EP1692689B1 - Optimized multiple coding method - Google Patents

Optimized multiple coding method Download PDF

Info

Publication number
EP1692689B1
EP1692689B1 EP04805538A EP04805538A EP1692689B1 EP 1692689 B1 EP1692689 B1 EP 1692689B1 EP 04805538 A EP04805538 A EP 04805538A EP 04805538 A EP04805538 A EP 04805538A EP 1692689 B1 EP1692689 B1 EP 1692689B1
Authority
EP
European Patent Office
Prior art keywords
coders
coder
coding
data rate
unit
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Not-in-force
Application number
EP04805538A
Other languages
German (de)
French (fr)
Other versions
EP1692689A1 (en
Inventor
David Virette
Claude Lamblin
Abdellatif Benjelloun Touimi
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Orange SA
Original Assignee
France Telecom SA
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by France Telecom SA filed Critical France Telecom SA
Priority to PL04805538T priority Critical patent/PL1692689T3/en
Publication of EP1692689A1 publication Critical patent/EP1692689A1/en
Application granted granted Critical
Publication of EP1692689B1 publication Critical patent/EP1692689B1/en
Not-in-force legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/002Dynamic bit allocation
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/02Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using spectral analysis, e.g. transform vocoders or subband vocoders
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/02Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using spectral analysis, e.g. transform vocoders or subband vocoders
    • G10L19/0212Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using spectral analysis, e.g. transform vocoders or subband vocoders using orthogonal transformation
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/04Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using predictive techniques
    • G10L19/08Determination or coding of the excitation function; Determination or coding of the long-term prediction parameters
    • G10L19/12Determination or coding of the excitation function; Determination or coding of the long-term prediction parameters the excitation function being a code excitation, e.g. in code excited linear prediction [CELP] vocoders
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/04Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using predictive techniques
    • G10L19/16Vocoder architecture
    • G10L19/18Vocoders using multiple modes

