input audio information 110 and to provide, based on this, a bit stream 112, which constitutes a coded audio information. The audio encoder 100 optionally comprises a preprocessor 120, which is configured to receive the input audio information 110 and to provide, based thereon, a preprocessed input audio information 110a. The audio encoder 100 also comprises a power-compressed time-domain-to-frequency-domain signal transformer 130, which is also referred to as a signal converter. The signal converter 130 is configured to receive the input audio information 110, 110a and to provide, based thereon, a frequency domain audio information 132, which preferably takes the form of a set of spectral values. For example, signal transformer 130 may be configured to receive a structure of input audio information 110, 110a (e.g., a block of time domain samples) and to provide a set of spectral values representing the audio content. of the respective audio structure. Furthermore, signal transformer 130 may be configured to receive various subsequent overlapping or non-overlapping audio structures of input audio information 110, 110a and to provide, based thereon, a frequency domain audio representation of time, which comprises a sequence of subsequent sets of spectral values, a set of spectral values associated with each structure.

[000197] The energy-compressed time-domain-to-frequency-domain signal transformer 130 may comprise an energy-compressed filter bank, which provides spectral values associated with different frequency bands, overlapping or non-overlapping. For example, signal transformer 130 may comprise a windowing MDCT transformer 130a, which is configured to window input audio information 110, 110a (or structure thereof) using a transform window, and to perform a transformation modified discrete cosine of the windowed input audio information 110, 110a (or their windowed structure). Thus, the frequency domain audio representation 132 may comprise a set of, for example, 1024 spectral values in the form of MDCT coefficients associated with a structure of the input audio information.

[000198] The audio encoder 100 may further optionally comprise a spectral post-processor 140, which is configured to receive the frequency domain audio representation 132 and provide, based thereon, a frequency domain audio representation post-processing 142. The spectral post-processor 140 may, for example, be configured to perform a form of temporal noise and/or a long-term forecast and/or any other spectral post-processing known in the art. The audio encoder further optionally comprises a frequency multiplier/quantifier 150, which is configured to receive the frequency domain audio representation 132 or its post-processed version 142 and provide a scaled frequency domain audio representation. and quantized 152.

[000199] Audio encoder 100 further optionally comprises a psychoacoustic model processor 160, which is configured to receive input audio information 110 (or the post-processed version 110a thereof) and provide, based thereon, a optional control information, which may be used for control of the power-compressed time-domain to frequency-domain signal transformer 130, for control of the optional spectral postprocessor 140, and/or for control of the frequency multiplier /optional quantifier 150. For example, the psychoacoustic model processor 160 may be configured to analyze input audio information to determine which components of the input audio information 110, 110a are particularly important to human perception of audio content and which components of the input audio information 110, 110a are least important for the perception of the audio content. Thus, the psychoacoustic model processor 160 can provide control information, which is used by the audio encoder 100 to adjust the scale of the frequency domain audio representation 132, 142 by the frequency multiplier/quantifier 150 and/or the quantization resolution applied by frequency multiplier/quantifier 150. Consequently, perceptually important scale factor bands (i.e., groups of adjacent spectral values that are particularly important to human perception of audio content) are scaled with a large factor of scaled and quantized with a comparably high resolution, while perceptually less important scale factor bands (i.e., groups of adjacent spectral values) are scaled with a comparably smaller scale factor and quantized with a comparably lower quantization resolution. Thus, the scaled spectral values of perceptually more important frequencies are typically significantly larger than the scaled spectral values of perceptually less important frequencies.

[000200] The audio encoder also comprises an arithmetic encoder 170, which is configured to receive the scaled and quantized version 152 of the frequency domain audio representation 132 (or, alternatively, the post-processed version 142 of the frequency domain audio representation 142). frequency domain audio 132, or even the frequency domain audio representation 132 itself) and providing arithmetic password information 172a based on this, so that the arithmetic password information represents the frequency domain audio representation 152.

[000201] Audio encoder 100 also comprises a bitstream payload formatter 190, which is configured to receive arithmetic password information 172a. The bitstream payload formatter 190 is also typically configured to receive additional information, such as scale factor information that describes which scale factors were applied by the frequency multiplier/quantifier 150. Additionally, the formatter Bitstream payload 190 may be configured to receive other control information. The bitstream payload formatter 190 is configured to provision the bitstream 112 based on the information received by assembling the bitstream according to a desired bitstream syntax, which will be discussed below.

[000202] Next, details regarding the arithmetic encoder 170 will be described. The arithmetic encoder 170 is configured to receive various post-processed and scaled and quantized spectral values from the frequency domain audio representation 132. The arithmetic encoder comprises an extractor of the most significant bit plane 174, or even of two spectral values, which is configured to extract a most significant bit plane m from a spectral value. It should be noted here that the most significant bit plane may comprise one or more bits (e.g., two or three bits), which are the most significant bits of the spectral value. Thus, the most significant bit plane extractor 174 provides a most significant bit plane value 176 from a spectral value.

[000203] Alternatively, however, the most significant bit plane extractor 174 may provide a combined value of the most significant bit plane m that combines the most significant bit plane of several spectral values (e.g., of spectral values a and b ). The most significant bit plane of the spectral value a is designated with m. Alternatively, the combined value of the most significant bit plane of several spectral values a,b is designated with m.

[000204] The arithmetic encoder 170 also comprises a first password determiner 180, which is configured to determine an arithmetic password acod_m [pki][m] that represents the value of the most significant bit plane m. Optionally, the password determiner 180 may also provide one or more escape passwords (also designated herein as “ARITH_ESCAPE”) indicating, for example, how many least significant bit planes are available (and, consequently, indicating the numerical weight of the bit plane). most significant bits). The first password determiner 180 may be configured to provide the password associated with a value of the most significant bit plane m that uses the selected cumulative frequency table having (or being referenced by) a cumulative frequency table index pki.

[000205] In order to determine which table of cumulative frequencies should be selected from, the arithmetic encoder preferably comprises a state tracker 182, which is configured to track the state of the arithmetic encoder, for example, observing which spectral values were previously encoded. The status tracker 182 consequently provides status information 184, for example, a status value designated with “s” or “t” or “c”. The arithmetic encoder 170 also comprises a cumulative frequency table selector 186, which is configured to receive state information 184 and provide information 188 that describes the selected cumulative frequency table to the password determiner 180. For example, the arithmetic code selector 180. cumulative frequency table 186 may provide a cumulative frequency table index “pki” that describes which cumulative frequency table, out of a set of 96 cumulative frequency tables, is selected for use by the password determiner. Alternatively, the cumulative frequency table selector 186 may provide the entire selected cumulative frequency table or a sub-table to the password determiner. Thus, the password determiner 180 may use the selected cumulative frequency table or sub-table for provisioning the password acod_m[pki][m] of the most significant bit plane value m, such that the current password acod_m[pki ][m] that encodes the value of the most significant bit plane m is dependent on the value of m and the index of the cumulative frequency table pki, and consequently on the current state information 184. Other details regarding the encoding process and the password format obtained will be described below.

[000206] It should be noted, however, that in some embodiments, state tracker 182 may be identical to, or take on the functionality of, state tracker 750, state tracker 1050, or state tracker 1250. It should also be It is noted that the cumulative frequency table selector 186 may, in some embodiments, be identical to, or take on the functionality of, the mapping rule selector 760, the mapping rule selector 1060, or the mapping rule selector 1260. Furthermore, the first password determiner 180 may, in some embodiments, be identical to, or take on the functionality of, the spectral value encoding 740.

[000207] The arithmetic encoder 170 further comprises a least significant bit plane extractor 189a, which is configured to extract one or more least significant bit planes from the scaled and quantized frequency domain audio representation 152, if one or more of the Spectral values to be encoded exceed the range of values that can be encoded using the most significant bit plane only. The least significant bit planes may comprise one or more bits as desired. Thus, the least significant bit plane extractor 189a provides information from the least significant bit plane 189b. The arithmetic encoder 170 also comprises a second password determiner 189c, which is configured to receive information from the least significant bit plane 189d and provide, based thereon, 0, 1 or more passwords "acod_r" representing the content of 0, 1 or more least significant bit planes. The second password determiner 189c may be configured to apply an arithmetic coding algorithm or any other coding algorithm to derive the least significant bit plane passwords “acod_r” from the least significant bit plane information 189b.

[000208] It should be noted here that the number of least significant bit planes may vary depending on the value of the scaled and quantized spectral values 152, so that there may be no less significant bit planes if the scaled and quantized spectral value to be encoded is comparably small, so that it may have a less significant bit plane if the current scaled and quantized spectral value to be encoded is of a medium range, and so that it may have more than one less significant bit plane if the spectral value scaled and quantized to be encoded has a comparably large value.

[000209] To summarize the above, the arithmetic encoder 170 is configured to encode scaled and quantized spectral values, which are described by information 152, which uses a hierarchical coding process. The most significant bit plane (comprising, for example, one, two or three bits per spectral value) of one or more spectral values, is encoded to obtain an arithmetic password “acod_m[pki][m]” of a value of the most significant bit plane m. One or more least significant bit planes (each of the least significant bit planes comprising, for example, one, two or three bits) of one or more spectral values are encoded to obtain one or more passwords “acod_r”. When encoding the most significant bit plane, the value m of the most significant bit plane is mapped to a password acod_m[pki][m]. For this purpose, 96 different cumulative frequency tables are available for encoding the value m depending on a state of the arithmetic encoder 170, i.e. depending on previously encoded spectral values. Thus, the password “acod_m[pki][m]” is obtained. Additionally, one or more passwords “acod_r” are provided and included in the bitstream if one or more less significant bit planes are present.

[000210] RESET DESCRIPTION

[000211] The audio encoder 100 may optionally be configured to decide whether an improvement in bit rate can be obtained by redefining the context, for example, by setting the state index to a default value. Thus, the audio encoder 100 may be configured to provide reset information (e.g., called “arith_reset_flag”) indicating whether the context for arithmetic encoding is reset, and also indicating whether the context for arithmetic decoding in a corresponding decoder must be reset.

[000212] Details regarding the bitstream format and applied cumulative frequency tables will be discussed below.

[000213] Audio decoder according to figure 2

[000214] Next, an audio decoder, according to an embodiment of the invention, will be described. Figure 2 shows a schematic block diagram of such an audio decoder 200.

[000215] The audio decoder 200 is configured to receive a bit stream 210, which represents encoded audio information and which may be identical to the bit stream 112 provided by the audio encoder 100. The audio decoder 200 provides an information decoded audio stream 212 based on bitstream 210.

[000216] Audio decoder 200 comprises an optional bitstream payload deformator 220, which is configured to receive bitstream 210 and to extract from bitstream 210 a frequency domain audio encoded representation 222. By For example, the payload deformator of bitstream 220 may be configured to extract from bitstream 210 arithmetically decoded spectral data such as an arithmetic password “acod_m [pki][m]” representing the value of the bit plane most significant m of a spectral value a, or of several spectral values a, b, and a password “acod_r” that represents a content of a least significant bit plane of the spectral value a, or of several spectral values a, b, of frequency domain audio representation. Thus, the frequency domain audio coded representation 222 constitutes (or comprises) an arithmetically coded representation of spectral values. The bitstream payload deformator 220 is further configured to extract from the bitstream additional control information, which is not shown in FIG. 2. Additionally, the bitstream payload deformator is optionally configured to extract from the bitstream of bits 210, a state reset information 224, which is also referred to as an arithmetic reset flag or “arith_reset_flag”.

[000217] The audio decoder 200 comprises an arithmetic decoder 230, which is also referred to as a “spectral silent decoder”. The arithmetic decoder 230 is configured to receive the encoded representation of frequency domain audio 220 and, optionally, the state reset information 224. The arithmetic decoder 230 is also configured to provide a decoded representation of the frequency domain audio 232. which may comprise a decoded representation of spectral values. For example, the frequency domain audio decoded representation 232 may comprise a decoded representation of spectral values, which are described by the frequency domain audio coded representation 220.

[000218] The audio decoder 200 also comprises an optional inverse frequency quantifier/remultiplier 240, which is configured to receive the decoded frequency domain audio representation 232 and provide, based thereon, a frequency domain audio representation. inversely quantized and rescaled 242.

[000219] The audio decoder 200 further comprises an optional spectral preprocessor 250, which is configured to receive the inversely quantized and scaled frequency domain audio representation 242 and provide, based thereon, a preprocessed version 252 of the inversely quantized and scaled frequency domain audio representation 242. The audio decoder 200 also comprises a time domain frequency domain signal transformer 260, which is also referred to as a “signal converter”. Signal transformer 260 is configured to receive the preprocessed version 252 of the scaled inversely quantized frequency domain audio representation 242 (or, alternatively, the scaled inversely quantized frequency domain audio representation 242 or the frequency domain audio decoded representation 232) and provide, based thereon, a time domain representation 262 of the audio information. The time-domain frequency-domain signal transformer 260 may, for example, comprise a transformer for performing an inverse modified discrete cosine transform (IMDCT) and appropriate windowing (as well as other auxiliary features, such as, for example, a overlapping and addition).

[000220] The audio decoder 200 may further comprise an optional time domain postprocessor 270, which is configured to receive the time domain representation 262 of the audio information and to obtain the decoded audio information 212 using a time domain post-processing. However, if post-processing is omitted, the time domain representation 262 may be identical to the decoded audio information 212.

[000221] It should be noted here that the inverse frequency quantifier/remultiplier 240, the spectral preprocessor 250, the time domain frequency domain signal transformer 260 and the time domain postprocessor 270 can be controlled depending on the control information, which is extracted from the bitstream 210 by the bitstream payload deformator 220.

[000222] To summarize the entire functionality of the audio decoder 200, a frequency domain audio decoded representation 232, e.g., a set of spectral values associated with an audio structure of the encoded audio information, may be obtained based on in the frequency domain coded representation 222 using the arithmetic decoder 230. Subsequently, the set of, for example, 1024 spectral values, which may be MDCT coefficients, is inversely quantized, rescaled and preprocessed. Thus, a set of spectrally inversely quantized and rescaled preprocessed spectral values (e.g. 1024 MDCT coefficients) is obtained. Subsequently, a time domain representation of an audio structure is derived from the spectrally preprocessed, inversely quantized and rescaled set of frequency domain values (e.g., MDCT coefficients). Thus, a time domain representation of an audio structure is obtained. The time domain representation of a given audio structure may be combined with the time domain representations of previous and/or subsequent audio structures. For example, an overlay and addition between the time domain representations of subsequent audio structures may be performed in order to smooth the transactions between the time domain representations of adjacent audio structures and in order to achieve an aliasing. For details regarding the reconstruction of the decoded audio information 212 based on the decoded time-frequency domain audio representation 232, reference is made, for example, to International Standard ISO/IEC 14496-3, part 3, subpart 4 where a detailed discussion is given. However, other more elaborate overlay and aliasing schemes can be used.

[000223] Next, some details regarding the arithmetic decoder 230 will be described. The arithmetic decoder 230 comprises a most significant bit plane determiner 284, which is configured to receive the arithmetic password acod_m [pki][m] that describes the value of the most significant bit plane m. The most significant bit plane determiner 284 may be configured to use the cumulative frequency table out of a set comprising multiple 96 cumulative frequency tables to derive the most significant bit plane value m from the arithmetic password “acod_m [pki] [m]”.

[000224] The most significant bit plane determiner 284 is configured to derive values 286 of a most significant bit plane from one or more spectral values based on the acod_m password. The arithmetic decoder 230 further comprises a least significant bit plane determiner 288, which is configured to receive one or more passwords “acod_r” representing one or more least significant bit planes of a spectral value. Thus, the least significant bit plane determiner 288 is configured to provide decoded values 290 from one or more least significant bit planes. The audio decoder 200 also comprises a bit plane combiner 292, which is configured to receive the decoded values 286 of the most significant bit plane of one or more spectral values and the decoded values 290 of one or more less significant bit planes. of the spectral values if such least significant bit planes are available for the current spectral values. Thus, the bit plane combiner 292 provides decoded spectral values, which are part of the decoded frequency domain audio representation 232. Naturally, the arithmetic decoder 230 is typically configured to provide several spectral values in order to obtain a complete set of decoded spectral values associated with a current structure of the audio content.

[000225] The arithmetic decoder 230 further comprises a cumulative frequency table selector 296, which is configured to select one of 64 cumulative frequency tables ari_cf_m[64][17] (each table ari_cf_m[pki][17], with 0 <pki<63, having 17 entries) depending on a state index 298 that describes a state of the arithmetic decoder. To select one of the cumulative frequency tables, the cumulative frequency table selector preferably evaluates the hash table ari_hash_m[742] as defined by the tabular representation of figures 22(1), 22(2), 22(3), 22( 4). Details regarding this evaluation of the ari_hash_m[742] hash table will be described below. The arithmetic decoder 230 further comprises a state tracker 299, which is configured to track a state of the arithmetic decoder depending on previously decoded spectral values. The state information may optionally be reset to a standard state information in response to the state reset information 224. Of course, the cumulative frequency table selector 296 is configured to provide an index (e.g., pki) of a table of selected cumulative frequencies, or a selected cumulative frequency table or the subtable itself, for application in decoding the value of the most significant bit plane m depending on the code “acod_m”.

[000226] To summarize the functionality of audio decoder 200, audio decoder 200 is configured to receive an efficiently encoded representation of frequency domain audio of bit rate 222 and to obtain a decoded representation of frequency domain audio with based on this. In the arithmetic decoder 230, which is used to obtain the frequency domain audio decoded representation 232 based on the frequency domain audio encoded representation 222, a probability of different combinations of values of the most significant bit plane of adjacent spectral values is explored using an arithmetic decoder 280, which is configured to apply the cumulative frequency table. In other words, statistical dependencies between spectral values are explored by selecting different cumulative frequency tables out of a set comprising 64 different cumulative frequency tables depending on a state index 298, which is obtained by observing the previously calculated decoded spectral values.

[000227] It should be noted that state tracker 299 may be identical to, or may take on the functionality of, state tracker 826, state tracker 1126, or state tracker 1326. The cumulative frequency table selector 296 may be identical to, or may take the functionality of, the mapping rule selector 828, the mapping rule selector 1128, or the mapping rule selector 1328. The most significant bit plane determiner 284 may be identical to, or may take the functionality of the spectral value determiner 824.