Definitions

  • the present invention relates to the encoding / decoding of digital signals, in applications for transmission or storage of multimedia signals such as audio signals (speech and / or sounds) or video.
  • the present invention is in the context of an optimization of " multiple coding " techniques, implemented as soon as a digital signal, or a portion of this signal, is coded according to several coding techniques.
  • This multiple coding can be performed simultaneously (in one pass) or not.
  • the processes can be carried out on the same signal, or possibly on versions derived from the same signal (for example according to different bandwidths).
  • Each coder carries out the compression of a version resulting from the decoding of the signal compressed by the preceding encoder.
  • Multiple coding is, for example, in the case of the same content which is coded according to several formats and then transmitted to terminals that do not support the same coding formats. If it is a real-time broadcast, the processing should be done simultaneously. If it is a question of access to a database, the codings can be carried out one after another, delayed. In these examples, the multiple coding makes it possible to code the same signal in different formats by using several coders (or possibly several rates or several modes of the same encoder), each encoder operating independently of the other coders.
  • multi-mode coding (with reference to the selection of a " mode " of coding).
  • mode the number of coders sharing a " common past " are required to encode the same signal portion.
  • the coding techniques used may be different, or from a single coding structure. However, they will not be completely independent unless they are " memory-free " techniques.
  • the second use case mentioned concerns multi-mode coding applications, allowing the selection of one encoder from a set for each portion of signal analyzed.
  • the selection requires the definition of a criterion, the most common aiming at the optimization of the rate-distortion compromise.
  • the signal being analyzed over successive time segments, at each segment several codings are evaluated.
  • the lowest bit rate coding for a given quality is then selected, or the best bit rate coding for a given bit rate. It will be noted that other constraints than those of flow / distortion can be used.
  • the coding selection is made “a priori” by an analysis of the signal on the segment in question (selection according to the characteristics of the signal).
  • selection according to the characteristics of the signal has led to propose a selection "a posteriori" the optimum mode after coding all the modes, however the complexity of high prices.
  • the decision a priori is made from a classification of the input signal.
  • signal classification There are many methods of signal classification.
  • US 6,581,032 discloses a speech compression system comprising four codecs selectively activated according to a signal to be compressed and according to a classification of the signals.
  • US 6,141,638 discloses an encoder using different code dictionaries according to parameters of the signal to be encoded.
  • the present invention improves the situation.
  • the above steps are implemented by a computer program product comprising program instructions for this purpose.
  • the present invention also aims at such a computer program product, intended to be stored in a memory of a processing unit, in particular a computer or a mobile terminal, or on a removable memory medium and intended to cooperate with a drive of the processing unit.
  • the present invention also aims at a device for aiding compression coding, for the implementation of the method according to the invention, and then including a memory adapted to store instructions of a computer program product of the aforementioned type.
  • FIG. 1a On which there is shown a plurality of encoders C0, C1, ..., CN, in parallel and each receiving an input signal s 0 .
  • Each encoder comprises functional blocks BF1 to BFn for implementing successive coding steps and ultimately outputting a coded bitstream BS0, BS1, ..., BSN. It is furthermore indicated that in an application in multi-mode coding, the outputs of the coders C0 to CN are connected to an optimal mode selection module MM and the bit stream BS of the optimal coder is transmitted (dotted line arrows of the figure 1a ).
  • Some BFi function blocks are sometimes identical from one mode (or encoder) to another, while others differ only in quantizer level. Usable relationships also exist when encoders from the same coding family are used, using similar models or computing parameters physically related to the signal.
  • the figure 1b illustrates the proposed solution.
  • the aforementioned " common " operations are performed once for at least a portion of the coders and, preferably, for all the coders, in an independent module MI which will redistribute the results. obtained at least part of the coders, or preferably all these coders. It is thus a sharing between at least part of all coders C0 to CN (or " pooling " below) the results obtained.
  • Such an independent module MI may be part of a device for a multiple compression coding as defined above.
  • the existing functional block or blocks BF1 to BFn of the same or more different coders is used, this or these coders being chosen according to criteria which will be described later.
  • the present invention can implement several strategies which, of course, may differ depending on the role of the functional block considered.
  • a first strategy is to use the parameters of the encoder whose bit rate is the lowest to focus the search parameters for all other modes.
  • a second strategy is to use the parameters of the encoder whose rate is the highest, then to " degrade " progressively to the encoder whose bit rate is the lowest.
  • the present invention makes it possible to reduce the complexity of the calculations preliminary to the a posteriori selection of an encoder performed in the last step, for example by the last module MM before the transmission of the bit stream BS.
  • MSPi partial selection module
  • FIG. 1d A more sophisticated variant of the multi-mode structure based on the functional block cutting described above is now proposed, with reference to the figure 1d .
  • the multi-mode structure of the figure 1d is called "trellis", with several possible paths in the trellis.
  • all the possible paths of the lattice are represented so that it is in a tree form.
  • each path of the trellis is defined by a combination of operating modes of the functional blocks, each functional block supplying several possible variants of the next functional block.