[000228] Silent Spectral Coding Tool Overview

[000229] Next, details regarding the encoding and decoding of the algorithm, which is performed, for example, by the arithmetic encoder 170 and the arithmetic decoder 230, will be explained.

[000230] Focus should be placed on describing the decoding of the algorithm. It should be noted, however, that a corresponding encoding algorithm can be performed in accordance with the teachings of the decoding algorithm, in which mappings between encoded and decoded spectral values are inverse, and in which the calculation of the mapping rule index value is substantially identical. In an encoder, the encoded spectral values take the place of the decoded spectral values. Furthermore, the spectral values to be encoded take the place of the spectral values to be decoded.

[000231] It should be noted that decoding, which will be discussed below, is used in order to enable so-called “silent spectral coding” of scaled and quantized spectral values typically post-processed. Silent spectral coding is used in an audio coding/decoding concept (or in any other coding/decoding concept) to further reduce the redundancy of the quantized spectrum, which is obtained, for example, by a time-domain transformer in frequency domain with energy compression. The spectral silent coding scheme, which is used in embodiments of the invention, is based on an arithmetic coding with a dynamically adapted context.

[000232] In some embodiments, according to the invention, the silent coding spectral scheme is based on 2 tuples, that is, two close spectral coefficients are combined. Each double tuple is divided into sign, the most significant 3-bit wide plane, and the remaining least significant bit planes. Silent encoding for the most significant bitwise plane m uses context-dependent cumulative frequency tables derived from four previously decoded double tuples. Silent encoding is fed, for example, by the quantized spectral values and uses the context-dependent cumulative frequency tables derived from four previously encoded close double tuples. Here, close in both time and frequency is preferably considered, as illustrated in figure 4. The cumulative frequency tables (which will be explained below) are then used by the arithmetic encoder to generate a binary code of variable length (and by the arithmetic decoder to derive decoded values from a variable-length binary code).

[000233] For example, the arithmetic encoder 170 produces a binary code for a given set of symbols and their respective probabilities (i.e., depending on the respective probabilities). The binary code is generated by mapping a probability range where the set of symbols lies in a password.

[000234] Silent encoding for the remaining least significant bit plane r uses, for example, a single cumulative frequency table. Cumulative frequencies correspond, for example, to a uniform distribution of symbols that occur in the least significant bit planes, that is, it is expected that there is the same probability that a 0 or a 1 occurs in the least significant bit planes. However, other solutions for encoding the remaining least significant bit plane or bit planes can be used.

[000235] Next, another brief overview of the silent spectral coding tool will be given. Silent spectral coding is used to further reduce the redundancy of the quantized spectrum. The spectral silent coding scheme is based on an arithmetic coding, with a dynamically adapted context. Silent coding is fed by the quantized spectral values and uses the context-dependent cumulative frequency tables derived from, for example, four previously decoded close double tuples of spectral values. Here, proximate, in time and frequency, is considered as illustrated in figure 4. The cumulative frequency tables are then used by the arithmetic encoder to generate a binary code of variable length.

[000236] The arithmetic coder produces a binary code for a given set of symbols and their respective probabilities. Binary code is generated by mapping a probability interval, where the set of symbols remains, into a code.

[000237] Decoding process

[000238] 11.1 Overview of the decoding process

[000239] Next, an overview of the coding process for a spectral value will be presented, with reference to figure 3, which shows a representation of the pseudo-program code of the decoding process for various spectral values.

[000240] The process of decoding various spectral values comprises an initialization 310 of a context. Context initialization 310 comprises a derivation of the current context from a previous context, which uses the function “arith_map_context(N, arith_reset_flag)”. The derivation of the current context from a previous context can selectively comprise a redefinition of the context. Context redefinition and the derivation of the current context from a previous context will be discussed below. Preferably, the function “arith_map_context(N, arith_reset_flag)” according to figure 5a can be used, but alternatively the function according to figure 5b can be used.

[000241] The decoding of various spectral values also comprises an iteration of a spectral value decoding 312 and a context update 313, which context update 313 is performed by the function “arith_update_context(i, a, b)” which is described below . Spectral value decoding 312 and context update 312 are repeated 1 g/2 times, where 1 g/2 indicates the number of 2-tuples of spectral values to be decoded (e.g., for an audio structure), unless the symbol called “ARITH_STOP” is detected. Furthermore, decoding a set of spectral values l g also comprises a signal decoding 314 and a finishing step 315.

[000242] Decoding 312 of a tuple of spectral values comprises a calculation of the context value 312a, a decoding of the most significant bit plane 312b, a detection of the arithmetic stop symbol 312c, an addition of the least significant bit plane 312d, and an update to the 312e matrix.

[000243] The calculation of the state value 312a comprises a call to the function “arith_get_context(c,i,N)” as shown, for example, in figure 5c or 5d. Preferably, the function “arith_get_context(c,i,N)” according to figure 5c is used. Thus, a numeric current context (state) value c is provided as a return value of the function call “arith_get_context(c,i,N)”. As can be seen, the numeric prior context value (also designated as “c”), which serves as an input variable in the function “arith_get_context(c,i,N)”, is updated to get, as a return value , the current numeric context value c.

[000244] Decoding the most significant bit plane 312b comprises an iterative execution of an algorithm decoding 312ba, and a derivation 312bb of values a, b of the result value m of the algorithm 312ba. In preparing the 312ba algorithm, the variable lev is initialized to zero. Algorithm 312ba is repeated until an “interrupt” instruction (or condition) is met. Algorithm 312ba comprises a calculation of a state index “pki” (which also serves as an index of the cumulative frequency table) depending on the current numeric context value c, and also depending on the level value “esc_nb” that uses the function “arith_get_pk()”, which is discussed below (and realizations of these are shown, for example, in Figs. 5e and 5f). Preferably, the function “arith_get_pk(c)” according to figure 5e is used. Algorithm 312ba also comprises selecting a table of cumulative frequencies depending on the state index “pki”, which is returned by calling the function “arith_get_pk”, wherein a variable “cum_freq” can be set to a starting address of one of the 96 cumulative frequency tables (or subtables) depending on the state index “pki”. A variable “cfl” can also be initialized to a length of the selected cumulative frequency table (or a sub-table), which is, for example, equal to a number of symbols in the alphabet, i.e. the number of different values that can be decoded. The length of all cumulative frequency tables (or subtables) from “ari_cf_m[pki=0][17]” to “ari_cf_m[pki=95][17]” available for decoding the most significant bit plane value m is 17, as 16 different values of the most significant bit plane and an escape symbol (“ARITH_ESCAPE”) can be decoded. Preferably, the cumulative frequency table ari_cf_m[64][17], as defined in the table representation according to figures 23(1), 23(2), 23(3), which define the cumulative frequency tables (or subtables) “ari_cf_m[pki=0][17]” to “ari_cf_m[pki=63][17]”, is evaluated, to obtain the selected cumulative frequency table (or sub-table).

[000245] Subsequently, a value of the most significant bit plane m can be obtained by executing a function “arith_decode()”, taking into account the selected cumulative frequency table (described by the variable “cum_freq” and the variable “cfl”). By deriving the value of the most significant bit plane m, bits called “acod_m” of bit stream 210 can be evaluated (see, for example, Figure 6g or Figure 6h). Preferably, the function “arith_decode(cum_freq,cfl)” according to Figure 5g is used, but alternatively the function “arith_decode(cum_freq,cfl)” according to Figures 5h and 5i can be used.

[000246] Algorithm 312ba also comprises checking whether the value of the most significant bit plane m is equal to an escape symbol “ARITH_ESCAPE”, or not. If the value of the most significant bit plane m is not equal to the arithmetic escape symbol, the 312ba algorithm is aborted (“break” condition) and the remaining instructions of the 312ba algorithm are then skipped. Thus, the execution of the process is continued with the definition of the value b and the value a in step 312bb. In contrast, if the decoded most significant bit plane value m is identical to the arithmetic escape symbol, or “ARITH_ESCAPE”, the level value “lev” is increased by one. The “esc_nb” level value is set to be equal to the “lev” level value, unless the “lev” variable is greater than seven, in which case, the “esc_nb” variable is set to be equal to seven. As mentioned, the 312ba algorithm is then repeated until the decoded most significant bit plane value m is different from the arithmetic escape symbol, in which a modified context is used (because of the function input parameter “arith_get_pk()” be adapted depending on the value of the variable “esc_nb”).

[000247] As soon as the most significant bit plane is decoded using one-period execution or iterative execution of algorithm 312ba, i.e., a value of the most significant bit plane m other than the arithmetic escape symbol has been decoded, the variable of the spectral value “b” is defined to be equal to several (e.g. 2) most significant bits of the most significant bit plane value m, and the spectral value variable “a” is defined (e.g. 2) in the bits least significant of the value of the most significant bit plane m. Details regarding this functionality can be seen, for example, in reference number 312bb.

[000248] Subsequently, it is checked in step 312c whether an arithmetic stop symbol is present. This is the case if the value of the most significant bit plane m is equal to zero and the variable “lev” is greater than zero. Thus, an arithmetic stop condition is signaled by an “unusual” condition, in which the value of the most significant bit plane m is equal to zero, while the variable “lev” indicates that an increased numerical weight is associated with the value of the plane of bits most significant m. In other words, an arithmetic stop condition is detected if the bit stream indicates that an increased numeric weight, greater than a minimum numeric weight, should be given to a value of the most significant bit plane that is equal to zero, which is a condition that does not occur in a normal coding situation. In other words, an arithmetic stop condition is signaled if a coded arithmetic escape symbol is followed by a coded most significant bit plane value of 0.

[000249] After an evaluation if there is an arithmetic stop condition, which is performed in step 212c, the least significant bit planes are obtained, for example, as shown in reference number 212d in Figure 3. For each bit plane least significant, two binary values are decoded. One of the binary values is associated with variable a (or the first spectral value of a tuple of spectral values) and one of the binary values is associated with variable b (or a second spectral value of a tuple of spectral values). A number of least significant bit planes are designated by the variable lev.

[000250] When decoding one or more less significant bit planes (if any) an algorithm 212da is performed iteratively, in which a number of executions of the algorithm 212da is determined by the variable “lev”. It should be noted here that the first iteration of algorithm 212da is performed based on the values of variables a, b as defined in step 212bb. Further iterations of the 212da algorithm should be performed based on the updated variable values of variable a, b.

[000251] At the beginning of an iteration, a cumulative frequency table is selected. Subsequently, an arithmetic decoding is performed to obtain a value of a variable r, wherein the value of the variable r describes several less significant bits, for example, a less significant bit associated with variable a and a less significant bit associated with variable b. The function “ARITH_DECODE” (for example, as defined in Figure 5g) is used to obtain the r value, where the cumulative frequency table “arith_cf_r” is used for arithmetic decoding.

[000252] Subsequently, the values of variables a and b are updated. For this purpose, the variable a is shifted left by one bit, and the least significant bit of the shifted variable a is set to the value defined by the least significant bit of the value r. The variable b is shifted left by one bit, and the least significant bit of the shifted variable b is set to the value defined by bit 1 of the variable r, wherein bit 1 of the variable r has a numerical weight of 2 in the binary representation of the variable r. The 412ba algorithm is then repeated until all least significant bits are decoded.

[000253] After decoding the least significant bit planes, a matrix “x_ac_dec” is updated in which the values of the variables a, b are stored in the entries of said matrix having matrix indices 2*i and 2*i+1.

[000254] Subsequently, the context state is updated by calling the function “arith_update_context(i,a,b)”, details of which will be explained below with reference to figure 5g. Preferably, the function “arith_update_context(i,a,b)”, as defined in figure 5l, can be used.

[000255] Subsequent to updating the context state, which is performed in step 313, algorithms 312 and 313 are repeated, until the execution variable i reaches the value of lg/2 or an arithmetic stop condition is detected.

[000256] Subsequently, an “arith_finish()” finishing algorithm is performed, as can be seen in reference number 315. Details of the “arith_finish()” finishing algorithm will be described below with reference to figure 5m.

[000257] Subsequent to finishing algorithm 315, the spectral value signals are decoded using algorithm 314. As can be seen, the spectral value signals that are non-zero are individually encoded. In algorithm 314, signals are read for all spectral values having indices i between i=0 and i=lg-1 that are not zero. For each non-zero spectral value having a spectral value index i between i=0 and i=lg-1, a value (typically a single bit) s is read from the bit stream. If the value of s, which is read from the bit stream, is equal to 1, the sign of said spectral value is inverted. For this purpose, access is made to the matrix “x_ac_dec”, to determine whether the spectral value having the index i is equal to zero and to update the signal of the decoded spectral values. However, it should be noted that the signs of variables a, b are not changed in signal decoding 314.

[000258] By performing the finishing algorithm 315 before decoding signals 314, it is possible to reset all necessary bins after an ARITH_STOP symbol.

[000259] It should be noted here that the concept of obtaining the values of the least significant bit planes is not of particular relevance in some embodiments according to the present invention. In some embodiments, decoding of any least significant bit planes may be omitted. Alternatively, different decoding algorithms can be used for this purpose.

[000260] 11.2 Decoding order according to figure 4

[000261] Next, the order of decoding the spectral values will be described.

[000262] The quantized spectral coefficients “x_ac_dec[]” are silently encoded and transmitted (e.g., in the bit stream) starting from the coefficient with the lowest frequency and continuing to the coefficient with the highest frequency.

[000263] Consequently, the quantized spectral coefficients “x_ac_dec[]” are silently encoded starting from the lowest frequency coefficient and continuing to the highest frequency coefficient. The quantized spectral coefficients are decoded by groups of two successive (e.g., adjacent in frequency) coefficients a and b assembling into a double tuple (a,b) (also designated as {a,b}). It should be noted here that quantized spectral coefficients are sometimes designated as “qdec”.

[000264] The decoded coefficients “x_ac_dec[]” for a frequency domain mode (e.g., decoded coefficients for an advanced audio encoding, e.g., obtained using a modified discrete cosine transformation, as discussed in ISO/IEC 14496 , part 3, subpart 4) are then stored in an array “x_ac_quant[g][win][sfb][bin]”. The transmission order of silently encoded passwords is such that when they are decoded in the order received and stored in the array, “bin” is the fastest-increasing index, and “g” is the slowest-increasing index. Within a password, the decryption order is a,b (that is, a, then b).

[000265] The decoded coefficients “x_ac_dec[]” for the transformation coded excitation (TCX) are stored, for example, directly in an array “x_tcx_invquant[win][bin]”, and a silent coding password transmission order is such that when they are decoded in the order received and stored in the array “bin” is the index that increases most quickly, and “win” is the index that increases most slowly. Within a password, a decryption order is a, b. In other words, if the spectral values describe a transformation-encoded excitation of the linear prediction filter of a speech encoder, the spectral values a, b are associated with the adjacent and rising frequencies of the transformation-encoded excitation. Spectral coefficients associated with a lower frequency are typically encoded and decoded before a spectral coefficient associated with a higher frequency.

[000266] Notably, the audio decoder 200 can be configured to apply the decoded frequency domain representation 232, which is provided by the arithmetic decoder 230, to a “direct” generation of a time domain audio signal representation that utilizes a frequency-domain to time-domain signal transformation and for an “indirect” provision of a time-domain audio signal representation that utilizes a time-domain frequency-domain decoder and an excited linear prediction filter by the output of the signal transformer from frequency domain to time domain.

[000267] In other words, the arithmetic decoder, the functionality of which is discussed here in detail, is well suited to decoding spectral values of a time-frequency domain representation of an audio content encoded in the frequency domain, and to the provision of a time-frequency domain representation of a stimulus signal to a linear prediction filter adapted to decode (or synthesize) a speech signal encoded in the linear prediction domain. Thus, the arithmetic decoder is well suited for use in an audio decoder that can handle both frequency domain encoded audio content and linear predictive frequency domain encoded audio content (transformation-domain encoded excitation mode). linear).

[000268] 11.3 Context initialization according to figures 5a and 5b

[000269] Next, context initialization (also referred to as a “context mapping”), which is performed in step 310, will be described.

[000270] Context initialization comprises a mapping between a previous context and a current context according to the “arith_map_context()” algorithm, a first example of what is shown in figure 5a and a second example of what is shown in figure 5b .

[000271] As can be seen, the current context is stored in a global variable “q[2][n_context]” which takes the form of a matrix having a first dimension of 2 and a second dimension of “n_context”. A previous context can optionally (but not necessarily) be stored in a variable “qs[n_context]” which takes the form of the table having a dimension of “n_context” (if used).

[000272] With reference to the example algorithm “arith_map_context” in figure 5a, the input variable N describes a length of a current window and the input variable “arith_reset_flag” indicates whether the context should be reset. Furthermore, the global variable “previous_N” describes a length of a previous window. It should be noted here that typically a number of spectral values associated with a window is at least approximately equal to half a length of said window in terms of time domain samples. Furthermore, it should be noted that a 2-tuple number of spectral values is consequently at least approximately equal to a quarter of a length of said window in terms of time domain samples.

[000273] First, it should be noted that the “arith_reset_flag” indicator determines whether the context should be reset.

[000274] Taking the example in figure 5a as a reference, context mapping can be carried out according to the “arith_map_context()” algorithm. It should be noted here that the “arith_map_context()” function sets the entries “q[0][j]” of the current context matrix q to zero for j=0 to j=N/4-1, if the “arith_reset_flag” indicator is active and consequently indicates that the context must be redefined. Otherwise, that is, if the “arith_reset_flag” indicator is inactive, the entries “q[0][j]” of the current context matrix q are derived from the entries “q[1][k]” of the current context matrix q. It should be noted that the function “arith_map_context()” according to figure 5a sets the entries “q[0][j]” of the current context matrix q to the values “q[1][k]” of the context matrix current q, if the number of spectral values associated with the current (e.g., frequency domain encoded) audio structure is identical to the number of spectral values associated with the previous audio structure for j=k=0 to j=k=N /4-1.

[000275] A more complicated mapping is performed if the number of spectral values that are associated with the current audio structure is different from the number of spectral values that are associated with the previous audio structure. However, details regarding the mapping in this case are not particularly relevant to this main idea of the present invention, so reference is made to the pseudo-program code of Figure 5a for details.