  • each coding mode is derived from the combination of operating modes of the functional blocks: the functional block 1 has N 1 operating modes, the functional block 2 has N 2 , and so on up to the block P.
  • a first feature of this structure is that it provides, for a given functional block, a common calculation module per output of the previous functional block. These common calculation modules perform the same operations, but on the basis of different signals since they come from different previous blocks.
  • the common calculation modules of the same level are pooled: the results of a given module usable by the following modules are provided to these following modules.
  • a partial selection, made at the end of the processing of each functional block advantageously makes it possible to eliminate the less efficient branches according to the chosen criterion. It is therefore possible to reduce the number of branches of the trellis to be evaluated.
  • the chosen trellis path is the one passing through the lowest flow functional block, or the highest rate functional block according to the coding context, and the results obtained from the lowest (or highest) bit rate functional block are adapted to the bit rates of at least a portion of the other functional blocks by a focused search of parameters for at least part of all other functional blocks, up to the highest (or lowest) rate functional block.
  • CELP Code Excited Linear Prediction
  • the reconstructed signal synthesis model is used at the encoder to extract the parameters modeling the signals to be coded.
  • These signals can be sampled at the frequency of 8 kHz (telephone band 300-3400 Hz) or at a higher frequency, for example at 16 kHz for wideband coding (bandwidth 50 Hz to 7 kHz).
  • the compression ratio varies from 1 to 16.
  • These encoders operate at rates of 2 to 16 kbit / s in the telephone band, and at speeds of 6 to 32 kbit / s in the extended band. .
  • the CELP type digital coding device currently the most useful synthesis analysis coder, is presented to the figure 3 in the form of main functional blocks.
  • the speech signal s 0 is sampled and converted into a sequence of frames of a number L of samples. Each frame is synthesized by filtering a waveform extracted from a directory (called " dictionary "), multiplied by a gain, through two filters varying in time.
  • the fixed excitation dictionary is a finite set of waveforms of the L samples.
  • the first filter is a long-term prediction filter. A " LTP " ( Long Term Prediction ) analysis makes it possible to evaluate the parameters of this long-term predictor which exploits the periodicity of the voiced sounds, this harmonic component being modeled in the form of an adaptive dictionary (block 32) .
  • the second filter is a short-term prediction filter.
  • the " LPC " ( Linear Prediction Coding ) analysis methods make it possible to obtain these short-term prediction parameters, which are representative of the vocal tract transfer function and characteristics of the envelope of the signal spectrum.
  • the method used to determine the innovation sequence is the synthetic analysis method which is summarized as follows. At the encoder, a large number of innovation sequences of the fixed excitation dictionary are filtered by the LPC filter (synthesis filter of the function block 34 of the figure 3 ). Beforehand, the adaptive excitation was obtained in a similar way.
  • the selected waveform is that producing the synthetic signal closest to the original signal (error minimization at function block 35), according to a perceptual weighting criterion (function block 36) which is generally known as criterion name " CELP ".
  • Decoding is, for its part, much less complex than coding.
  • the bitstream generated by the coder enables the decoder, after demultiplexing, to obtain the quantization index of each parameter.
  • the decoding of the parameters and the application of the synthesis model then make it possible to reconstruct the signal.
  • the first embodiment relates to the perceptual frequency coder called " TDAC " and described in particular in the published document. US 2001 / 027,393 .
  • This TDAC encoder is used to encode digital audio signals sampled at 16 kHz (wide band).
  • the figure 4a illustrates the main functional blocks of this encoder.
  • An audio signal x (n) limited in band at 7 kHz and sampled at 16 kHz is cut into frames of 320 samples (20 ms).
  • a Modified Discrete Cosine Transform (or " MDCT ") is applied (function block 41) on 640 sample input signal frames with 50% overlap, thus with a refresh of MDCT analysis every 20 ms .
  • the spectrum is limited to 7225 Hz by setting the last 31 coefficients to zero (only the first 289 coefficients are different of 0).
  • a masking curve (block 42) is determined from this spectrum and all masked coefficients are set to zero.
  • the spectrum is divided into 32 bands of unequal widths. Any masked bands are determined according to the transformed coefficients of the signals. For each band of the spectrum, the energy of the MDCT coefficients is calculated (to obtain scale factors).
  • the 32 scale factors constitute the spectral envelope of the signal which is then quantized and coded by entropy encoding (function block 43), and finally transmitted in the coded frame s c .
  • the dynamic allocation of the bits is based on a band masking curve (functional block 42) calculated from the decoded and dequantized version of the spectral envelope. This measurement makes it possible to have compatibility between the bit allocation of the encoder and the decoder.
  • the normalized MDCT coefficients in each band are then quantized (function block 45) by vector quantizers using size-nested dictionaries, the dictionaries being composed of a type II permutation code union.
  • the information on the tone (coded here on a bit B 1 ) and the voicing (coded here on a bit B 0 ), as well as the spectral envelope eq (i) and the coefficients coded y q (j) are multiplexed ( block 46 of the figure 4a ) and transmitted in frames.
  • the functional blocks shared by the coders bear the same reference as those of a single TDAC coder as represented in FIG. figure 4a .
  • the bit allocation block 44 is used in several passes, and the number of bits allocated is adjusted for the transquantification performed by each coder (blocks 45_1,..., 45_ (K-2), 45_ (K -1)), as will be seen below. Note further that these transquantifications use the results obtained by the quantization function block 45_0 for a chosen encoder, index 0 (the lowest rate encoder in the example described).
  • the only functional blocks of the encoders which act without real interaction are the multiplexing blocks 46_0, 46_1, ..., 46_ (K-2), 46_ (K-1), although they all use the same information of voicing and tone, as well as the same coded spectral envelope. As such, it is simply stated that a partial pooling of the multiplexing can be conducted, again.
  • the strategy employed is to exploit the results of the two bit allocation and quantization functional blocks made for the bit stream (0), at the lowest bit rate D 0 , to speed up the operations of the two corresponding function blocks for the K-1 other bitstream (k) (1 ⁇ k ⁇ K ). It is also possible to consider the multi-rate coding scheme which uses a bit-allocation functional block per bit stream (without factorization provided for this block) but mutualizes a part of the quantization operations thereafter.
  • the multiple coding techniques presented below are advantageously based on intelligent transcoding used for the reduction of the coded audio stream bit rate, generally located in a node of the network.
  • bit streams k , 0 ⁇ k ⁇ K are classified in an increasing order of rates ( D 0 ⁇ D 1 ⁇ ... ⁇ D K-1 ) .
  • bit stream 0 corresponds to the lowest bit rate.
  • a second phase is used to perform the readjustment. This step is preferably done by a succession of iterative operations based on a perceptual criterion that adds or removes bits from the bands.
  • the bits are added to the bands where the perceptual improvement is the most important. This perceptual improvement is measured by the variation of the noise to mask ratio between the initial and final allocation of the bands. The rate is increased for the band where this variation is greatest. In the opposite case where the total number of distributed bits is greater than that available, the extraction of bits on the bands is dual to the latter procedure.
  • the first determination step by the above formula can be done once based on the lowest bit rate D 0 .
  • the notation ⁇ ( k ) indicates the parameter used in the processing performed to obtain the bitstream of the encoder k.
  • the parameters without this exponent being calculated once and for all for the bit stream 0. They are independent of the flow (or mode) considered.
  • the MPEG-1 Layer I & II coder presented to the figure 6a uses a filter bank with 32 uniform sub-bands (block 61 of the figure 6a ) to perform the time / frequency transformation of the input audio signal s 0.
  • the output samples of each subband are grouped and then normalized by a common scale factor (determined by the function block 67) before being quantified (block 62).
  • the number of levels of the uniform scalar quantizer used for each subband results from a dynamic bit allocation procedure (performed by block 63). This procedure uses a psychoacoustic model (block 64) to determine the bit distribution that makes the quantization noise as noticeable as possible.
  • the hearing models proposed in the standard are based on the estimation of the spectrum obtained by a fast Fourier transform (FFT) of the input temporal signal (made by block 65).
  • FFT fast Fourier transform
  • the frame s c multiplexed by the block 66 of the figure 6a and which is finally transmitted, contains, after a header field H D , the set of samples of the quantized sub-bands E SB , which represent the main information, and complementary information used for the decoding operation constituted by the scaling factors F E and the bit allocation A i .
  • the two blocks 64 and 65 already provide the signal to mask ratios (SMR arrows Figures 6a and 7 ), used for the bit allocation procedure (block 70 of the figure 7 ).
  • Steps 1 and 2 are repeated iteratively until the total number of available bits, corresponding to the operating rate, is distributed.
  • the result is then a bit distribution vector ( b 0 , b 1 , ..., b M -1 ) .
  • the K outputs of this bit allocation block then feed the quantization blocks for each of the bit streams at the given bit rate.
  • the last exemplary embodiment relates to the coding of the multi-mode speech with a posteriori decision from the 3GPP NB-AMR (" Narrow-Band Adaptive Multi-Rate ”) coder which is a multi-rate bandband speech encoder. adaptive, according to a 3GPP standard.
  • This encoder which belongs to the well-known family of CELP coders whose principle was briefly described above, has eight modes (or bit rates) ranging from 12.2 kbit / s to 4.75 kbit / s, all based on the technique ACELP (for " Algebraic Code Excited Linear Prediction ").
  • the figure 8 gives the coding scheme in functional blocks of this encoder. This structure was exploited in order to realize a post-decision multi-mode encoder, based on 4 modes of the NB-AMR encoder (7.4, 6.7, 5.9, 5.15).
  • the complexity is even smaller.
  • the non-identical functional block calculations for some modes are accelerated by exploiting those of another mode or a common processing module, as will be seen below.
  • the results of the four encodings thus shared are then different from those of the four codings in parallel.
  • the functional blocks of these four modes are used for trellis multi-mode coding, as has been seen above with reference to FIG. figure 1d .
  • the 3GPP NB-AMR coder is working on a 3.4 kHz band-limited speech signal sampled at 8 kHz cut into 20 ms frames (160 samples). Each frame has 4 subframes of 5 ms (40 samples) grouped 2 by 2 in " super subframes " of 10 ms (80 samples). For all modes, the same types of parameters are extracted from the signal but with variants of modeling and / or quantification of these parameters. In the NB-AMR encoder, five types of parameters are to be analyzed and coded. Line Spectral Pairs (LSP) parameters are processed once per frame for all modes, except for 12.2 mode (so once per super subframe). The other parameters (in particular the LTP delay, the gain of the adaptive excitation, the fixed excitation, the gain of the fixed excitation) are processed once per subframe.
  • LSP Line Spectral Pairs
  • the four modes considered here (7.4, 6.7, 5.9, 5.15) are distinguished essentially by the quantifications of their parameters.
  • the binary allocation of these 4 modes is summarized in Table 1 below. ⁇ b> Table 1: ⁇ / b> Bit allocation of the 4 modes (7.4, 6.7, 5.9, 5.15) of the 3GPP NB-AMR encoder Mode (kbit / s) 7.4 6.7 5.9 5.15
  • These 4 modes of the NB-AMR encoder (7.4, 6.7, 5.9, 5.15) have identical modules such as preprocessing, analysis of linear prediction coefficients, signal calculation weighted.
  • the signal preprocessing is 80 Hz high-pass cut-off filtering to suppress the continuous components combined with division of the input signals to avoid overflows.