[000276] Additionally, an initialization value for the current numeric context value c is returned by the “arith_map_context()” function. This initialization value is, for example, equal to the value of the input “q[0][0]” shifted left by 12 bits. Thus, the numeric context (current) value c is appropriately initialized for an iterative update.

[000277] Furthermore, figure 5b shows another example of an “arith_map_context()” algorithm that can alternatively be used. For details, reference is made to the pseudo-program code in Figure 5b.

[000278] To summarize the above, the “arith_reset_flag” indicator determines whether the context should be reset. If the indicator is true, a reset sub-algorithm 500a of the “arith_map_context()” algorithm is called. Alternatively, however, if the “arith_reset_flag” indicator is inactive (which indicates that no context reset should be performed), the decoding process begins with an initialization phase where the context element vector (or matrix) q is updated by copying and mapping the previous structure context elements stored in q[1][] into q[0][]. The context elements within q are stored in 4 bits by 2 tuples. The copying and/or mapping of the context element is performed in a sub-algorithm 500b.

[000279] Furthermore, it should be noted that if the context cannot be reliably determined, for example, if the previous structure data is not available, and if “arith_reset_flag” is not defined, then decoding of the spectral data cannot be continued and reading the current element “arith_data()” must be skipped.

[000280] In the example of figure 5b, the decoding process begins with an initialization phase where a mapping is made between the previous saved context stored in qs and the context of the current structure q. The previous context qs is stored in 2 bits per frequency line.

[000281] 11.4 Calculation of the state value according to figures 5c and 5d

[000282] Next, the calculation of the state value 312a will be described in more detail.

[000283] A first example algorithm will be described with reference to figure 5c and a second example algorithm will be described with reference to figure 5d.

[000284] It should be noted that the current numeric context value c (as shown in figure 3) can be obtained as a return value from the function “arith_get_context(c,i,N)”, a representation of the pseudo-program code which is shown in figure 5c. Alternatively, however, the current numeric context value c can be obtained as a return value from the function “arith_get_context(c,i)”, a representation of the pseudo-program code that is shown in Figure 5d.

[000285] Regarding the calculation of the state value, reference is also made to figure 4, which shows the context used for a state evaluation, that is, for the calculation of a numerical current context value c. Figure 4 shows a two-dimensional representation of spectral values, over time and frequency. An abscissa 410 describes time, and an ordinate 412 describes frequency. As can be seen in Figure 4, a tuple 420 of spectral values to decode (preferably using the current numeric context value), is associated with a time index t0 and a frequency index i. As can be seen, for time index t0, the tuples having frequency indexes i-1, i-2, and i-3 are already decoded at the time whose spectral values of tuple 120, having frequency index i, must be decoded. As can be seen from Figure 4, a spectral value 430 having a time index t0 and a frequency index i-1 is already decoded before the tuple 420 of spectral values are decoded, and the tuple 430 of spectral values is considered for the context that is used for decoding the 420-tuple of spectral values. Similarly, a tuple 440 of spectral values having a time index t0-1 and a frequency index of i-1, a tuple 450 of spectral values having a time index t0-1 and a frequency index of i, and a tuple 460 of spectral values having a time index t0-1 and a frequency index of i+1, is already decoded before the tuple 420 of spectral values is decoded, and are considered for determining the context, which is used to decode the 420-tuple of spectral values. The spectral values (coefficients) already decoded at the time when the spectral values of tuple 420 are decoded and considered for context are shown by a shaded square. In contrast, some other spectral values already decoded (at the time when the spectral values of tuple 420 are decoded) but not considered for context (for decoding the spectral values of tuple 420) are represented by squares having dashed lines, and others Spectral values (which are not yet decoded in time when the spectral values of tuple 420 are decoded) are shown by circles having dashed lines. Tuples represented by squares having dashed lines and tuples represented by circles having dashed lines are not used to determine the decoding context of the spectral values of tuple 420.

[000286] However, it should be noted that some of these spectral values, which are not used for the “regular” or “normal” calculation of the decoding context of the spectral values of tuple 420 can, however, be evaluated for the detection of different values previously decoded adjacent spectra that fulfill, individually or together, a predetermined condition regarding their magnitudes. Details regarding this issue will be discussed below.

[000287] Now with reference to figure 5c, details of the “arith_get_context(c,i,N)” algorithm will be described. Figure 5c shows the functionality of said function “arith_get_context(c,i,N)” in the form of a pseudo-program code, which uses the conventions of the well-known C language and/or C++ language. Thus, other details regarding the calculation of the current numeric context value “c” which is performed by the function “arith_get_context(c,i,N)” will be described.

[000288] It should be noted that the function “arith_get_context(c,i,N)” receives, as input variables, an “old state context”, which can be described by a numerical previous context value c. The function “arith_get_context(c,i,N)” also receives, as an input variable, an index i of a double tuple of spectral values to decode. The index i is typically a frequency index. An input variable N describes a window length of a window, in which spectral values are decoded.

[000289] The function “arith_get_context(c,i,N)” provides, as an output value, an updated version of the input variable c, which describes an updated state context, and which can be considered as a context value numeric current. To summarize, the function “arith_get_context(c,i,N)” takes a previous numeric context value c as an input variable and provides an updated version of it, which is considered as a current numeric context value. Furthermore, the “arith_get_context” function considers the variables i, N, and also evaluates the “global” matrix q[][].

[000290] Referring to the details of the function “arith_get_context(c,i,N)”, it should be noted that the variable c, which initially represents the numeric previous context value in a binary form, is shifted to the right by 4 bits in a step 504a. Thus, the four least significant bits of the numeric prior context value (represented by the input variable c) are discarded. Furthermore, the numeric weights of other bits of the numeric prior context values are reduced, for example, by a factor of 16.

[000291] Furthermore, if the index i of the double tuple is less than N/4-1, that is, it does not assume a maximum value, the current numeric context value is modified in that the value of the input q[0][ i+1] is added to bits 12 to 15 (i.e., to bits having a numerical weight of 212, 213, 214, and 215) of the changed context value that is obtained in step 504a. For this purpose, the entry q[0][i+1] of the matrix q[][] (or, more precisely, a binary representation of the value represented by said entry) is left-shifted by 12 bits. The changed version of the value represented by the input q[0][i+1] is then added to the context value c, which is derived in step 504a, i.e., the numeric representation of the previous numeric context value one bit changed ( shifted right by 4 bits). It should be noted here that the entry q[0][i+1] of the matrix q[][] represents a subregion value associated with a prior piece of audio content (e.g., a piece of audio content having time index t0-1, as defined with reference to figure 4), and with a higher frequency (e.g., a frequency having a frequency index i+1, as defined with reference to figure 4) than the tuple of values spectral data currently being decoded (which uses the current numeric context value c issued by the “arith_get_context(c,i,N)” function). In other words, if the 420-tuple of spectral values is to be decoded using the current numeric context value, the input q[0][i+1] can be based on the 460-tuple of previously decoded spectral values.

[000292] A selective addition of the entry q[0][i+1] of the matrix q[][] (shifted left by 12 bits) is shown in reference number 504b. As can be seen, the addition of the value represented by the input q[0][i+1] is naturally performed only if the frequency index i does not designate a tuple of spectral values having the highest frequency index i=N/4 -1.

[000293] Subsequently, in a step 504c, a Boolean AND operation is performed, in which the value of variable c is AND- combined with a hexadecimal value of 0xFFF0 to obtain an updated value of variable c. By performing such an AND operation, the four least significant bits of the variable c are effectively set to zero.

[000294] In a step 504d, the value of the input q[1][i-1] is added to the value of the variable c, which is obtained by step 504c, to then update the value of the variable c. However, said update of variable c in step 504d is only performed if the frequency index i of the double tuple to decode is greater than zero. It should be noted that the input q[1][i-1] is a context subregion value based on a tuple of previously decoded spectral values of the current part of the audio content for frequencies lower than the frequencies of the spectral values to be decoded using the current numeric context value. For example, the entry q[1][i-1] of the matrix q[][] can be associated with the tuple 430 having the time index t0 and the frequency index i-1, if it is assumed that the tuple 420 of Spectral values must be decoded using the current numeric context value returned by the current execution of the “arith_get_context(c,i,N)” function.

[000295] To summarize, bits 0, 1, 2, and 3 (i.e., a portion of four least significant bits) of the numeric prior context value are discarded in step 504a by toggling them from the binary numeric representation of the context value numerical preview. Furthermore, bits 12, 13, 14, and 15 of the changed variable c (i.e., of the previous numeric changed context value) are set to have values defined by the value of the context subregion q[0][i+ 1] in step 504b. Bits 0, 1, 2, and 3 of the numeric prior changed context value (that is, bits 4, 5, 6, and 7 of the original numeric prior context value) are overwritten by the value of the context subregion q[ 1][i-1] in steps 504c and 504d.

[000296] Consequently, it can be said that bits 0 to 3 of the numeric prior context value represent the value of the context subregion associated with tuple 432 of spectral values, bits 4 to 7 of the numeric prior context value represent the value of the context subregion associated with a tuple 434 of previously decoded spectral values, bits 8 to 11 of the numeric prior context value represent the value of the context subregion associated with the tuple 440 of previously decoded spectral values, and the bits 12 to 15 of the numerical prior context value represent a context subregion value associated with the 450-tuple of previously decoded spectral values. The numeric prior context value, which is entered in the function “arith_get_context(c,i,N)”, is associated with a decoding of the 430 tuple of spectral values.

[000297] The current numeric context value, which is obtained as an output variable of the function “arith_get_context(c,i,N)”, is associated with a decoding of the tuple 420 of spectral values. Thus, bits 0 to 3 of the numeric current context value describe the value of the context subregion associated with the 430 tuple of the spectral values, bits 4 to 7 of the numeric current context value describe the value of the context subregion associated with the tuple 440 of spectral values, bits 8 to 11 of the numeric current context value describe the numeric value of the subregion associated with the tuple 450 of spectral value, and bits 12 to 15 of the numeric current context value describe the value of the context subregion associated with the 460-tuple of spectral values. Thus, it can be seen that a part of the numeric prior context value, namely bits 8 to 15 of the numeric prior context value, are also included in the numeric current context value, such as bits 4 to 11 of the numeric prior context value. numeric current. In contrast, bits 0 through 7 of the numeric prior current context value are discarded when deriving the numeric representation of the numeric current context value from the numeric prior context value representation.

[000298] In a step 504e, the variable c representing the current numeric context value is selectively updated if the frequency index i of the double tuple to decode is greater than a predetermined number of, for example, 3. In this case, either that is, if i is greater than 3, it is determined whether the sum of the values of the context subregion q[1][i-3], q[1][i-2], and q[1][i-1 ] is less than (or equal to) a predetermined value of, for example, 5. If it is discovered that the sum of said context subregion values is less than said predetermined value, a hexadecimal value of, e.g. 0x10000, is added to variable c. Thus, the variable c is defined such that the variable c indicates whether there is a condition in which the context subregion values q[1][i-3], q[1][i-2], and q[ 1][i-1] comprise a particularly small sum value. For example, bit 16 of the current numeric context value may act as an indicator to indicate such a condition.

[000299] To conclude, the return value of the function “arith_get_context(c,i,N)” is determined by steps 504a, 504b, 504c, 504d, and 504e, where the current numeric context value is derived from the context value numerical prior in steps 504a, 504b, 504c, and 504d, and in which an indicator indicating an environment of previously decoded spectral values having, on average, particularly small absolute values, is derived in step 504e and added to the variable c. Thus, the value of the variable c obtained in steps 504a, 504b, 504c, 504d is returned, in a step 504f, as a return value of the function “arith_get_context(c,i,N)”, if a condition evaluated in step 504e is not fulfilled. In contrast, the value of variable c, which is derived in steps 504a, 504b, 504c, and 504d, is increased by the hexadecimal value of 0x10000 and the result of this increase operation is returned, in step 504e, if the condition evaluated in step 540e is fulfilled.

[000300] To summarize the above, it should be noted that the silent decoder outputs 2 tuples of unsigned quantized spectral coefficients (as will be described in more detail below). First, the state c of the context is calculated based on the previously decoded spectral coefficients “around” the double tuple to decode. In a preferred embodiment, the state (which is, for example, represented by a numeric context value) is further updated using the context state of the last decoded double tuple (which is designated as a numeric prior context value), considering only two new double tuples (e.g. 2 tuples 430 and 460). The state is encoded in 17 bits (for example, using a numeric representation of a current numeric context value) and is returned by the “arith_get_context()” function. For details, reference is made to the program code representation of Figure 5c.

[000301] Furthermore, it should be noted that a pseudo-program code of an alternative realization of the “arith_get_context()” function is shown in figure 5d. The function “arith_get_context(c,i)” according to figure 5d is similar to the function “arith_get_context(c,i,N)” according to figure 5c. However, the function “arith_get_context(c,i)” according to figure 5d does not comprise a special handling or decoding of tuples of spectral values that comprise a minimum frequency index of i=0 or a maximum frequency index of i= N/4-1.

[000302] 11.5 Mapping rule selection

[000303] Next, the selection of a mapping rule, for example, the cumulative frequency table that describes a mapping of a password value to a symbol code, will be described. The selection of the mapping rule is made depending on a context state, which is described by the numeric current context value c.

[000304] 11.5.1 Selection of the mapping rule using the algorithm according to figure 5e

[000305] Next, the selection of a mapping rule using the “arith_get_pk(c)” function will be described. It should be noted that the function “arith_get_pk()” is called at the beginning of sub-algorithm 312ba when decoding a code value “acod_m” to provide a tuple of spectral values. It should be noted that the function “arith_get_pk(c)” is called with different arguments in different iterations of algorithm 312b. For example, in a first iteration of algorithm 312b, the function “arith_get_pk(c)” is called with an argument that is equal to the current numeric context value c, provided by the previous execution of the function “arith_get_context(c,i,N) ” in step 312a. In contrast, in other iterations of the 312ba sub-algorithm, the function “arith_get_pk(c)” is called with an argument that is a sum of the current numeric context value c provided by the function “arith_get_context(c,i,N)” in step 312a, and a bit-shifted version of the value of the variable “esc_nb”, in which the value of the variable “esc_nb” is left-shifted by 17 bits. Thus, the current numeric context value c provided by the function “arith_get_context(c,i,N)” is used as an input value of the function “arith_get_pk()” in the first iteration of the 312ba algorithm, that is, in decoding values comparably small spectra. In contrast, when decoding comparably larger spectral values, the input variable of the “arith_get_pk()” function is modified where the value of the variable “esc_nb” is taken into account, as shown in figure 3.

[000306] Now referring to figure 5e, which shows a representation of the pseudo-program code of a first realization of the function “arith_get_pk(c)”, it should be noted that the function “arith_get_pk()” receives the variable c as a input value, where the variable c describes the state of the context, and where the input variable c of the “arith_get_pk()” function is equal to the current numeric context value provided as a return variable of the “arith_get_context()” function ” at least in some situations. Furthermore, it should be noted that the function “arith_get_pk()” provides, as an output variable, the variable “pki”, which describes an index of a probability model and which can be considered as an index value of the rule of thumb. mapping.

[000307] With reference to figure 5e, it can be seen that the function “arith_get_pk()” comprises an initialization of the variable 506a, in which the variable “i_min” is initialized to have the value of -1. Similarly, the variable i is defined to be equal to the variable “i_min”, so that the variable i is also initialized to a value of -1. The variable “i_max” is initialized to assume a value that is smaller, by 1, than the number of entries in the table “ari_lookup_m[]” (details that will be described with reference to figure 21). Thus, the variables “i_min” and “i_max” define an interval. For example, i_max can be initialized to the value 741.

[000308] Subsequently, a search 506b is performed to identify an index value designating an entry of the table “ari_hash_m”, such that the value of the input variable c of the function “arith_get_pk()” within a range defined by said entrance and an adjacent entrance.

[000309] In search 506b, a sub-algorithm 506ba is repeated as long as a difference between the variables “i_max” and “i_min” is greater than 1. In the sub-algorithm 506ba, the variable i is defined to be equal to one-half arithmetic of the values of the variables “i_min” and “i_max”. Consequently, the variable i designates a table entry “ari_hash_m[]” (as defined in the tabular representation of figures 22(1), 22(2), 22(3) and 22(4)) in the middle of a range of the table defined by the values of the variables “i_min” and “i_max”. Subsequently, variable j is set to be equal to the value of the entry “ari_hash_m[i]” of the table “ari_hash_m[]”. through a table range defined by the variables “i_min” and “i_max”. Subsequently, the range defined by the variables “i_min” and “i_max” is updated if the value of the input variable c of the function “arith_get_pk()” is different from a state value defined by the upper bits of the table entry “j=ari_hash_m [i]” of the “ari_hash_m[]” table. For example, the “high bits” (bits 8 and above) of the “ari_hash_m[]” table entries describe meaningful state values. Thus, the value “j>>8” describes a meaningful state value represented by the entry “j=ari_hash_m[i]” of the table “ari_hash_m[]” designated by the hash-value table of index i. Thus, if the value of variable c is less than the value “j>>8”, this means that the state value described by variable c is less than a significant state value described by the table entry “ari_hash_m[i]” “ari_hash_m[]”. In this case, the value of the variable “i_max” is set to be equal to the value of the variable i, which in turn has the effect that a range size defined by “i_min” and “i_max” is reduced, whereby the new range is approximately equal to the lower half of the previous range. If it is observed that the input variable c of the function “arith_get_pk()” is greater than the value “j>>8”, which means that the context value described by the variable c is greater than a significant state value described by the input “ari_hash_m[i]” from the array “ari_hash_m[]”, the value of the variable “i_min” is set to be equal to the value of the variable i. Thus, the size of the interval defined by the values of the variables “i_min” and “i_max” is reduced to approximately half the size of the previous interval, defined by the previous values of the variables “i_min” and “i_max”. To be more precise, the interval defined by the updated value of the variable “i_min” and the previous value of the (unchanged) variable “i_max” is approximately equal to the upper half of the previous interval in the case where the value of the variable c is greater than the meaningful state value defined by the “ari_hash_m[i]” entry.