  • the quantization of the LSP parameters from , 15 kbit / s is done on 23 bits, that of the other three modes on 26 bits.
  • the Cartesian product vector quantization (so-called " split VQ ") of the LSP parameters divides the LSP parameters into 3 sub-vectors, of size 3, 3 and 4.
  • the first sub-vector composed of The first 3 LSP is quantized on 8 bits by the same dictionary for the four modes.
  • the second sub-vector composed of the following 3 LSPs is quantized for the 3 high-speed modes by a dictionary of size 512 (9 bits) and for the mode with 5,15 by half of this dictionary (one vector out of 2).
  • the third and last sub-vector composed of the last 4 LSPs is quantized for the 3 high-speed modes by a dictionary of size 512 (9 bits) and for the mode of lower bit rate by a dictionary of size 128 (7 bits).
  • the transformation in the normalized frequency domain, the calculation of the squared error criterion weights and the MA prediction (for " Moving Average ") of the LSP residue to be quantized are identical for the 4 modes.
  • the three broadband modes use the same dictionaries to quantify the LSPs, they can share, in addition to the same vector quantization module, the inverse transformation (to return from the normalized frequency domain to the cosine domain), as well as the calculation of the LSPs.
  • the closed-loop searches of the adaptive and fixed excitations are done sequentially and require the calculation of the impulse response of the weighted synthesis filter and then of the target signals beforehand.
  • the impulse response of the weighted synthesis filter (A i (z / ⁇ 1 ) / [A Q i (z) A i (z / ⁇ 2 )]) is identical for the 3 high-speed modes (7.4, 6, 7; 5,9).
  • the calculation of the target signal for the adaptive excitation depends on the weighted signal (regardless of the mode); the quantized filter A Q i (z) (identical for 3 modes) and the past subframe (different for each subframe other than the first subframe).
  • the target signal for the fixed excitation is obtained by removing from the previous target signal the contribution of the filtered adaptive excitation of this subframe (which is different from one mode to another except for the first subframe of the first 3 modes).
  • the search in this dictionary of absolute delays is focused around the delay found in open loop (range of ⁇ 5 for the 5.15 mode, ⁇ 3 for the other modes).
  • the target signal and the open-loop delay being identical, the result of this closed-loop search is also identical.
  • the other two dictionaries are of differential type and make it possible to code the difference between the current delay and the integer delay T i-1 closest to the fractional delay of the preceding sub-frame.
  • the first 5-bit differential dictionary used for the odd subframes of the 7.4 mode, is 1/3 resolution around the entire delay T i-1 in the interval [T i-1 -5 + 2 / 3, T i-1 +4 + 2/3].
  • the second 4-bit differential dictionary included in the first one, is used for the odd subframes of the modes at 6.7 and 5.9 as well as for the last three subframes of the 5.15 mode.
  • This second dictionary is of integer resolution around the integer delay T i-1 in the interval [T i-1 -5, T i-1 +4] plus a resolution of 1/3 in the interval [T i-1 -1 + 2/3, T i-1 + 2/3].
  • ACELP Interleaved Single-Pulse Permutation
  • the four modes (7.4, 6.7, 5.9, 5.15) use the same slice of the 40 samples of a 5-track subframe of length 8 interleaved, as shown in Table 2a.
  • Table 2b shows, for the 3 modes (7.4, 6.7, 5.9) the dictionary rate, the number of pulses and their distribution in the tracks.
  • the distribution of the 2 pulses of the ACELP 9-bit dictionary of the 5.15 mode is even more constrained.
  • the gains of the adaptive and fixed excitations are quantified on 7 or 6 bits (with an MA prediction applied to the gain of the fixed excitation) by a joint vector quantization minimizing the CELP criterion.
  • Non-identical functional blocks can be accelerated by exploiting those of another mode or a common processing module. Depending on the constraints of the application (in terms of quality and / or complexity), different variants can be used. Some examples are described below. It is also possible to rely on intelligent transcoding techniques between CELP coders.
  • This implementation gives a result identical to that of the non-optimized multi-mode coding. If one wishes to further reduce the complexity of the quantization, one can stop at step 1 and take Y 1 as a quantized vector for the high-speed modes if this vector is considered sufficiently close to Y. This simplification can therefore give a different result from an exhaustive search.
  • FIG. 1d it is proposed to realize a multi-mode trellis coder for several combinations of functional blocks, each functional block having at least two modes of operation (or flows).
  • This new encoder was constructed from the four NB-AMR encoder rates mentioned above (5.15, 5.90, 6.70, 7.40).
  • this encoder there are four functional blocks: the LPC block, the LTP block, the fixed excitation block and the gain block. Referring to Table 1 presented above, Table 3a below summarizes for each of these functional blocks, its number of flow rates and its flow rates.
  • Functional block Number of flows Flow of the functional blocks LPC (LSP) 2 26 and 23 LTP delay 3 26, 24 and 20 Fixed excitation 4 68, 56, 44 and 36 Earnings 2 28 and 24
  • the multi-rate encoder thus obtained has a high granularity in rates, with 32 possible modes given in Table 3b. However, it is indicated that the encoder thus obtained is not interoperable with the aforementioned NB-AMR encoder.
  • Table 3b the modes corresponding to the three flows of the NB-AMR (5.15, 5.90, 6.70) are shown in bold, the exclusion of the highest bit rate of the LTP functional block eliminating the flow of 7 40.
  • the present invention makes it possible to provide an effective solution to the problem of the complexity of multiple codings by pooling and accelerating the calculations implemented by the various coders.
  • the coding structures can therefore be represented using functional blocks describing the various operations performed during a treatment.
  • the functional blocks of the different encodings implemented in multiple coding have strong relationships that are exploited within the meaning of the present invention. These relations are particularly strong when the different codings correspond to different modes of the same structure.
  • the present invention is flexible from the point of view of complexity. It is indeed possible to decide a priori the maximum complexity of the multiple coding and to adapt the number of coders explored as a function of this complexity.