[000310] If, however, it is observed that the context value described by the input variable c of the algorithm “arith_get_pk()” is equal to the meaningful state value defined by the input “ari_hash_m[i]” (i.e. c== (j>>8)), a mapping rule index value defined by the lower 8 bits of the “ari_hash_m[i]” input is returned as the return value of the “arith_get_pk()” function (“return (j&0xFF) )”).

[000311] To summarize the above, an input “ari_hash_m[i]”, the upper bits (bits 8 and above) of which describe a meaningful state value, is evaluated at each 506ba iteration, and the context value (or numeric current context value) described by the input variable c of the function “arith_get_pk()” is compared with the meaningful state value described by said table entry “ari_hash_m[i]”. If the context value represented by the input variable c is less than the meaningful state value represented by the table entry “ari_hash_m[i]”, the upper limit (described by the value “i_max”) of the table range is reduced, and if the context value described by the input variable c is greater than the meaningful state value described by the table entry “ari_hash_m[i]”, the lower bound (which is described by the value of the variable “i_min”) of the table range is increased. In both said cases, sub-algorithm 506ba is repeated unless the size of the interval (defined by the difference between “i_max” and “i_min”) is less than, or equal to, 1. If, in contrast, the context value described by the variable c is equal to the meaningful state value described by the table entry “ari_hash_m[i]”, the function “arith_get_pk()” is aborted, where the return value is defined by the lowest 8 bits of the table entry “ari_hash_m[i]”.

[000312] If, however, the 506b search is terminated because the range size reaches its minimum value (“i_max” - “i_min” is less than, or equal to, 1), the return value of the function “arith_get_pk( )” is determined by an entry “ari_lookup_m[i_max]” of the table “ari_lookup_m[]”, which can be seen in reference number 506c. The ari_lookup_m[] table is preferably chosen as defined in the table representation of figure 21, and can then be equal to the ari_lookup_m[742] table. Thus, the entries of the “ari_hash_m[]” table (which is preferably the same as the ari_hash_m[742] table as defined in figures 22(1), 22(2), 22(3), 22(4)) define both values of significant state and interval limits. In sub-algorithm 506ba, the search range limits “i_min” and “i_max” are adapted iteratively, so that the entry “ari_hash_m[i]” of the table “ari_hash_m[]”, an index of the hash table i which remains at least approximately in the center of the search range defined by the range boundary values “i_min” and “i_max”, at least approximates a context value described by the input variable c. It is then reached that the context value described by the input variable c within a range defined by “ari_hash_m[i_min]” and “ari_hash_m[i_max]” after the completion of the iterations of sub-algorithm 506ba, unless the value of context described by the input variable c is equal to a meaningful state value described by an entry in the “ari_hash_m[]” table.

[000313] If, however, the iterative repetition of sub-algorithm 506ba is terminated because the size of the interval (defined by “i_max” - i_min”) reaches or exceeds its minimum value, it is assumed that the context value described by input variable c is not a meaningful state value. In this case, the index “i_max”, which designates an upper limit of the range, is however used. The upper value “i_max” of the range, which is reached in the last iteration of the 506ba sub-algorithm, is reused as a table index value for an access to the “ari_lookup_m” table (which may be the same as the ari_lookup_m[742] table of figure 21). The “ari_lookup_m[]” table describes the mapping rule index values associated with ranges of multiple adjacent numeric context values. The ranges, in which the mapping rule index values described by the “ari_lookup_m[]” table entries are associated, are defined by the significant state values described by the “ari_hash_m[]” table entries. ” define both significant state values and the limits of adjacent numeric context value ranges. In executing algorithm 506b, it is determined whether the numeric context value described by the input variable c is equal to a significant state value, and if this is not the case, whose range of numeric context values (of the various ranges, limits of which are defined by the meaningful state values) the context value described by the input variable c remains. Thus, algorithm 506b fulfills a dual functionality to determine whether the input variable c describes a meaningful state value and, if not the case, to identify a range, bounded by meaningful state values, in which the represented context value by the input variable c remains. Thus, the 506e algorithm is particularly efficient and requires only a comparably small number of table accesses.

[000314] To summarize the above, the context state c determines the cumulative frequency table used to decode the most significant bitwise plane m. The mapping of c to the corresponding cumulative frequency table index “pki” as performed by the “arith_get_pk()” function. A representation of the pseudo-program code of said function “arith_get_pk()” was explained with reference to figure 5e.

[000315] To further summarize the above, the value m is decoded using the function “arith_decode()” (which is described in more detail below) called with the cumulative frequency table “arith_cf_m[pki][]”, where “ pki” corresponds to the index (also designated as mapping rule index value) returned by the function “arith_get_pk()”, which is described with reference to Figure 5e in the form of pseudo C code.

[000316] 11.5.2 Selection of the mapping rule using the algorithm according to figure 5f

[000317] In the following, another realization of a selection of the mapping rule algorithm “arith_get_pk()” will be described with reference to figure 5f which shows a representation of the pseudo-program code of such an algorithm, which can be used in decoding a tuple of spectral values. The algorithm, according to figure 5f, can be considered an optimized version (e.g., speed optimized version) of the algorithm, “get_pk()” or the “arith_get_pk()” algorithm.

[000318] The “arith_get_pk()” algorithm according to figure 5f receives, as an input variable, a variable c that describes the state of the context. The input variable c can, for example, represent a numeric current context value.

[000319] The “arith_get_pk()” algorithm provides, as an output variable, a variable “pki”, which describes the index of a probability distribution (or probability model) associated with a state of the context described by the input variable w. The variable “pki” can, for example, be a mapping rule index value.

[000320] The algorithm according to figure 5f comprises a definition of the contents of the “i_diff[]” matrix. As can be seen, a first entry of the array “i_diff[]” (having an array index 0) is equal to 299 and other entries of the array (having array indices 1 to 8) take the values of 149, 74, 37 , 18, 9, 4, 2, and 1. Thus, the step size for selecting a value from the “i_min” hash table index is reduced with each iteration, as the entries of the “i_diff[]” arrays define the said step sizes. For details, reference is made to the discussion below.

[000321] However, different step sizes, for example, different contents of the array “i_diff[]” can generally be chosen, whereby the contents of the array “i_diff[]” can naturally be adapted to the size of the hash table “ari_hash_m[ i]”.

[000322] It should be noted that the variable “i_min” is initialized to assume a correct value of 0 at the beginning of the “arith_get_pk()” algorithm.

[000323] In an initialization step 508a, a variable s is initialized depending on the input variable c, wherein a numerical representation of the variable c is left shifted by 8 bits in order to obtain the numerical representation of the variable s.

[000324] Subsequently, a table lookup 508b is performed in order to identify a hash table index value “i_min” from a hash table entry “ari_hash_m[]”, such that the context value described by the value of context c within a range that is bounded by the context value described by the hash table entry “ari_hash_m[i_min]” and a context value described by another hash table entry “ari_hash_m” whose other entry “ari_hash_m” is adjacent (in terms of its hash table index value) to the hash table entry “ari_hash_m[i_min]”. Thus, algorithm 508b allows determining an index value of the hash table “i_min” by assigning an entry “j=ari_hash_m[i_min]” of the hash table “ari_hash_m[]”, such that the hash table entry “ari_hash_m[i_min] ” at least approximate the context value described by the input variable c.

[000325] Table lookup 508b comprises an iterative execution of a sub-algorithm 508ba, wherein the sub-algorithm 508ba is executed for a predetermined number of, for example, nine iterations. In the first step of sub-algorithm 508ba, the variable i is set to a value that is equal to a sum of a value from a variable “i_min” and a value from a table entry “i_diff[k]”. It should be noted here that k is an execution variable, which is increased, starting from an initial value of k=0, with each iteration of the 508ba sub-algorithm. The “i_diff[]” array defines predetermine the increase values, where the increase values decrease with the increase in the table index k, that is, with the increase in the number of iterations.

[000326] In a second step of sub-algorithm 508ba, a value from a table entry “ari_hash_m[]” is copied into a variable j. Preferably, the upper bits of the “ari_hash_m[]” table entries describe meaningful state values of a numeric context value, and the lower bits (bits 0 to 7) of the “ari_hash_m[]” table entries describe index values. mapping rule associated with their respective significant state values.

[000327] In a third step of sub-algorithm 508ba, the value of variable S is compared with the value of variable j, and the variable “i_min” is selectively set to the value “i+1” if the value of variable s is greater than the value of variable j. Subsequently, the first step, the second step, and the third step of sub-algorithm 508ba are repeated for a predetermined number of times, e.g., nine times. Thus, in each execution of sub-algorithm 508ba, the value of the variable “i_min” is increased by i_diff[]+1, if and only if the context value described by the currently valid hash table index i_min + i_diff[] is less than the context value described by the input variable c. Thus, the hash table index value “i_min” is (iteratively) increased in each execution of sub-algorithm 508ba if (and only if) the context value described by the input variable c and, consequently, by the variable s, is greater than the context value described by the entry “ari_hash_m[i=i_min + diff[k]]”.

[000328] Furthermore, it should be noted that only a single comparison, namely, the comparison as if the value of variable s were greater than the value of variable j, is performed in each execution of sub-algorithm 508ba. Thus, the 508ba algorithm is particularly computationally efficient. Furthermore, it should be noted that there are different possible results regarding the final value of the “i_min” variable. For example, it is possible that the value of the variable “i_min” after the last execution of sub-algorithm 512ba is such that the context value described by the table entry “ari_hash_m[i_min]” is smaller than the context value described by input variable c, and that the context value described by the table entry “ari_hash_m[i_min +1]” is greater than the context value described by the input variable c. Alternatively, it may happen that after the last execution of sub-algorithm 508ba, the context value described by the hash table entry “ari_hash_m[i_min -1]” is smaller than the context value described by the input variable c, and that the context value described by the input “ari_hash_m[i_min]” is greater than the context value described by the input variable c. Alternatively, however, it may happen that the context value described by the hash table entry “ari_hash_m[i_min]” is identical to the context value described by the input variable c.

[000329] For this reason, a return value provision based on the 508c decision is made. The variable j is defined to have the value of the hash table entry “ari_hash_m[i_min]”. Subsequently, it is determined whether the context value described by the input variable c (and also by the variable s) is greater than the context value described by the input “ari_hash_m[i_min]” (first case defined by a condition “s>j” ), or if the context value described by the input variable c is smaller than the context value described by the hash table entry “ari_hash_m[i_min]” (second case defined by a condition “c<j>>8”), or if the context value described by the input variable c is equal to the context value described by the input “ari_hash_m[i_min]” (third case).

[000330] In the first case, (s>j), an entry “ari_lookup_m[i_min +1]” from the table “ari_lookup_m[]” designated by the table index value “i_min+1” is returned as the output value of the function “arith_get_pk()”. In the second case (c<(j>>8)), an entry “ari_lookup_m[i_min]” from table “ari_lookup_m[]” designated by the index value of table “i_min” is returned as the return value of the “arith_get_pk()” function. In the third case (that is, if the context value described by the input variable c is equal to the meaningful state value described by the table entry “ari_hash_m[i_min]”), a mapping rule index value described by the 8 bits lowermost hash table entry “ari_hash_m[i_min]” is returned as the return value of the “arith_get_pk()” function.

[000331] To summarize the above, a particularly simple table lookup is performed in step 508b, wherein the table lookup provides a variable value from a variable “i_min” without distinguishing whether the context value described by the input variable c is equal to a meaningful state value defined by one of the state table entries “ari_hash_m[]” or not. In step 508c, which is performed subsequent to table lookup 508b, a magnitude relationship between the context value described by the input variable c and a significant state value described by the hash table entry “ari_hash_m[i_min]” is evaluated, and the return value of the function “arith_get_pk()” is selected depending on a result of said evaluation, wherein the value of the variable “i_min”, which is determined in the evaluation of table 508b, is considered to select a rule index value mapping even if the context value described by the input variable c is different from the meaningful state value described by the hash table entry “ari_hash_m[i_min]”.

[000332] It should also be noted that the comparison in the algorithm should preferably (or alternatively) be made between the context index (numeric context value) c and j=ari_hash_m[i]>>8. Furthermore, each entry in the “ari_hash_m[]” table represents a context index, encoded beyond the 8th bit, and its corresponding probability model encoded in the first 8 bits (least significant bits). In the current implementation, we are mainly interested in knowing whether the present context c is greater than ari_hash_m[i]>>8, which is equivalent to detecting whether s=c<<8 is also greater than ari_hash_m[i].

[000333] To summarize the above, since the context state is calculated (which can, for example, be obtained using the “arith_get_context(c,i,N)” algorithm according to figure 5c, or the “arith_get_context(c,i,N)” algorithm arith_get_context(c,i)” according to figure 5d, the most significant bitwise plane is decoded using the “arith_decode” algorithm (which will be described below) called with the appropriate cumulative frequency table corresponding to the corresponding probability model to the context state.The matching is done by the “arith_get_pk()” function, for example, the “arith_get_pk()” function that was discussed with reference to figure 5f.

[000334] 11.6 Arithmetic decoding

[000335] 11.6.1 Arithmetic decoding using the algorithm according to figure 5g

[000336] In the following, the functionality of a preferred implementation of the “arith_decode()” function will be discussed in detail with reference to figure 5g. Figure 5g shows a C pseudo-code that describes the algorithm used.

[000337] It should be noted that the “arith_decode()” function uses the help function “arith_first_symbol (void)”, which returns TRUE if it is the first symbol in the sequence and FALSE otherwise. The “arith_decode()” function also uses the “arith_get_next_bit(void)” helper function, which obtains and provides the next bit of the bit stream.

[000338] Furthermore, the “arith_decode()” function uses the global variables “low”, “high” and “value”. Furthermore, the “arith_decode()” function receives, as an input variable, the variable “cum_freq[]”, which points towards a first entry or element (having the element index or entry index 0) of the selected cumulative frequencies or subtable of cumulative frequencies (preferably one of the ari_cf_m[pki=0][17] to ari_cf_m[pki=63][17] subtables of the ari_cf_m[64][17] subtable, as defined by the representation in table of figures 23(1), 23(2), 23(3)). Furthermore, the “arith_decode()” function uses the input variable “cfl”, which indicates the length of the selected cumulative frequency table or cumulative frequency sub-table designated by the variable “cum_freq[]”.

[000339] The “arith_decode()” function comprises, as a first step, an initialization of the variable 570a, which is performed if the “arith_first_symbol()” helper function indicates that the first symbol of a sequence of symbols is being decoded. Initializing the value 550a initializes the “value” variable depending on several, for example, 16 bits, which are obtained from the bit stream using the “arith_get_next_bit” helper function, so that the “value” variable has the value represented by said bits. Furthermore, the “low” variable is initialized to have the value of 0, and the “high” variable is initialized to have the value of 65535.

[000340] In a second step 570b, the variable “range” is set to a value, which is greater, by 1, than the difference between the values of the variables “high” and “low”. The “cum” variable is set to a value that represents a relative position of the value of the “value” variable between the value of the “low” variable and the value of the “high” variable. Thus, the “cum” variable has, for example, a value between 0 and 216 depending on the value of the “value” variable.

[000341] The pointer p is initialized to a value that is smaller, by 1, than the starting address of the selected cumulative frequency table or sub-table.

[000342] The “arith_decode()” algorithm also comprises an iterative search of the cumulative frequency table 570c. The iterative search of the cumulative frequency table is repeated until the variable cfl is less than or equal to 1. In the iterative search of the cumulative frequency table 570c, the pointer variable q is set to a value, which is equal to a sum of the current value of the pointer variable p and half of the value of the “cfl” variable. If the value of the entry *q of the selected cumulative frequency table, whose input is driven by the pointer variable q, is greater than the value of the variable “cum”, the pointer variable p is set to the value of the pointer variable q, and the variable “cfl” is increased. Finally, the “cfl” variable is shifted right by one bit, thus effectively dividing the value of the “cfl” variable by 2 and ignoring the modulus part.

[000343] Thus, the iterative search of the cumulative frequency table 570c effectively compares the value of the variable “cum” with several entries of the selected cumulative frequency table in order to identify a range within the selected cumulative frequency table, which is linked by the cumulative frequency table entries so that the cum value remains within the identified range. Thus, the entries of the selected cumulative frequency table define the intervals, where a respective symbol value is associated with each of the intervals of the selected cumulative frequency table. Furthermore, the widths of the intervals between two adjacent values of the cumulative frequency table define the probabilities of the symbols associated with said intervals, so that the selected cumulative frequency table in its entirety defines a probability distribution of different symbols (or symbol values ). Details regarding the available cumulative frequency tables will be discussed below with reference to figure 23.

[000344] Referring again to figure 5g, the value of the symbol is derived from the value of the pointer variable p, wherein the value of the symbol is derived as shown in reference number 570d. Thus, the difference between the value of the pointer variable p and the initial address “cum_freq” is evaluated in order to obtain the value of the symbol, which is represented by the variable “symbol”.

[000345] The “arith_decode” algorithm also comprises an adaptation 570e of the “high” and “low” variables. If the value of the symbol represented by the “symbol” variable is other than 0, the “high” variable is updated, as shown in reference number 570e. Furthermore, the value of the “low” variable is updated, as shown in reference number 570e. The “high” variable is set to a value that is determined by the value of the “low” variable, the “range” variable and the entry having the index “symbol -1” of the selected cumulative frequency table. The variable “low” is increased, where the magnitude of the increase is determined by the variable “range” and the selected cumulative frequency table entry having the index “symbol”. Thus, the difference between the values of the “low” and “high” variables is adjusted depending on the numerical difference between the two adjacent entries of the selected cumulative frequency table.

[000346] Thus, if a symbol value having a low probability is detected, the interval between the values of the “low” and “high” variables is reduced to narrow width. In contrast, if the value of the detected symbol comprises a relatively large probability, the width of the interval between the values of the “low” and “high” variables is adjusted to the comparably large value. Again, the width of the interval between the “low” and “high” variable values is dependent on the detected symbol and the corresponding cumulative frequency table entries.

[000347] The “arith_decode()” algorithm also comprises a renormalization of the range 570f, in which the range determined in step 570e is iteratively changed and scaled until the “break” condition is reached. In the renormalization of the 570f range, a selective 570fa shift-down operation is performed. If the “high” variable is less than 32768, nothing is done, and the range renormalization continues with an operation to increase the size of the 570fb range. If, however, the “high” variable is not less than 32768 and the “low” variable is greater than or equal to 32768, the “values”, “low” and “high” variables are all reduced to 32768, so that a range defined by the variables “low” and “high” is changed downwards, and so that the value of the variable “value” is also changed downwards. If, however, it is observed that the value of the “high” variable is not less than 32768, and that the “low” variable is not greater than or equal to 32768, and that the “low” variable is greater than or equal to 16384 and that the “high” variable is less than 49152, the “value”, “low” and “high” variables are all reduced to 16384, thus reducing the interval between the values of the “high” and “low” variables and also the value of the variable “value”. If, however, none of the above conditions are met, range renormalization is aborted.

[000348] If, however, any of the conditions mentioned above, which are evaluated in step 570fa, are met, the range increase operation 570fb is performed. When increasing the 570fb range, the value of the “low” variable is doubled. Furthermore, the value of the “high” variable is doubled, and the result of the doubling operation is increased by 1. Furthermore, the value of the “value” variable is doubled (shifted to the left by one bit), and one bit of the bits, which is obtained by the help function “arith_get_next_bit” is used as the least significant bit. Thus, the size of the interval between the values of the “low” and “high” variables is approximately doubled, and the precision of the “value” variable is increased by using a new bit of the bit stream. As mentioned above, steps 570fa and 570fb are repeated until the “break” condition is reached, that is, until the interval between the values of the “low” and “high” variables is large enough.

[000349] With reference to the functionality of the “arith_decode()” algorithm, it should be noted that the range between the values of the “low” and “high” variables is reduced in step 570e depending on two adjacent entries of the cumulative frequency table referenced by the variable “cum_freq”. If an interval between two adjacent values of the selected cumulative frequency table is small, that is, if the adjacent values are comparably close together, the interval between the values of the variables “low” and “high”, which is obtained in step 570e, will be comparably small. In contrast, if two adjacent entries of the cumulative frequency table are spaced further apart, the interval between the values of the “low” and “high” variables, which is obtained in step 570e, will be comparably large.

[000350] Consequently, if the interval between the values of the variables “low” and “high”, which is obtained in step 570e, is comparably small, a large number of renormalization of interval steps will be performed to resize the interval to the size “ sufficient” (so that none of the conditions of a 570fa condition assessment are met). Thus, a comparably large number of bits from the bit stream will be used in order to increase the precision of the “value” variable. If, in contrast, the size of the interval obtained in step 570e is comparably large, only a smaller number of repetitions of the interval normalization steps 570fa and 570fb will be necessary in order to renormalize the interval between the values of the variables “low” and “ high” to “sufficient” size. Thus, only a comparably small number of bits from the bit stream will be used to increase the precision of the “value” variable and to prepare a decoding of the next symbol.

[000351] To summarize the above, if a symbol is decoded, which comprises a comparably high probability, and in which a large range is associated by the selected cumulative frequency table entries, only a comparably small number of bits will be read from the stream of bits to allow decoding of a subsequent symbol. In contrast, if a symbol is decoded which comprises a comparably small probability and in which a small range is associated by the selected cumulative frequency table entries, a comparably large number of bits will be considered from the bit stream in order to prepare a decode. of the next symbol.

[000352] Thus, the entries of the cumulative frequency tables reflect the probabilities of different symbols and also reflect a number of bits required to decode a sequence of symbols. By varying the cumulative frequency table depending on a context, i.e. depending on previously decoded symbols (or spectral values), for example by selecting different cumulative frequency tables depending on the context, stochastic dependencies between different symbols can be explored, which allow efficient encoding of the particular bitrate of subsequent (or adjacent) symbols.

[000353] To summarize the above, the function “arith_decode()”, which was described with reference to figure 5g, is called with the cumulative frequency table “arith_cf_m[pki][]”, corresponding to the index “pki” returned by a function “arith_get_pk()” to determine the value of the most significant bit plane m (which can be set to the value of the symbol represented by the return variable “symbol”).

[000354] To summarize the above, the arithmetic decoder is an integer implementation using the scaled identification generation method. For details, reference is made to the book “Introduction to Data Compression” by K. Sagood, Third Edition, 2006, Elsevier Inc.

[000355] The computer program code according to figure 5g describes the algorithm used according to an embodiment of the invention.

[000356] 11.6.2 Arithmetic decoding using the algorithm according to figures 5h and 5i

[000357] Figures 5h and 5i show a representation of the pseudo-program code of another realization of the “arith_decode()” algorithm, which can be used as an alternative to the “arith_decode” algorithm described with reference to Figure 5g.

[000358] It should be noted that the algorithms according to figure 5g and figures 5h and 5i can be used in the “values_decode()” algorithm according to figure 3.

[000359] To summarize, the value m is decoded using the function “arith_decode()” called with the cumulative frequency table “arith_cf_m[pki][]” (which is preferably a sub-table of the ari_cf_m[67] table). [17] defined in the table representation of figures 23(1), 23(2), 23(3)) where “pki” corresponds to the index returned by the “arith_get_pk()” function. The arithmetic encoder (or decoder) is an integer implementation using the scaled ID generation method. For details, reference is made to the book “Introduction to Data Compression” by K. Sagood, Third Edition, 2006, Elsevier Inc. The computer program code according to figure 5h and 5i describes the algorithm used.

[000360] 11.7 Exhaust mechanism

[000361] Next, the escape mechanism, which is used in decoding the “values_decode()” algorithm according to figure 3, will be briefly discussed.

[000362] When the decoded value m (which is provided as a return value of the “arith_decode()” function) is the escape symbol “ARITH_ESCAPE”, the variables “lev” and “esc_nb” are increased by 1, and the another value m is decoded. In this case, the function “arith_get_pk()” (or “get_pk()”) is called once again with the value “c+ esc_nb<<17” as input argument, where the variable “esc_nb” describes the number of symbols of escape previously decoded to the same double tuple and bounded to 7.

[000363] To summarize, if an escape symbol is identified, the value of the most significant bit plane m is assumed to comprise an increased numerical weight. Furthermore, the current numeric decoding is repeated, where a modified numeric current context value “c+ esc_nb<<17” is used as an input variable to the “arith_get_pk()” function. Thus, a different mapping rule index value “pki” is typically obtained in different iterations of sub-algorithm 312ba.

[000364] 11.8 Arithmetic stop mechanism

[000365] Next, the arithmetic stop mechanism will be described. The arithmetic stopping mechanism allows reduction of the number of bits required in the case where the upper frequency part is completely quantized to 0 in an audio encoder.

[000366] In one embodiment, an arithmetic stopping mechanism can be implemented as follows: since the value m is not the escape symbol, “ARITH_ESCAPE”, the decoder checks whether the successive m forms an “ARITH_STOP” symbol. If a condition “(esc_nb >0&&m==0)” is true, the “ARITH_STOP” symbol is detected and the decoding process is terminated. In this case, the decoder goes directly to the signal decoding described below or the “arith_finish()” function that will be described below. A condition means that the rest of the structure is made up of 0 values.

[000367] 11.9 Decoding the least significant bit plane

[000368] Next, the decoding of one or more less significant bit planes will be described. Decoding the least significant bit plane is performed, for example, in step 312d shown in Figure 3. Alternatively, however, the algorithms as shown in Figures 5j and 5n can be used, wherein the algorithm of Figure 5j is a preferred algorithm.

[000369] 11.9.1 Decoding the least significant bit plane according to figure 5j

[000370] Now referring to figure 5j, it can be seen that the values of variables a and b are derived from the value m. For example, the numeric representation of the value m is shifted right by 2 bits to obtain the numeric representation of the variable b. Furthermore, the value of variable a is obtained by subtracting a bit-shifted version of the value of variable b, left-shifted by 2 bits, from the value of variable m.

[000371] Subsequently, an arithmetic decoding of the values of the least significant bit plane r is repeated, where the number of repetitions is determined by the value of the variable “lev”. A value of the least significant bit plane r is obtained using the function “arith_decode”, in which the cumulative frequency table adapted to the decoding of the least significant bit plane is used (cumulative frequency table “arith_cf_r”). A least significant bit (having a numerical weight of 1) of the variable r describes a least significant bit plane of the spectral value represented by the variable a, and a bit having a numerical weight of 2 of the variable r describes a least significant bit of the spectral value represented by variable b. Thus, the variable a is updated by shifting the variable a to the left by 1 bit and adding the bit having the numerical weight of 1 of the variable r as the least significant bit. In a similar way, the variable b is updated by shifting the variable b to the left by one bit and adding the bit having the numerical weight of 2 to the variable r.

[000372] Thus, the two bits that carry the most significant information of variables a, b are determined by the value of the most significant bit plane m, and one or more less significant bits (if any) of the values a and b are determined by one or more values of the least significant bit plane r.

[000373] To summarize the above, if the symbol “ARITH_STOP” is not found, the remaining bit planes are then decoded, if any, to the double tuple present. The remaining bit planes are decoded from most significant to least significant level by calling the “arith_decode()” lev function multiple times with the cumulative frequency table “arith_cf_r[]”. The decoded bit planes r allow the refinement of the previously decoded value m according to the algorithm, a pseudo-program code that is shown in figure 5j.

[000374] 11.9.2 Decoding the least significant bit range according to figure 5n

[000375] Alternatively, however, the algorithm of a pseudo-program code representation that is shown in Figure 5n can also be used for decoding the least significant bit plane. In this case, if the “ARITH_STOP” symbol is not found, the remaining bit planes are then decoded, if any, to the present double tuple. The remaining bit planes are decoded from most significant to least significant level by calling “lev” times “arith_decode()” with the cumulative frequency table “arith_cf_r()”. The decoded bit planes r allow the refinement of the previously decoded value m according to the algorithm shown in figure 5n.

[000376] 11.10 Context update

[000377] 11.10.1 Context update according to figure 5k, 5l, and 5m

[000378] Next, the operations used to complete the decoding of the tuple of spectral values will be described, with reference to figures 5k and 5l. Furthermore, an operation will be described that is used to complete a decoding of a set of tuples of spectral values associated with a current part (e.g., a current structure) of an audio content.

[000379] It should be noted that algorithms according to figures 5k, 5l and 5m are preferred, even in algorithms can be used.

[000380] Now with reference to figure 5k, it can be seen that the entry having the entry index 2*i of the array “x_ac_dec[]” is defined to be equal to a, and that the entry having the entry index “2* i+1” of the array “x_ac_dec[]” is set to be equal to b after least significant bit 312d decoding. In other words, at the point after least significant bit decoding 312d, the unsigned value of the double tuple (a,b), is completely decoded. It is saved in the element (e.g. the “x_ac_dec[]” array) that holds the spectral coefficients according to the algorithm shown in figure 5k.

[000381] Subsequently, the context “q” is also updated to the next double tuple. It should be noted that this context update must also be performed for the last double tuple. This context update is performed by the “arith_update_context()” function, a representation of the pseudo-program code of which is shown in figure 5l.

[000382] Now with reference to figure 5l, it can be seen that the function “arith_update_context(i,a,b)” receives, as input variables, unsigned decoded quantized spectral coefficients (or spectral values) a, b of the double tuple . Furthermore, the “arith_update_context” function also receives, as an input variable, an index i (e.g. a frequency index) of the quantized spectral coefficient to decode. In other words, the input variable i can, for example, be an index of the tuple of spectral values, absolute values that are defined by the input variables a, b. As can be seen, the entry “q[1][i]” of the matrix “q[][]” can be set to a value that is equal to a+b+1. Furthermore, the value of the entry “q[1][i]” of the matrix “q[][]” can be limited to the hexadecimal value of “0xF”. Thus, the entry “q[1][i]” of the matrix “q[][]” is obtained by calculating a sum of absolute values of the currently decoded tuple {a,b} of spectral values having the frequency index i, and adding 1 to the result of said sum.

[000383] It should be noted here that the entry “q[1][i]” of the matrix “q[][]” can be considered as a value of the context subregion, as it describes a subregion of the context that is used for subsequent decoding of additional spectral values (or tuples of spectral values).

[000384] It should be noted here that the sum of the absolute values a and b of the two currently decoded spectral values (signed versions that are stored in the “x_ac_dec[2*i]” and “x_ac_dec[2*i+1]” entries of the matrix “ x_ac_dec[]”), can be considered as the calculation of a norm (e.g. an L1 norm) of the decoded spectral values.

[000385] It has been observed that the values of the context subregion (i.e., entries of the matrix “q[][]”), which describe a norm of a vector formed by several previously decoded spectral values, are particularly meaningful and with memory efficient. It was observed that such a norm, which is calculated based on several previously decoded spectral values, comprises significant context information in a compact form. It has been observed that the sign of the spectral values is typically not particularly relevant to the choice of context. It was also observed that the formation of a norm across the various previously decoded spectral values typically maintains the most important information, although some details are discarded. Furthermore, it has been observed that limiting the current numeric context value to the maximum value typically does not result in a serious loss of information. Preferably, it has been observed that it is more efficient to use the same context state for significant spectral values that are greater than a predetermined threshold value. Therefore, limiting the values of the context subregion brings with it further improvement of memory efficiency. Furthermore, it was observed that limiting the values of the context subregion to a certain maximum value allows a computationally efficient and particularly simple update of the current numerical context value, which was described, for example, with reference to Figures 5c and 5d. By limiting the values of the context subregion to a comparably small value (for example, to a value of 15), a context state that is based on several values of the context subregion can be represented in the efficient way that was discussed. taking figures 5c and 5d as reference.

[000386] Furthermore, it was observed that a limitation of the values of the context subregion of values between 1 and 15, brings with it a particularly good compromise between precision and memory efficiency, as 4 bits are sufficient in order to store such a value of the context subregion.

[000387] However, it should be noted that in other embodiments, a context subregion value may be based on a single spectral decoded value only. In this case, the formation of a standard can optionally be omitted.

[000388] The next double tuple of the structure is decoded after completing the “arith_update_context” function by increasing i by 1 and repeating the same process as described above, starting with the “arith_get_context()” function.

[000389] When the 2 lg/2 tuples are decoded within the structure, or with the stop symbol according to “ARITH_STOP” occurs, the spectral amplitude decoding process ends and the decoding of the signals begins.

[000390] Details regarding the decoding of the signals were discussed with reference to figure 3, in which the decoding of the signals is shown at reference number 314.

[000391] Since all unsigned quantized spectral coefficients are decoded, the agreement signal is added. For each non-zero quantized value of “x_ac_dec”, one bit is read. If the read bit value equals 0, the quantized value is positive, nothing is done, and the signed value is equal to the previously decoded unsigned value. Otherwise (i.e., if the read bit value is equal to 1), the decoded coefficient (or spectral value) is negative and the two's complement is considered to be the unsigned value. Signal bits are read from low to high frequencies. For details, reference is made to Figures 3 and the explanations regarding signal decoding 314.

[000392] Decoding is completed by calling the “arith_finish()” function. The remaining spectral coefficients are set to 0. The respective context states are updated correspondingly.

[000393] For details, reference is made to figure 5m, which shows a representation of the pseudo-program code of the “arith_finish()” function. As can be seen, the “arith_finish()” function receives an input variable lg that describes the decoded quantized spectral coefficients. Preferably, the input variable lg of the “arith_finish” function describes a number of currently decoded spectral coefficients, leaving the spectral coefficients disregarded, where a value of 0 has been allocated in response to the detection of an “ARITH_STOP” symbol. An input variable N of the “arith_finish” function describes a window length of a current window (i.e., a window associated with the current part of the audio content). Typically, a number of spectral values associated with a window of length N is equal to N/2 and a 2-tuple number of spectral values associated with a window of window length N is equal to N/4.

[000394] The “arith_finish” function also receives, as an input value, a vector “x_ac_dec” of decoded spectral values, or at least a reference to such a vector of decoded spectral coefficients.

[000395] The “arith_finish” function is configured to set the entries of the matrix (or vector) “x_ac_dec”, for which no spectral value has been decoded due to the presence of an arithmetic stop condition, to 0. Furthermore, the function “arith_finish” sets values from the context subregion “q[1][i]”, which are associated with a spectral value for which no value has been decoded due to the presence of an arithmetic stop condition, to the predetermined value of 1. The default value of 1 corresponds to the tuple of spectral values where both spectral values are equal to 0.

[000396] Thus, the “arith_finish()” function allows updating the entire matrix (or vector) “x_ac_dec[]” of spectral values and also the entire matrix of values of the context subregion “q[1][i] ”, even in the presence of an arithmetic stop condition.

[000397] 11.10.2 Context update according to figures 5o and 5p

[000398] Next, another realization of the context update will be described with reference to figures 5o and 5p. At the point where the unsigned value of the double tuple (a,b) is completely decoded, the context q is then updated to the next double tuple. The update is also performed if the present double tuple is the last double tuple. Both updates are made by the “arith_update_context()” function, the representation of the pseudo-program code shown in figure 5o.

[000399] The next double tuple of the structure is then decoded by increasing i by 1 and calling the arith_decode() function. If the 2 lg/2 tuples have already been decoded with the structure, or if the stop symbol “ARITH_STOP” has occurred, the function “arith_finish()” is called. The context is saved and stored in the “qs” matrix (or vector) for the next structure. A pseudo-program code for the “arith_save_context()” function is shown in figure 5p.

[000400] Since all unsigned quantized spectral coefficients are decoded, the signal is then additional. For each non-quantized value of “qdec”, one bit is read. If the read bit value equals 0, the quantized value is positive, nothing is done, and the signed value is equal to the previously decoded unsigned value. Otherwise, the decoded coefficient is negative and the two's complement is taken from the unsigned value. Signaled bits are read from low to high frequencies.

[000401] 11.11 Summary of the decoding process

[000402] Next, the decoding process will be briefly summarized. For details, reference is made to the above discussion and also to Figures 3, 4, 5a, 5c, 5e, 5g, 5j, 5k, 5l, and 5m. The quantized spectral coefficients “x_ac_dec[]” are silently decoded starting from the lowest frequency coefficient and continuing to the highest frequency coefficient. They are decoded by groups of two successive coefficients a,b assembling into a so-called double tuple (a,b) (also designated with{a,b}).