Abstract

The invention relates to the compression coding of digital signals such as multimedia signals (audio or video), and more particularly a method for multiple coding, wherein several encoders each comprising a series of functional blocks receive an input signal in parallel. Accordingly, a method is provided in which, a) the functional blocks forming each encoder are identified, along with one or several functions carried out of each block, b) functions which are common to various encoders are itemized and c) said common functions are carried out definitively for a part of at least all of the encoders within at least one same calculation module.

Description

La présente invention concerne le codage/décodage de signaux numériques, dans des applications de transmission ou de stockage de signaux multimédia tels que les signaux audio (parole et/ou sons) ou vidéo.The present invention relates to the encoding / decoding of digital signals, in applications for transmission or storage of multimedia signals such as audio signals (speech and / or sounds) or video.

Pour offrir mobilité et continuité, les services de communication multimédia modernes et innovants doivent pouvoir fonctionner dans une grande variété de conditions. Le dynamisme du secteur de la communication multimédia, l'hétérogénéité des réseaux, de l'accès et des terminaux ont engendré une prolifération de formats de compression.To provide mobility and continuity, modern and innovative multimedia communication services must be able to operate under a wide variety of conditions. The dynamism of the multimedia communication sector, the heterogeneity of networks, access and terminals have led to a proliferation of compression formats.

La présente invention s'inscrit dans ie contexte d'une optimisation des techniques de "codage multiple", mises en oeuvre dès lors qu'un signal numérique, ou une portion de ce signal, est codé selon plusieurs techniques de codage. Ce codage multiple peut être effectué de manière simultanée (en une seule passe) ou non. Les traitements peuvent s'effectuer sur le même signal, ou éventuellement sur des versions dérivées du même signal (par exemple selon des bandes passantes différentes). On distingue donc le "codage multiple" des "transcodages", où chaque codeur effectue la compression d'une version issue du décodage du signal compressé par le codeur précédent.The present invention is in the context of an optimization of " multiple coding " techniques, implemented as soon as a digital signal, or a portion of this signal, is coded according to several coding techniques. This multiple coding can be performed simultaneously (in one pass) or not. The processes can be carried out on the same signal, or possibly on versions derived from the same signal (for example according to different bandwidths). One thus distinguishes the " multiple coding " of the " transcodings ", where each coder carries out the compression of a version resulting from the decoding of the signal compressed by the preceding encoder.

Le codage multiple se présente par exemple dans le cas d'un même contenu qui est codé selon plusieurs formats et transmis ensuite à des terminaux ne supportant pas les mêmes formats de codage. S'il s'agit d'une diffusion en temps réel, le traitement devra être effectué en simultané. S'il s'agit d'accès à une base de données, les codages pourront être effectués l'un après l'autre, en différé. Dans ces exemples, le codage multiple permet de coder un même signal selon des formats différents en utilisant plusieurs codeurs (ou éventuellement plusieurs débits ou plusieurs modes d'un même codeur), chaque codeur fonctionnant de manière indépendante des autres codeurs.Multiple coding is, for example, in the case of the same content which is coded according to several formats and then transmitted to terminals that do not support the same coding formats. If it is a real-time broadcast, the processing should be done simultaneously. If it is a question of access to a database, the codings can be carried out one after another, delayed. In these examples, the multiple coding makes it possible to code the same signal in different formats by using several coders (or possibly several rates or several modes of the same encoder), each encoder operating independently of the other coders.