[000403] The decoded coefficients “x_ac_dec[]” for the frequency domain (i.e., for a frequency domain mode) are then stored in the matrix “x_ac_quant[g][win][sfb][bin]”. One transmission order of silently encoded passwords is such that when they are decoded in the order received and stored in the array, “bin” is the fastest-increasing index and “g” is the slowest-increasing index. Within a password, a decryption order is a, then b. The decoded coefficients “x_ac_dec[]” for “TCX” (i.e. for an audio decoding using a transform-coded excitation) are stored (e.g. directly) in the matrix “x_tcx_invquant[win][bin]” and a Transmission order of silently encoded passwords is such that when they are decoded in the order received and stored in the array, “bin” is the fastest-increasing index and “win” is the slowest-increasing index. Within a password, the decryption order is a, then b.

[000404] First, the “arith_reset_flag” indicator determines whether the context should be reset. If the indicator is true, this is considered in the “arith_map_context” function.

[000405] The decoding process begins with an initialization phase where the context element vector “q” is updated by copying and mapping the context elements of the previous structure stored in “q[1][]” into “q[0 ][]”. The context elements within “q” are stored in a bit with 4 by 2 tuples. For details, reference is made to the pseudo-program code in Figure 5a.

[000406] The silent decoder outputs 2 tuples of unsigned quantized spectral coefficients. Firstly, the state c of the context is calculated based on the previously decoded spectral coefficients around the double tuple to decode. In this way, the state is additionally updated using the context state of the last double tuple decoded considering only two new double tuples. The state is decoded into 17 bits and is returned by the “arith_get_context” function. A representation of the pseudo-program code of the “arith_get_context” function is shown in figure 5c.

[000407] The context state c determines the cumulative frequency table used to decode the most significant bitwise plane m. The mapping of c to the corresponding cumulative frequency table index “pki” is performed by a function “arith_get_pk()”. A representation of the pseudo-program code of the “arith_get_pk()” function is shown in figure 5e.

[000408] The value m is decoded using the function “arith_decode()” called with the cumulative frequency table, “arith_cf_m[pki][]”, where “pki” corresponds to the index returned by “arith_get_pk()”. The arithmetic encoder (and decoder) is an integer implementation using a scaled identification generation method. The pseudo-program code according to figure 5g describes the algorithm used.

[000409] When the decoded value m is the escape symbol “ARITH_ESCAPE”, the variables “lev” and “esc_nb” are increased by 1 and another value m is decoded. In this case, the “get_pk()” function is called once again with the value “c+ esc_nb<<17” as input argument, where “esc_nb” is the number of previously decoded escape symbols for the same double delimited tuple to 7.

[000410] Since the value m is not the “ARITH_ESCAPE” escape symbol, the decoder checks whether the successive m forms an “ARITH_STOP” symbol. If the condition “(esc_nb>0&&m==0)” is true, the symbol “ARITH_STOP” is detected and the decoding process ends. The decoder goes directly to the signal decoding described later. The condition means that the rest of the structure is made up of 0 values.

[000411] If the symbol “ARITH_STOP” is not found, the remaining bit planes are then decoded, if any, to the present double tuple. The remaining bit planes are decoded from most significant to least significant level by calling “arith_decode()” lev multiple times with the cumulative frequency table “arith_cf_r[]”. The decoded bit planes r allow the refinement of the previously decoded value m, according to the algorithm of a pseudo-program code shown in figure 5j. At this point, the unsigned value of the double tuple (a,b) is completely decoded. A representation of the pseudo-program code is saved in the element that maintains the spectral coefficients according to the algorithm, which is shown in figure 5k.

[000412] The context “q” is also updated to the next double tuple. It should be noted that this context update must also be performed for the last double tuple. This context update is performed by the “arith_update_context()” function, a representation of the pseudo-program code that is shown in figure 5l.

[000413] The next double tuple of the structure is then decoded by increasing i by 1 and repeating the same process as described above, starting with the “arith_get_context()” function. When the 2 lg/2 tuples are decoded within the structure, or when the stop symbol “ARITH_STOP” occurs, the spectral amplitude decoding process ends and decoding of the signals begins.

[000414] Decoding is completed by calling a function “arith_finish()”. The remaining spectral coefficients are set to 0. The respective context states are updated correspondingly. A pseudo-program code representation of the “arith_finish” function is shown in figure 5m.

[000415] Since all unsigned quantized spectral coefficients are decoded, the agreement signal is added. For each non-zero quantized value of “x_ac_dec”, one bit is read. If the read bit value is equal to 0, the quantized value is positive, and nothing is done, and the signed value is equal to the previously decoded unsigned value. Otherwise, the decoded coefficient is negative and the two's complement is taken from the unsigned value. Signaled bits are read from low to high frequencies.

[000416] 11.12 Subtitles

[000417] Figure 5q shows a legend of the definitions that is related to the algorithms according to figures 5a, 5c, 5e, 5f, 5g, 5j, 5k, 5l, and 5m.

[000418] Figure 5r shows a legend of the definitions that is related to the algorithms according to figures 5b, 5d, 5f, 5h, 5i, 5n, 5o, and 5p.

[000419] Mapping tables

[000420] In one embodiment, according to the invention, the particularly advantageous tables “ari_lookup_m”, “ari_hash_m”, and “ari_cf_m” are used for executing the function “arith_get_pk()” according to figure 5e or figure 5f, and for executing the function “arith_decode()” as discussed with reference to figures 5g, 5h and 5i. However, it should be noted that different tables may be used in some embodiments according to the invention.

[000421] 12.1 “ari hash m[742]” table according to figures 22(1), 22(2), 22(3) and 22(4)

[000422] A content of a particularly advantageous implementation of the table “ari_hash_m”, which is used by the function “arith_get_pk”, a first preferred embodiment which has been described with reference to figure 5e, and a second embodiment which has been described with reference to figure 5f , is shown in the table in figures 22(1) to 22(4). It should be noted that the table in figures 22(1) to 22(4) lists the 742 entries of the table (or matrix) “ari_hash_m[742]”. It should also be noted that the table representation of figures 22(1) to 22(4) shows the elements in the order of the element indices, so that the first value “0x00000104UL” corresponds to a table entry “ari_hash_m[0] ” having an element index (or table index) 0, and such that the last value “0xFFFFFF00UL” corresponds to a table entry “ari_hash_m[741]” having element index or table index 741. It must still be noted that “0x” indicates that the table entries of the “ari_hash_m[]” table are represented in a hexadecimal format. Furthermore, it should be noted here that the “UL” suffix indicates that the table entries of the “ari_hash_m[]” table are represented as unsigned “long” integer values (having a precision of 32-bits).

[000423] Furthermore, it should be noted that the table entries of the “ari_hash_m[]” table according to figures 22(1) to 22(4) are arranged in a numerical order in order to allow the search to be performed from table 506b, 508b, 510b of the “arith_get_pk()” function.

[000424] It should further be noted that the 24 most significant bits of the table entries of the “ari_hash_m” table represent certain significant state values (and can be considered as a first sub-entry), while the 8 least significant bits represent values of mapping rule index “pki” (and can be considered as a second sub-entry). Thus, the “ari_hash_m[]” table entries describe a “direct impact” mapping of a context value to a “pki” mapping rule index value.

[000425] However, the highest 24 bits of the table entries “ari_hash_m[]” represent, at the same time, limits of ranges of numeric context values, in which the same mapping rule index value is associated. Details regarding this concept have already been discussed above.

[000426] 12.2 “ari lookup m” table according to figure 21

[000427] A content of a particularly advantageous embodiment of the “ari_lookup_m” table is shown in the table of figure 21. It should be noted here that the table of figure 21 lists the entries of the “ari_lookup_m” table. Entries are referenced by a one-dimensional integer entry index (also designated as “element index” or “array index” or “table index”) which is, for example, designated as “i_max” or “i_min ” or “i”. It should be noted that the “ari_lookup_m” table, which comprises a total of 742 entries, is well suited for use by the “arith_get_pk” function according to figure 5e or figure 5f. It should also be noted that the “ari_lookup_m” table according to figure 21 is adapted to cooperate with the “ari_hash_m” table according to figure 22.

[000428] It should be noted that the entries of the table “ari_lookup_m[742]” are listed in ascending order of the index of table “i” (e.g., “i_min” or “i_max”) between 0 and 741. The term “0x ” indicates that table entries are described in a hexadecimal format. Thus, the first table entry “0x01” corresponds to the table entry “ari_lookup_m[0]” having the table index 0 and the last table entry “0x27” corresponds to the table entry “ari_lookup_m[741]” having the index from table 741.

[000429] It should also be noted that entries in the “ari_lookup_m[]” table are associated with the ranges defined by adjacent entries in the “arith_hash_m[]” table. Thus, the “ari_lookup_m” table entries describe mapping rule index values associated with numeric context value ranges, where the ranges are defined by the “arith_hash_m” table entries.

[000430] 12.3. Table “ari cf m[64][17]” according to figures 23(1), 23(2) and 23(3)

[000431] Figure 23 shows a set of 64 cumulative frequency tables (or subtables) "ari_cf_m[pki][17]", one which is selected by the audio encoder 100, 700 or an audio decoder 200, 800, for example, for executing the “arith_decode()” function, that is, for decoding the value of the most significant bit plane. The one selected from the 64 cumulative frequency tables (or subtables) shown in figures 23(1) to 23(3) considers a function from the “cum_freq[]” table when executing the “arith_decode()” function.

[000432] As can be seen from figures 23(1) to 23(3), each sub-block or line represents the cumulative frequency table having 17 entries. For example, a first subblock 2310 represents the 17 cumulative frequency table entries for “pki=0”. A second subblock or line 2312 represents the 17 cumulative frequency table entries for “pki=1”. Finally, a 64th subblock or line 2364 represents the 17 cumulative frequency table entries for “pki=63”. Thus, figures 23(1) to 23(3) effectively represent the 64 different cumulative frequency tables (or subtables) from “pki=0” to “pki=95”, where each of the 64 cumulative frequency tables is represented by a sub-block (surrounded by coiled supports) or line, and wherein each of said cumulative frequency tables comprises 17 entries.

[000433] Within a sub-block or line (e.g., a sub-block or line 2310 or 2312, or a sub-block or line 2396), a first value (e.g., a first value 708 of the first sub- block 2310) describes a first cumulative frequency table entry (having a matrix index or table index of 0) represented by the subblock or row, and a last value (e.g., a last value 0 of the first subblock or line 2310) describes a last cumulative frequency table entry (having a matrix index or table index of 16) represented by the sub-block or line.

[000434] Thus, each sub-block 2310, 2312, 2364 of the table representation of figure 23 represents the cumulative frequency table entries for use by the “arith_decode” function according to figure 5g, or according to figures 5h and 5i. The input variable “cum_freq[]” of the “arith_decode” function describes which of the 64 cumulative frequency tables (represented by the individual sub-blocks of 17 entries of the “arith_cf_m” table) should be used for decoding the current spectral coefficients.

[000435] 12.4 Table “ari cf r[]” according to figure 24

[000436] Figure 24 shows the contents of the “ari_cf_r[]” table.

[000437] The four entries of said table are shown in figure 24. However, it should be noted that the “ari_cf_r” table may eventually be different in other embodiments.

[000438] Overview, performance evaluation and advantages

[000439] Embodiments according to the invention utilize the updated functions (or algorithms) and a set of updated tables, as discussed above, to obtain an improved trade-off between computational complexity, memory requirement and coding efficiency.

[000440] Generally speaking, the embodiments, according to the invention, create an improved silent spectral coding. Embodiments according to the present invention describe an improvement of silent spectral coding in USAC (unified audio and speech coding).

[000441] The embodiments, according to the invention, create an updated proposal for CE in improved silent spectral coding of spectral coefficients, based on the schemes as presented in the MPEG input papers m16912 and m17002. Both proposals were evaluated, potential outcomes eliminated, and forces combined. Furthermore, embodiments of the invention comprise an update of the silent spectral coding tables for application to a current USAC specification.

[000442] 13.1. Overview

[000443] Next, a brief overview will be given. In the course of ongoing standardization of USAC (Unified Speech and Audio Coding), a silent spectral coding scheme (aka entropy coding scheme) in USAC has been proposed. This spectral silent coding scheme helps to more efficiently encode quantized spectral coefficients in a lossless manner. In this way, the spectral coefficients are mapped to corresponding passwords of variable length. This entropy coding scheme is based on a context-based arithmetic coding scheme: The context (i.e., the nearby spectral coefficients) of a spectral coefficient determines a probability distribution (cumulative frequency table), which is used to the arithmetic coding of the spectral coefficient.

[000444] Embodiments according to the present invention use an updated set of tables for the spectral coding scheme, as previously proposed in the USAC context. As a basis, it should be noted that conventional silent spectral coding technology consists first of an algorithm and second of a set of trained tables (or, at least, comprises an algorithm and a set of trained tables). This set of conventional trained tables is based on WD4 USAC bitstreams. Since USAC has now advanced to WD7, and significant changes have been applied to the USAC specification, however, a new set of retrained tables is used in the embodiments according to the invention, which is based on the latest version of USA WD7. The algorithm itself remains unchanged. As a side effect, the retrained tables provide better compression performance than any of the previously presented schemes.

[000445] According to the present invention, it is proposed to replace conventional trained tables with retrained tables as presented here, which results in increased coding performance.

[000446] 13.2. Introduction

[000447] Next, an introduction will be provided.

[000448] For the USAC work item, several proposals on updating the silent coding scheme were brought up during the last few meetings collaboratively. However, this work was basically started at the 89th meeting. Thus it has been common practice for all proposals in spectral coefficient encoding to show performance results based on USAC's WD4 reference quality bitstreams and training on a WD4 training database.

[000449] However, major improvements in other fields of USAC, in particular stereo processing and windowing, have been incorporated into the USAC specification so far. It was observed that these improvements also slightly affect the statistics for silent spectral coding. The results shown for silent coding CEs can thus be referred to as suboptimal, as long as they do not correspond to the latest revision of WD.

[000450] Certainly, silent spectral coding tables are suggested as better adapted to the updated algorithms and statistics of the spectral values to be encoded and decoded.

[000451] 13.3. Brief description of the algorithm

[000452] Next, a brief description of the algorithm will be provided.

[000453] To overcome the issue of memory consumption and computational complexity, an improved silent coding scheme is proposed to replace the scheme as in working design 6/7 (WD6/7). The main focus in development was placed on reducing memory demand while maintaining compression efficiency and not increasing computational complexity. More specifically the goal was to achieve the best tradeoff in the multidimensional complexity space of compression performance, complexity and memory requirements.

[000454] The proposed new coding scheme borrows the main feature of the WD6/7 silent encoder, namely context adaptation. Context is derived using previously decoded spectral coefficients, as close as in WD6/7 to past and present structure. However, spectral coefficients are now encoded by combining two coefficients together to form a double tuple. Another difference remains in the fact that the spectral coefficients are now divided into three parts, the signal, the most significant bits (MSBs) and the least significant bits (LSBs). The signal is encoded independently of the magnitude which is further divided into two parts, the two most significant bits and the rest of the bits, if any. The 2-tuples in which the magnitude of the two elements is less than or equal to 3 are directly encoded by encoding MSBs. Otherwise, an escape password is passed first to signal any additional bit planes. In the base version, the missing information, the LSBs and the signal, are both encoded using the uniform probability distribution. Alternatively, a different probability distribution can be used.

[000455] Reducing the size of the table is still possible, since:

[000456] only probabilities for 17 symbols need to be stored: symbol {[0;+3], [0;+3]}+ESC;

[000457] there is no need to store a grouping table (e groups, d groups, dg vectors);

[000458] the size of the hash table can be reduced with training carried out.

[000459] 13.3.1 Coding of MSBs

[000460] Next, an encoding of MSBs will be described.

[000461] As already mentioned, the main difference between WD6/7, previous proposals and the current proposal, is the size of the symbols. In WD6/7 4-tuples were considered for context generation and silent encoding. In previous submissions, 1-tuple was used instead to reduce ROM requirements. During our development, 2-tuples were considered the best compromise for reducing ROM requirements without increasing computational complexity. Instead of considering four 4-tuples for context derivation, now four double-tuples are considered. As shown in figure 25, three double tuples come from the previous structure and one from the present structure.

[000462] The reduction in table size is made due to three main factors. First, only probabilities for 17 symbols need to be stored (i.e. symbol {[0;+3], [0;+3]} + ESC). The clustering tables (i.e., e-groups, d-groups, and dg-vectors) are no longer needed. Finally, the size of the hash table was reduced by performing appropriate training.

[000463] Although the dimension was reduced from four to two, the complexity was maintained in the WD6/7 range of the USAC Design Standard. It was achieved by simplifying both context generation and hash table access.

[000464] The different simplifications and optimizations were made in such a way that the coding performance was not affected, and more slightly improved.

[000465] 13.3.2 LSB coding

[000466] LSBs are encoded with a uniform probability distribution. Compared to WD6/7, LSBs are now considered to be within double-tuples instead of 4-t-tuples. However, different encoding of the least significant bits is possible.

[000467] 13.3.3 Signal coding

[000468] The signal is encoded without using the arithmetic core encoder to reduce complexity. The signal is transmitted in 1 bit only when the corresponding magnitude is non-zero. 0 means a positive value and 1 means a negative value.

[000469] 13.4. Update of proposed tables

[000470] This contribution provides an updated set of tables for the USAC silent coding spectral scheme scheme. The tables have been retrained based on current USA WD6/7 bitstreams. Far from the current tables, which result from a training process, the algorithm remains unchanged.

[000471] To investigate the effect of retraining, the code efficiency and memory requirement of the new tables are compared with the previous proposal (M17558) and WD6. WD6 is selected as a reference point since a) the results at the 92nd meeting were given with respect to this reference and b) the differences between WD6 and WD7 are only very small (security bugs only, with no effect on coding entropy or distribution of spectral coefficients).

[000472] 13.4.1 Coding efficiency

[000473] Firstly, the coding efficiency of the proposed new set of tables is compared with USAC WD6 and CE as proposed in M17558. As can be seen in the table representation of figure 26, by a simple retraining the average increase in coding efficiency (compared to WD6) could be increased from 1.74% (M17558) to 2.45% (new proposal, according to with an embodiment of the invention). Compared to M17558, the compression gain could then be increased by approximately 0.7% in the embodiments according to the invention.

[000474] Figure 27 visualizes the compression gain for all operating points. As can be seen, a minimum compression gain of at least 2% can be achieved using the embodiments according to the invention compared to WD6. For low rates, such as 12 kbit/s and 16 kbit/s, the compression gain is even slightly increased. Good performance is also retained at higher bitrates such as 64 kbit/s, where a significant increase in coding efficiency of more than 3% can be observed.

[000475] It should be noted that lossless transcoding of all WD6 reference quality bit streams has proven possible without violating bit reservoir constraints. More detailed results will be given in section 13.6.

[000476] 13.4.2 Memory demand and complexity

[000477] Then, the memory demand and complexity are compared with USAC WD6 and CE as proposed in M17558. The table in figure 28 compares the memory demand for the silent encoder as in WD6, proposed in M17558 and the New Proposal according to an embodiment of the invention. As can be clearly seen, the memory demand is significantly reduced by adopting the new algorithm, as proposed in M17558. It can still be seen that for the New Proposal the total table size could still be slightly reduced by approximately 80 words (32 bits), resulting in a total ROM demand of 1441 words, and a total RAM demand of 64 words (32 bit) per audio channel. The small savings in ROM demand are the result of a better trade-off between the number of probability models and the size of the hash table, found by the automatic training algorithm based on the new training bitstreams from the WD6 set. For more details, reference is made to the table in figure 29.

[000478] In terms of complexity, the computational complexity of the most recent proposed schemes was compared with an optimized version of the current silence in USAC. It was observed by a “pen and paper” method and code instruction that the new coding scheme has the same order of complexity as the current scheme. As reported in the table in Figure 30 for 32 kbps stereo and the table in Figure 31 for the 12 kbps mono operating points, the estimated complexity shows an increase of 0.006 in MOPS weight and 0.024 in MOPS weight respectively over an optimized implementation of WD6 silent decoder. Compared to the overall complexity of approximately 11.7 PCU [2], these differences can be considered insignificant.

[000479] 13.5. Conclusion

[000480] Next, some conclusions will be provided.

[000481] A new set of tables for the USAC silent coding spectral scheme has been presented. In contrast to the previous proposal, which is the result of training based on older bitstreams, the proposed new tables are now trained on current WD USAC bitstreams, where an advanced training concept was used. By this retraining, coding efficiency in current USAC bitstreams could be improved, without sacrificing low memory demand or increasing complexity compared to previous proposals. Compared to the USAC WD6, the memory demand could be significantly reduced.

[000482] 13.6. Detailed information about transcoding WD6 bitstreams

[000483] Detailed information about the transcoding of Work Project 6 (WD6) bit streams can be seen in the table representation of figures 32, 33, 34, 35 and 36.

[000484] Figure 32 shows a tabular representation of the average bit rates produced by the arithmetic encoder in an embodiment according to the invention and in WD6.

[000485] Figure 33 shows a table representation of the minimum, maximum and average USAC bitrates on a structure basis using the proposed scheme.

[000486] Figure 34 shows a tabular representation of the average bit rates produced by a USAC encoder using WD6 arithmetic encoder and an encoder according to an embodiment according to the invention ("New Proposal").

[000487] Figure 35 shows a tabled representation of the best and worst cases for an embodiment according to the invention.

[000488] Figure 36 shows a table representation of the bit reservoir limit for an embodiment according to the invention.

[000489] 14. Changes when compared to work project 6 or work project 7

[000490] Next, changes in silent coding when compared to conventional silent coding will be described. Of course, an accomplishment is defined in terms of modifications when compared to work design 6 or work design 7 of the USAC Design Standard.

[000491] In particular, changes to the WD text will be described. In other words, this section lists the complete set of changes against the USAC WD7 specification.

[000492] 14.1. Changes to the technical description

[000493] The proposed new silent encoding generates the modifications in MPEG USAC WD that will be described below. The main differences are marked.

[000494] 14.1.1. Syntax and payload changes

[000495] Figure 7 shows a representation of an arithmetically encoded data syntax “arith_data()”. The main differences are marked.

[000496] Next, changes regarding Silent Spectral Encoder Payloads will be described.

[000497] Spectral coefficients of the “linear prediction domain” coded signal and the “frequency domain” coded signal are scalar quantized and then silently coded by an arithmetic coding in a context-dependent adaptive manner. The quantized coefficients are joined into 2-tuples before being transmitted from lower frequency to higher frequency. It should be noted that the use of 2-tuples constitutes a change when compared to previous versions of silent spectral coding.

[000498] However, it is still a change that each 2-tuple is divided into sign s, the most significant bit plane, m, and the remaining least significant bit planes, r. Further, it is a change that the value m is encoded according to coefficient proximity, and that the remaining least significant bit planes, r, are entropy encoded without considering context. Furthermore, it is a change from some previous versions that the values m and r form the symbols of the arithmetic encoder. Finally, it is a change from some previous versions that the s signals are encoded outside the arithmetic encoder using 1 bit per non-zero quantized coefficient.

[000499] A detailed arithmetic decoding procedure is described below in section 14.2.3.

[000500] 14.1.2 Changes to definitions and help elements

[000501] Changes to definitions and help elements are shown in the representation of definitions and help elements in figure 38.

[000502] 14.2 Silent spectral coding

[000503] In the following, silent spectral coding according to an embodiment will be summarized.

[000504] 14.2.1 Tool description

[000505] Silent spectral coding is used to further reduce the redundancy of the quantized spectrum.

[000506] The silent spectral coding scheme is based on an arithmetic coding in conjunction with a dynamically adapted context. Silent coding is input by quantized spectral values and uses the context-dependent cumulative frequency tables derived from the previously decoded next four. Here, proximity in time and frequency is considered, as illustrated in figure 25. The cumulative frequency tables are then used by the arithmetic coder to generate a binary code of variable length.

[000507] The arithmetic coder produces a binary code for a given set of symbols and their respective probabilities. Binary code is generated by mapping a probability interval, where the set of symbols lies, into a code.

[000508] 14.2.2 Definitions

[000509] Definitions and help elements are described in figure 39. Changes when compared to previous versions of arithmetic coding are marked.

[000510] 14.2.3 Coding process

[000511] Quantized spectral coefficients qdec are silently encoded starting from the lowest frequency coefficient and continuing to the highest frequency coefficient. They are decoded by groups of two successive coefficients a and b joining together into a so-called 2-tuple {a,b}.

[000512] The decoded coefficients for AAC are then stored in the matrix x_ac_quant[g][win][sfb][bin]. The transmission order of silent encoding codes is such that when they are decoded in the order received and stored in the array, bin is the fastest additional index and g is the slowest additional index. Within a code the order of decoding is a and then b.

[000513] The decoded coefficients for TCX are stored in the matrix x_tcx_invquant[win][bin], and the order of transmission of the silent encoding codes is such that when they are decoded in the order received and stored in the matrix, the bin is the index additional fastest and win is the slowest additional index. Within a code the order of decoding is a and then b.

[000514] The encoding process begins with an initialization phase where a mapping is made between the previously saved context stored in qs and the context of the current structure q. The previous context qs is stored in 2 bits per frequency line.

[000515] For details, reference is made to the pseudo-program code representation of the “arith_map_context” algorithm in figure 40a.

[000516] The silent decoder outputs 2 tuples of quantized spectral unsigned coefficients. Firstly, the state c of the context is calculated based on the previously decoded spectral coefficients surrounding the 2 tuples to decode. The state is gradually updated using the context state of the last double tuples decoded considering only the two new double tuples. The state is encoded in 17 bits and is returned by the arith_get_context() function.

[000517] A representation of the pseudo-program code of the “arith_get_context()” function is shown in figure 40b.

[000518] Since the context state c is calculated, the most significant bitwise plane m is decoded using the inserted arith_decode() with the appropriate cumulative frequency table corresponding to the probability model corresponding to the context state. The matching is done by the arith_get_pk() function.

[000519] A pseudo-program code representation of the arith_get_pk() function is shown in figure 40c.

[000520] The value m is decoded using the arith_decode() function called with the cumulative frequency table, arith_cf_m[pki][], where pki corresponds to the index returned by arith_get_pk(). The arithmetic encoder is an integer implementation using the scaled tag generation method. The C pseudo-code shown in figures 40d and 40e describes the algorithm used.

[000521] When the decoded value m is the escape symbol, ARITH_ESCAPE, the variables lev and esc_nb are added into one and another value m is encoded. In this case, the get_pk() function is called once again with the value c&esc_nb<<17 as the input argument, where esc_nb is the number of previously decoded escape symbols for the same 7-bound double tuples.

[000522] Since the value m is not the escape symbol, ARITH_ESCAPE, the decoder checks whether the successive m forms an ARITH_STOP symbol. If a condition (esc_nb>0 && m==0) is true, the ARITH_STOP symbol is detected and the encoding process ends. The decoder goes directly to the arith_save_context() function. A condition means that the rest of the structure is made up of zero values.

[000523] If the ARITH_STOP symbol is not reached, the remaining bit planes are then decoded if they exist for the present double tuple. The remaining bit planes are decoded from most significant to least significant level by calling lev periods arith_decode() with the cumulative frequency table arith_cf_r[]. The decoded bit planes r allow refining the previously decoded value m by the function or algorithm of a representation of the pseudo-program code in which is shown in figure 40f.

[000524] At this point, the unsigned value of the double tuples {a,b} is completely decoded. The context q is then updated for the next double tuples. If it's the last double tuples, too. Both updates are done by the arith_update_context() function, a representation of the pseudo-program code of which is shown in figure 40g.

[000525] The next double tuples of the structure are then decoded by increasing i by one and calling the function. If lg/2 double tuples have already been decoded with the structure or if the ARITH_STOP stop symbol has occurred, the arith_save_context() function is called. The context is saved and stored in qs for the next structure. A pseudo-program code representation of the arith_save_context() function or algorithm is shown in figure 40h.

[000526] Since all unsigned quantized spectral coefficients are decoded, the signal is then added. For each non-zero quantized value of qdec, one bit is read. If the read bit value is zero, the quantized value is positive, nothing is done, and the signed value is equal to the unsigned previously decoded value. Otherwise, the decoded coefficient is negative and the two component is of the unsigned value. Signal bits are read from low to high frequencies.

[000527] 14.2.4 Updated tables

[000528] A set of retrained tables for use with the algorithms described above is shown in figures 41(1), 41(2), 42(1), 42(2), 42(3), 42(4), 43 (1), 43(2), 43(3), 43(4), 43(5), 43(6) and 44.

[000529] Figures 41(1) and 41(2) show a table representation of a content of a table “ari_lookup_m[742]”, according to an embodiment of the invention;

[000530] Figures 42 (1),(2),(3),(4) show a table representation of a content of a table “ari_hash_m[742]”, according to an embodiment of the invention;

[000531] Figures 43 (1),(2),(3),(4),(5),(6) show a table representation of a content of a table “ari_cf_m[96][17]”, according to an embodiment of the invention; It is

[000532] Figure 44 shows a table representation of a table “ari_cf_r[4]”, according to an embodiment of the invention.

[000533] To summarize the above, it can be seen that embodiments, according to the present invention, provide a good trade-off between computational complexity, memory requirements and coding efficiency.

[000534] 15. Bitstream syntax

[000535] 15.1 Silent Spectral Encoder Payloads

[000536] Next, some details regarding the silent spectral encoder payloads will be described. In some embodiments, there are several different coding modes, such as the “linear prediction domain” coding mode and the “frequency domain” coding mode. In the linear prediction domain coding mode, a noise shape is realized based on a linear prediction analysis of the audio signal, and a noise shape signal is encoded in the frequency domain. In frequency domain coding mode, a noise form is performed based on a psychoacoustic analysis and a noise form version of the audio content is encoded in the frequency domain.

[000537] Spectral coefficients of the “linear prediction domain” encoded signal and the “frequency domain” encoded signal are quantized by scaling and then silently encoded by a context-dependent arithmetic encoding in an adaptive manner. The quantized coefficients are joined into 2-tuples before being transmitted from the lower frequency to the higher frequency. Each double tuple is divided into a sign s, the most significant bit plane m, and one or more remaining least significant bit planes r (if any). The value m is encoded according to a context defined by nearby spectral coefficients. In other words, m is coded according to nearby coefficients. The remaining least significant bit planes r are entropy encoded without considering context. By means of m and r, the amplitude of these spectral coefficients can be reconstructed on the decoder side. For all non-zero symbols, the s signals are encoded outside the arithmetic encoder using 1 bit. In other words, the values m and r form the symbols of the arithmetic encoder. Finally, the signals s, are encoded outside the arithmetic encoder using 1 bit per non-zero quantized coefficient.

[000538] A detailed arithmetic coding procedure will be described here.

[000539] 15.2 Syntax elements according to figures 6a to 6j

[000540] In the following, the bitstream syntax of a bitstream carrying arithmetically encoded spectral information will be described with reference to figures 6a to 6j.

[000541] Figure 6a shows a syntax representation of the USAC raw data block (“usac_raw_data_block()”).

[000542] The USAC raw data block comprises one or more single channel elements (“single_channel_element()”) and/or one or more channel pair elements (“channel_pair_element()”).

[000543] Now with reference to figure 6b, the syntax of a single channel element is described. Single channel elements comprise a linear prediction domain channel stream (“lpd_channel_stream()”) or a frequency domain channel stream (“fd_channel_stream()”) depending on the core mode.

[000544] Figure 6c shows a syntax representation of an element of the channel pair. An element of the channel pair comprises code mode information (“core_mode0”, “core_mode1”). Furthermore, the channel pair element may comprise an “ics_info()” configuration information. Additionally, depending on the code mode information, the channel pair element comprises a linear prediction domain channel stream or a frequency domain channel stream associated with a first of the channels, and the channel pair element also comprises a linear prediction domain channel stream or a frequency domain channel stream associated with a second of the channels.

[000545] The configuration information “ics_info()”, a syntax representation that is shown in figure 6d, comprises several different items of configuration information, which are not of particular relevance to the present invention.

[000546] A frequency domain channel stream (“fd_channel_stream()”), a syntax representation of which is shown in Figure 6e, comprises a gain information (“global_gain”) and a configuration information (“ics_info()) ”). Additionally, the frequency domain channel stream comprises scale factor data (“scale_factor_data()”), which describes scale factors used to scale spectral values of different scale factor ranges, and which is applied, for example, by frequency multiplier 150 and frequency remultiplier 240. The frequency domain channel stream also comprises arithmetically encoded spectral data (“ac_spectral_data()”), which represents arithmetically encoded spectral values.

[000547] The arithmetically encoded spectral data (“ac_spectral_data()”), a syntax representation of which is shown in Figure 6f, comprises an optional arithmetic reset flag (“arith_reset_flag”), which is used to selectively reset the context as per described above. Furthermore, the arithmetically decoded spectral data comprises several arithmetic data blocks (“arith_data”), which carry the arithmetically encoded spectral values. The structure of the arithmetically encoded data blocks depends on the number of frequency bands (represented by the variable “num_bands”) and also on the state of the arithmetic reset flag, as will be discussed below.

[000548] Next, the structure of the arithmetically coded data block will be described with reference to figure 6g, which shows a syntax representation of said arithmetically coded data blocks. The representation of data within the arithmetically encoded data block depends on the lg number of spectral values to be encoded, the status of the arithmetic reset flag and also the context, i.e. the previously encoded spectral values.

[000549] The context for encoding the current set (e.g., 2 tuples) of spectral values is determined according to the context determination algorithm shown in reference number 660. Details regarding the context determination algorithm have been explained above, taking figures 5a and 5b as reference. The arithmetically encoded data block comprises lg/2 for defining the passwords, each set of passwords representing several (e.g., a double tuple) spectral values. A set of passwords comprises an arithmetic password “acod_m[pki][m]” that represents a value from the most significant bit plane m of the tuple of spectral values using between 1 and 20 bits. Furthermore, the set of passwords comprises one or more passwords “acod_r[r]” if the tuple of spectral values requires more bit planes than the most significant bit plane for correct representation. The password “acod_r[r]” represents a least significant bit plane using between 1 and 14 bits.

[000550] If, however, one or more less significant bit planes are required (in addition to the most significant bit plane) for a correct representation of the spectral values, this is signaled using one or more arithmetic escape passwords (“ARITH_ESCAPE”) . Thus, it can generally be said that for a spectral value, it is determined how many bit planes (the most significant bit plane and possibly one or more additional less significant bit planes) are needed. If one or more least significant bit planes are required, this is signaled by one or more arithmetic escape passwords “acod_m[pki][ARITH_ESCAPE]”, which are encoded according to a currently selected cumulative frequency table, an index of the cumulative frequency table which is given by the variable “pki”. Furthermore, the context is adapted, as can be seen from reference numbers 664, 662, if one or more arithmetic escape passwords are included in the bitstream. Following one or more escape arithmetic passwords, an arithmetic password “acod_m[pki][m]” is included in the bitstream as shown in reference number 663, where “pki” designates the currently valid index of the probability model (considering the context adaptation caused by the inclusion of the arithmetic escape passwords under consideration) and where m designates the value of the most significant bit plane of the spectral value to be encoded or decoded (where m is different from the password “ARITH_ESCAPE”) .

[000551] As discussed above, the presence of any least significant bit plane results in the presence of one or more passwords “acod_r[r]”, each representing 1 bit of a least significant bit plane of a first spectral value and each also representing 1 bit of a least significant bit plane of a second spectral value. One or more passwords “acod_r[r]” are encoded according to a corresponding cumulative frequency table, which may, for example, be constant and context-independent. However, different mechanisms for selecting the cumulative frequency table for a decryption of one or more passwords “acod_r[r]” are possible.

[000552] Furthermore, it should be noted that the context is updated after encoding each tuple of spectral values, as shown in reference number 668, so that the context is typically different from encoding and decoding double tuples of spectral values subsequent.

[000553] Figure 6i shows a legend of definitions and help elements that define the syntax of the arithmetically encoded data block.

[000554] Furthermore, an alternative arithmetic data syntax “arith_data()” is shown in figure 6h, with a corresponding legend of definitions and help elements shown in figure 6j.

[000555] To summarize the above, a bitstream format has been described, which can be provided by audio encoder 100 and which can be evaluated by audio decoder 200. The bitstream of arithmetically encoded spectral values is encoded in way that fits the decoding algorithm discussed above.