Un autre usage de codage multiple se rencontre dans des structures de codage où plusieurs codeurs se trouvent en compétition pour coder un segment de signal, un seul codeur étant finalement sélectionné pour le codage de ce segment. Le choix du codeur sélectionné peut s'effectuer à l'issue du traitement de ce segment, voire même ultérieurement (par décision retardée). Dans ce qui suit, on désignera par "codage multi-modes" ce type de structure (en référence à la sélection d'un "mode" de codage). Dans ces structures multi-modes, plusieurs codeurs partageant un "passé commun" sont amenés à coder la même portion de signal. Les techniques de codage utilisées peuvent être différente, ou issues d'une unique structure de codage. Elles ne seront cependant pas totalement indépendantes sauf s'il s'agit de techniques "sans mémoire". En effet, dans le cas (courant) de techniques de codage mettant en oeuvre des traitements récursifs, le traitement d'un segment donné de signal dépend de la manière dont ce signal a été codé dans le passé. Il y a donc une certaine dépendance entre les codeurs, dès lors qu'un codeur devra prendre en compte dans ses mémoires la sortie d'un autre codeur.Another use of multiple coding occurs in encoding structures where multiple encoders compete to encode a signal segment, with only one encoder ultimately selected for encoding that segment. The choice of the selected encoder can be made at the end of the processing of this segment, or even later (by delayed decision). In what follows, this type of structure will be referred to as " multi-mode coding " (with reference to the selection of a " mode " of coding). In these multi-mode structures, several coders sharing a " common past " are required to encode the same signal portion. The coding techniques used may be different, or from a single coding structure. However, they will not be completely independent unless they are " memory-free " techniques. Indeed, in the (current) case of coding techniques implementing recursive processing, the processing of a given signal segment depends on the way in which this signal has been coded in the past. There is therefore a certain dependence between the encoders, since an encoder will have to take into account in its memories the output of another encoder.

Dans ces différents contextes, la notion de "codage multiple" a été introduite ainsi que les conditions d'usage de telles techniques. Cependant la complexité de mise en oeuvre peut s'avérer rédhibitoire.In these different contexts, the notion of " multiple coding " has been introduced as well as the conditions of use of such techniques. However, the complexity of implementation can be prohibitive.

Par exemple, dans le cas de serveurs de contenus qui diffusent un même contenu sous plusieurs formats adaptés aux conditions d'accès, de réseaux et terminaux de différents clients, cette opération devient extrêmement complexe à mesure qu'augmente le nombre de formats désiré. S'il s'agit d'une diffusion temps réel, on se trouve rapidement limité par les ressources du système étant donné que les différents formats sont codés en parallèle.For example, in the case of content servers that broadcast the same content in several formats adapted to the access conditions, networks and terminals of different clients, this operation becomes extremely complex as the number of desired formats increases. If it is a real-time broadcast, one is quickly limited by the resources of the system since the different formats are coded in parallel.

Le deuxième cas d'usage mentionné concerne les applications de codage multi-modes, permettant la sélection d'un codeur parmi un ensemble pour chaque portion de signal analysé. La sélection demande la définition d'un critère, les plus courants visant à l'optimisation du compromis débit-distorsion. Le signal étant analysé sur des segments temporels successifs, à chaque segment plusieurs codages sont évalués. On sélectionne ensuite le codage de débit le plus faible pour une qualité donnée, ou celui de meilleure qualité pour un débit donné. On notera que d'autres contraintes que celles de débit/distorsion peuvent être utilisées.The second use case mentioned concerns multi-mode coding applications, allowing the selection of one encoder from a set for each portion of signal analyzed. The selection requires the definition of a criterion, the most common aiming at the optimization of the rate-distortion compromise. The signal being analyzed over successive time segments, at each segment several codings are evaluated. The lowest bit rate coding for a given quality is then selected, or the best bit rate coding for a given bit rate. It will be noted that other constraints than those of flow / distortion can be used.

En général, dans de telles structures, la sélection du codage s'effectue "a priori" par une analyse du signal sur le segment considéré (sélection selon les caractéristiques du signal). Cependant, la difficulté de produire une classification robuste du signal pour cette sélection a conduit à proposer une sélection "a posteriori" du mode optimal après codage de l'ensemble des modes, au prix toutefois d'une complexité élevée.In general, in such structures, the coding selection is made "a priori " by an analysis of the signal on the segment in question (selection according to the characteristics of the signal). However, the difficulty of producing a robust classification signal for this selection has led to propose a selection "a posteriori" the optimum mode after coding all the modes, however the complexity of high prices.

Des méthodes intermédiaires combinant les deux approches ont été proposées pour alléger le coût de calcul. Ces stratégies sont cependant sous-optimales et s'avèrent moins performantes que l'exploration de tous les modes. L'exploration de tous les modes ou d'une grande partie des modes constitue une application de codage multiple qui présente une complexité potentiellement élevée, difficilement compatible a priori avec le codage en temps réel par exemple.Intermediate methods combining the two approaches have been proposed to reduce the cost of calculation. These strategies, however, are suboptimal and are less efficient than exploring all modes. The exploration of all the modes or a large part of the modes constitutes a multiple coding application which presents a potentially high complexity, hardly compatible a priori with the real-time coding for example.