[000556] Furthermore, it should generally be noted that encoding is the inverse operation of decoding, so that it can generally be assumed that the encoder performs a table lookup using the tables mentioned above, which is approximately inverse to the table lookup performed by the decoder. Generally, it can be said that a person skilled in the art who knows the decoding algorithm and/or the desired bitstream syntax can easily design an arithmetic encoder, which provides the data defined in the bitstream syntax and needed by an arithmetic decoder. .

[000557] Furthermore, it should be noted that the mechanisms for determining the current numeric context value and for deriving a mapping rule index value may be identical in an audio encoder and an audio decoder, as it is typically desired that the audio decoder uses the same context as the audio encoder, so that decoding is adapted to encoding.

[000558] 15.3. Syntax elements according to figures 6k, 6l, 6m, 6n, 6o and 6p

[000559] In the following, an extraction of an alternative bitstream syntax will be described with reference to figures 6k, 6l, 6m, 6n, 6o and 6p.

[000560] Figure 6k shows a syntax representation of an element of the bitstream “UsacSingleChannelElement(indepFlag)”. Said “UsacSingleChannelElement(indepFlag)” syntax element comprises a “UsacCoreCoderData” syntax element that describes a core encoder channel.

[000561] Figure 6l shows a syntax representation of an element of the bitstream “UsacChannelPairElement(indepFlag)”. Said “UsacChannelPairElement(indepFlag)” syntax element comprises a “UsacCoreCoderData” syntax element that describes one or two channels of the core encoder, depending on a stereo configuration.

[000562] Figure 6m shows a syntax representation of an element of the “ics_info()” bitstream, which comprises definitions of a number of parameters, as can be seen in Figure 6m.

[000563] Figure 6n shows a syntax representation of an element of the “UsacCoreCoderData()” bitstream. The bitstream element “UsacCoreCoderData()” comprises one or more linear prediction domain channel streams “lpd_channel_stream()” and/or one or more frequency domain channel streams “fd_channel_stream()”. Other control information can optionally also be included in the “UsacCoreCoderData()” bitstream element, as can be seen in figure 6n.

[000564] Figure 6o shows a syntax representation of an element of the bitstream “fd_channel_stream()”. The “fd_channel_stream()” bitstream element comprises, among other optional bitstream elements, a “scale_factor_data()” bitstream element and an “ac_spectral_data()” bitstream element.

[000565] Figure 6p shows a syntax representation of an element of the “ac_spectral_data()” bitstream. The bitstream element “ac_spectral_data()” optionally comprises a bitstream element “arith_reset_flag”. Furthermore, the bitstream element also comprises a number of arithmetically encoded data “arith_data()”. The arithmetically encoded data may, for example, follow the bitstream syntax described with reference to Figure 6g.

[000566] 16. Alternative implementations

[000567] Although some aspects have been described in the context of an apparatus, it is clear that these aspects also represent a description of the corresponding method, where a block or device corresponds to a method step or a characteristic of a method step. Equivalently, aspects described in the context of a method step also represent a description of a corresponding block or item or feature of a corresponding apparatus. Some or all of the steps of the method may be performed by (or using) a hardware device such as, for example, a microprocessor, a programmable computer or an electronic circuit. In some embodiments, one or more of the most important steps of the method may be performed by such an apparatus.

[000568] The inventive coded audio signal may be stored on a digital storage medium or may be transmitted on a transmission medium such as a wireless transmission medium or a wired transmission medium such as the Internet.

[000569] Depending on certain implementation requirements, embodiments of the invention can be implemented in hardware or in software. The implementation can be carried out using a digital storage medium, for example, a floppy disk, a DVD, a Blue-Ray, a CD, a ROM, a PROM, an EPROM, an EEPROM or a FLASH memory, having electrically controlled control signals readable stored therein, which cooperates (or is capable of cooperating) with a programmable computer system so that the respective method is carried out. In this way, the digital storage medium can be computer readable.

[000570] Some embodiments, according to the invention, comprise a data loader having electrically readable control signals, which can cooperate with a programmable computer system, so that one of the methods described herein is carried out.

[000571] Generally, embodiments of the present invention may be implemented as a computer program product with a program code, the program code being operative to carry out one of the methods when the computer program product operates on a computer. Program code can, for example, be stored in a machine-readable loader.

[000572] Other embodiments comprise the computer program for carrying out one of the methods described herein, stored in a machine-readable loader.

[000573] In other words, an embodiment of the inventive method is, in this way, a computer program having a program code for carrying out one of the methods described here, when the computer program is executed on a computer.

[000574] Another embodiment of the inventive method is, in this way, a data loader (or a digital storage medium, or a computer-readable medium) comprising, registered therein, the computer program for carrying out one of the methods described here. The data carrier, digital storage medium, or recorded medium is typically tangible and/or non-transitory.

[000575] Another embodiment of the inventive method is, therefore, a data stream or a sequence of signals that represents the computer program for carrying out one of the methods described here. The data stream or sequence of signals may, for example, be configured to be transferred via a data communication connection, for example via the Internet.

[000576] Another embodiment comprises a processing means, for example, a computer, or a programmable logic device, configured or adapted to perform one of the methods described here.

[000577] Another embodiment comprises a computer having computer program installed therein to perform one of the methods described herein.

[000578] Another embodiment, according to the invention, comprises an apparatus or a system configured to transfer (for example, electronically or optically) a computer program for carrying out one of the methods described here to a receiver. The receiver may, for example, be a computer, a mobile device, a memory device or the like. The apparatus or system may, for example, comprise a file server for transferring the computer program to the receiver.

[000579] In some embodiments, a programmable logic device (e.g., a field-programmable gate array) may be used to perform some or all of the functionality of the methods described herein. In some embodiments, a field-programmable gate array may cooperate with a microprocessor to perform one of the methods described herein. Generally, the methods are preferably performed by any hardware apparatus.

[000580] The embodiments described above are merely illustrative of the principles of the present invention. It is understood that modifications and variations of the arrangements and details described here will be apparent to others skilled in the art. It is intended, therefore, to be limited only by the scope of the claims of the impending patent and not by the specific details presented in the form of a description and explanation of the embodiments herein.

[000581] 17. Conclusions

[000582] To conclude, the embodiments, according to the invention, comprise one or more of the following aspects, wherein the aspects can be used individually or in combination.

[000583] Hashing context mechanism state

[000584] According to one aspect of the invention, the states in the hash table are considered to be meaningful states and group boundaries. This allows you to significantly reduce the size of the required tables.

[000585] Additional context update

[000586] According to one aspect, some embodiments, according to the invention, comprise a computationally efficient way to update the context. Some realizations use an additional context update in which a numeric current context value is derived from a numeric previous context value.

[000587] Context derivation

[000588] According to one aspect of the invention, using a sum of two absolute spectral values is the association of a truncation. It is a type of gain vector quantization of spectral coefficients (as opposed to conventional gain vector quantization). It aims to limit the order of the context, while carrying the most significant information from the next one.

[000589] Updated tables

[000590] According to one aspect of the invention, the optimized tables ari_hash_m[742], ari_lookup_m[742] and ari_cf_m[64][17], which provide a particularly good compromise between coding efficiency and computational complexity are applied.

[000591] Other technologies, which are applied in the embodiments, according to the invention, are described in the non-pre-published patent applications PCT EP2101/065725, PCT EP2010/065726, and PCT EP 2010/065727. Furthermore, in some embodiments according to the invention, a stop symbol is used. Furthermore, in some embodiments, only unsigned values are considered for context.

[000592] However, non-pre-published International patent applications reveal aspects that are still in some embodiments according to the invention.

[000593] For example, a zero region identification is used in some embodiments of the invention. Thus, the so-called “small value indicator” is defined (e.g. bit 16 of the current numeric context value c).

[000594] In some embodiments, region-dependent context calculation may be used. However, in other embodiments, a region-dependent context calculation may be omitted in order to keep the complexity and size of the tables reasonably small.

[000595] Furthermore, the hashing context using a hash function is an important aspect of the invention. The hashing context may be based on the two-table concept that is described in the non-pre-published International patent applications referenced above. However, specific adaptations to the hashing context can be used in some realizations in order to increase computational efficiency. However, in other embodiments, according to the invention, the hashing context that is described in the non-prepublished International patent applications referenced above may be used.

[000596] Furthermore, it should be noted that the additional hashing context is quite simple and computationally efficient. Furthermore, the context independence of value signals, which is used in some embodiments of the invention, helps to simplify the context, thus keeping memory requirements reasonably low.

[000597] In some embodiments of the invention, a context derivation using a sum of two spectral values and a context limitation is used. These two aspects can be combined. Both aim to limit the context order by carrying the most significant information from the next one.

[000598] In some embodiments, a small value indicator is used that may be similar to identifying a group of several zero values.

[000599] In some embodiments, according to the invention, an arithmetic stop mechanism is used. The concept is similar to the use of an “end of block” symbol in JPEG, which has a comparable function. However, in some embodiments of the invention, the symbol (“ARITH_STOP”) is not explicitly included in the entropy encoder. Furthermore, a combination of existing symbols, which could not previously occur, is used, that is, “ESC+0”. In other words, the audio decoder is configured to detect a combination of existing symbols, which are not normally used to represent a numeric value, and to interpret the occurrence of such a combination of existing symbols as an arithmetic stop condition.

[000600] An embodiment, according to the invention, uses a two-table context hashing mechanism.

[000601] To summarize, some embodiments, according to the invention, may comprise one or more of the five main aspects.

[000602] Improved tables;

[000603] extended context to detect both zero regions and small amplitude regions in the near;

[000604] hashing context;

[000605] context state generation: additional context state update; It is

[000606] context derivation: specific quantization of context values including sum of amplitudes and limiting.

[000607] To conclude, one aspect of the embodiments according to the present invention remains in a further context update. The realizations according to the invention comprise an efficient concept for updating the context, which avoids the extensive calculations of the working project (e.g. of working project 5). Furthermore, simple change operations and logical operations are used in some realizations. The simple context update makes context calculation significantly easier.

[000608] In some embodiments, the context is independent of the sign of the values (e.g., the decoded spectral values). This independence of the context from the sign of the values brings with it a reduced complexity of the variable context. This concept is based on the observation that a negation of the signal in context does not bring with it a severe degradation of coding efficiency.

[000609] According to one aspect of the invention, the context is derived using the sum of two spectral values. Thus, the memory requirements for storing the context are significantly reduced. Thus, the use of a context value, which represents the sum of two spectral values, can be considered advantageous in some cases.

[000610] Furthermore, limiting the context brings a significant improvement in some cases. In addition to context derivation using the sum of two spectral values, the context matrix entries “q” are limited to a maximum value of “0xF” in some realizations which in turn results in a limitation of memory requirements. This limitation of the matrix of context values matrix “q” brings some advantages.

[000611] In some embodiments, the so-called “small value indicator” is used. By obtaining the context variable c (which is also designated as a numeric current context value), an indicator is defined if the values of some inputs “q[1][i-3]” to “q[1][i- 1]” are too small. Thus, context calculation can be performed with high efficiency. Particularly meaningful context values (e.g. numeric current context value) can be obtained.

[000612] In some embodiments, an arithmetic stop mechanism is used. The “ARITH_STOP” mechanism allows efficient stopping of arithmetic encoding or decoding if there are only zero values left. Thus, coding efficiency can be improved at moderate costs in terms of complexity.

[000613] According to one aspect of the invention, a two-table hashing context mechanism is used. Context mapping is performed using a range splitting algorithm that evaluates the “ari_hash_m” table in combination with a subsequent lookup table evaluation of the “ari_lookup_m” table. This algorithm is more efficient than the WD3 algorithm.

[000614] Next, some additional details will be discussed.

[000615] It should be noted here that the tables “arith_hash_m[742]” and “arith_lookup_m[742]” are two distinct tables. The first is used to map a single context index (e.g. numeric context value) to the probability model index (e.g. mapping rule index value) and the second is used to map a group of consecutive contexts , delimited by the context indices in “arith_hash_m[]”, in the single probability model.

[000616] It should also be noted that the “arith_cf_msb[64][16]” table can be used as an alternative to the “ari_cf_m[64][17]” table, although the dimensions are slightly different. “ari_cf_m[][]” and “ari_cf_msb[][]” can refer to the same table, as the 17th coefficients of the probability model are always zero. Sometimes it is not taken into account when counting the space required to store tables.

[000617] To summarize the above, some embodiments, according to the invention, provide a new proposed silent encoding (encoding or decoding), which generate modifications in the USAC MPEG working project (for example, in the USAC MPEG 5 working project ). Said modifications can be seen in the attached figures and also in the related description.

[000618] As a conclusion, it should be noted that the prefix “ari” and the prefix “arith” in the names of variables, matrices, functions, and so on, are used interchangeably.

Claims (6)

1. AUDIO DECODER (200; 800) to provide decoded audio information (212; 812) based on encoded audio information (210; 810), the audio decoder is characterized by comprising: an arithmetic decoder (230 ; 820) to provide various decoded spectral values (232; 822) based on an arithmetically encoded representation (222; 821) of the spectral values comprised in the encoded audio information (210; 810); and a frequency domain to time domain converter (260; 830) for providing a time domain audio representation (262; 812) using the decoded spectral values (232; 822), to obtain the decoded audio information ( 212; 812); wherein the arithmetic decoder (232; 820) is configured to select a mapping rule (297; cum_freq[]) that describes a mapping of a code value (value) to the arithmetically encoded representation of the spectral values representing one or more of the values spectral values, or a most significant bit plane of one or more of the spectral values, in a coded form, in a symbol code representing one or more of the spectral values, or a most significant bit plane of one or more of the spectral values, in a decoded form, in dependence on a context state (s) described by a current numeric context value (c); configured to determine the current numerical context value (c) depending on the various previously decoded spectral values; wherein the arithmetic decoder is configured to evaluate a hash table (ari_hash_m[]), whose entries define both the significant state values among the numeric context values and the limits of the ranges of the non-significant state values among the numeric context values , to select the mapping rule, where the hash table ari_hash_m is defined as shown in figures 22(1), 22(2), 22(3) and 22(4); wherein the arithmetic decoder is configured to evaluate the hash table (ari_hash_m), to determine whether the current numeric context value is identical to a table context value described by a hash table entry (ari_hash_m), or to determine a described range by the entries of the hash table (ari_hash_m) in which the current numeric context value is found, and to derive a mapping rule index (pki) value that describes a mapping rule selected in dependence on an evaluation result, where a mapping rule index value is individually associated with a numeric context value being a meaningful state value. 2. AUDIO DECODER, according to claim 1, characterized in that the arithmetic decoder is configured to compare the current numeric context value (c), or a scaled version (s) of the current numeric context value, with a series of numerically ordered entries (j=ari_hash_m[i]) or hash table sub-entries (ari_hash_m[]), to iteratively obtain a hash table index value (i_max) from a table entry (ari_lookup_m[i_max]), such that the current numeric context value (c) lies within a range defined by the obtained hash table entry (ari_hash_m[i_max]) designated by the obtained hash table index value (i_max) and an adjacent hash table entry ( ari_hash_m[i_max-1]), and wherein the arithmetic decoder is configured to determine a next entry in the series of hash table entries (ari_hash_m[]) depending on a result of a comparison between the current numeric context value (c ), or a scaled version(s) of the current numeric context value, and a current entry or sub-entry (ari_hash_m[i]) of the hash table. 3. AUDIO DECODER, according to claim 2, characterized in that the arithmetic decoder is configured to select a mapping rule defined by a second sub-entry (j&0xFF) of the hash table (ari_hash_m) designated by the current index value of the hash table (i) if it is observed that the current numeric context value (c) or a scaled version (s) thereof is equal to the first sub-entry (j>>8) of the hash table (j=ari_hash_m[i]) designated by the value of the current index of the hash table (i). 4. AUDIO DECODER, according to claim 2 or 3, characterized in that the arithmetic decoder is configured to select a mapping rule defined by an entry or sub-entry (ari_lookup_m[i_max]) of an ari_lookup_m mapping table if is not observed that the current numeric context value is equal to a sub-entry of the hash table (ari_hash_m), where the arithmetic decoder is configured to choose the entry or sub-entry of the mapping table depending on the index value of the Iteratively obtained hash table (i_max). 5. AUDIO DECODER, according to one of claims 1 to 4, characterized in that the arithmetic decoder is configured to selectably provide an index value of the mapping rule defined by the hash table entry designated by the current index value of the hash table if it is observed that the current numeric context value (c) is equal to a value (j>>8) defined by the entry (ari_hash_m[i]) of the hash table designated by the current index value of the hash table (i) . 6. METHOD FOR PROVIDING A DECODED AUDIO INFORMATION (212; 812) BASED ON A CODED AUDIO INFORMATION (210; 810), the method is characterized by comprising: providing several decoded spectral values (232; 822) based on a arithmetically encoded representation (222; 821) of the spectral values comprised in the encoded audio information (210; 810); and providing a time domain audio representation (262; 812) using the decoded spectral values (232; 822), to obtain the decoded audio information (212; 812); wherein providing a plurality of decoded spectral values comprises selecting a mapping rule (297; cum_freq[]) that describes a mapping of a code value (value) to the arithmetically encoded representation of the spectral values representing one or more of the spectral values, or a plane of one or more of the spectral values, in a form encoded in a symbol code representing one or more of the spectral values, or a most significant bit plane of one or more of the spectral values, in a form decoded, depending on a context state (s) described by a current numeric context value (c); wherein the current numerical context value (c) is determined depending on the various previously decoded spectral values; wherein a hash table (ari_hash_m[]), whose entries define both the significant state values among the numeric context values and the boundaries of the ranges of the non-significant state values among the numeric context values, is evaluated to select the rule mapping, where the hash table ari_hash_m is defined as shown in figures 22(1), 22(2), 22(3) and 22(4); wherein the hash table (ari_hash_m) is evaluated to determine whether the current numeric context value is identical to a table context value described by a hash table entry (ari_hash_m) or to determine a range described by the hash table entries ( ari_hash_m) in which the current numeric context value lies, and in which a mapping rule index (pki) value describing a selected mapping rule is derived depending on an evaluation result; wherein a mapping rule index value is individually associated with a numeric context value being a meaningful state value.