Actuellement, la plupart des opérations de codage multiple et de transcodage ne prennent pas en compte les interactions entre les formats et entre le format et son contenu. Quelques techniques de codage multi-modes ont été proposées, mais la décision du mode utilisé se fait généralement a priori soit sur le signal (par classification, comme par exemple le codeur SMV pour "Selectable Mode Vocoder"), soit en fonction des conditions du réseau (par exemple dans les codeurs AMR pour "Adaptive Multi-Rate"). Currently, most multi-coding and transcoding operations do not take into account the interactions between formats and between the format and its content. Some multi-mode coding techniques have been proposed, but the decision of the mode used is generally a priori on the signal (by classification, for example the SMV coder for " Selectable Mode Vocoder ") , depending on the network conditions (for example in AMR encoders for Adaptive Multi-Rate ) .

Dans les documents :

  • " An overview of variable rate speech coding for cellular networks", Gersho, A.; Paksoy, E.; Wireless Communications, 1992. Conference Proceedings, 1992 IEEE International Conference on Selected Topics, 25-26 Jun 1992 Page(s) : 172-175 ,
  • " A variable rate speech coding algorithm for cellular networks", Paksoy, E.; Gersho, A.; Speech Coding for Telecommunications, 1993. Proceedings, IEEE Workshop 1993, Page(s): 109-110 ,
  • " Variable rate speech coding for multiple access wireless networks", Paksoy E.; Gersho A.; Electrotechnical Conference, 1994, Proceedings, 7th Mediterranean, 12-14 Apr 1994 Page(s): 47 -50 vol.1 ,
plusieurs modes de sélection sont présentés, en particulier une décision contrôlée par la source et une décision contrôlée par le réseau.In the documents:
  • " An overview of variable rate speech coding for cellular networks ", Gersho, A., Paksoy, E. Wireless Communications, 1992. Conference Proceedings, 1992 IEEE International Conference on Selected Topics, 25-26 Jun 1992 Page (s): 172- 175 ,
  • " A variable rate speech coding algorithm for cellular networks ", Paksoy, E .; Gersho, A. Speech Coding for Telecommunications, 1993. Proceedings, IEEE Workshop 1993, Page (s): 109-110 ,
  • " Variable rate speech coding for multiple access wireless networks ", Paksoy E, Gersho A, Electrotechnical Conference, 1994, Proceedings, 7th Mediterranean, 12-14 Apr 1994 Page (s): 47 -50 vol.1 ,
several selection modes are presented, in particular a decision controlled by the source and a decision controlled by the network.

Dans le cas d'une décision contrôlée par la source, la décision a priori s'effectue à partir d'une classification du signal d'entrée. Il existe alors de nombreuses méthodes de classification du signal.In the case of a decision controlled by the source, the decision a priori is made from a classification of the input signal. There are many methods of signal classification.

Dans le cas d'une décision contrôlée par le réseau, il est plus simple de réaliser un codeur multi-modes dont le débit est choisi par un module externe plutôt que par la source. La méthode la plus simple consiste à élaborer une famille de codeurs chacun à débit fixe mais dont les débits sont différents entre codeurs et de commuter entre ces différents débits pour obtenir un mode courant désiré.In the case of a decision controlled by the network, it is simpler to produce a multi-mode encoder whose bit rate is chosen by an external module rather than by the source. The simplest method is to develop a family of coders each fixed rate but whose flow rates are different between coders and to switch between these different rates to obtain a desired current mode.

Quelques travaux ont aussi été présentés sur la possibilité de combiner plusieurs critères pour sélectionner a priori le mode qui doit être utilisé, notamment dans les documents :

  • " Variable-rate for the basic speech service in UMTS" Berruto, E.; Sereno, D.; Vehicular Technology Conference, 1993 IEEE 43rd, 18-20 May 1993 Page(s): 520-523
  • " A VR-CELP codec implementation for CDMA mobile communications" Cellario, L.; Sereno, D.; Giani, M.; Blocher, P.; Hellwig, K.; Acoustics, Speech, and Signal Processing, 1994, ICASSP-94, 1994 IEEE International Conference, Volume: 1 , 19-22 Apr 1994 Page(s): I/281 -I/284 vol.1 .
Some work was also presented on the possibility of combining several criteria to select a priori the mode to be used, particularly in documents:
  • " Variable-rate for basic speech service in UMTS "Berruto, E. Sereno, D. Vehicular Technology Conference, 1993 IEEE 43rd, 18-20 May 1993 Page (s): 520-523
  • " A VR-CELP Codec Implementation for CDMA Mobile Communications "Cellario, L., Sereno, D., Giani, M. Blocher, P., Hellwig, K., Acoustics, Speech, and Signal Processing, 1994, ICASSP-94, 1994 IEEE International Conference, Volume: 1, 19-22 Apr 1994 Page (s): I / 281 -I / 284 vol.1 .

Tous les algorithmes de codage muiti-modes avec sélection du mode de codage a priori souffrent d'un même inconvénient, en particulier lié à des problèmes de robustesse de la classification a priori.All the multi-mode coding algorithms with selection of the prior coding mode suffer from the same drawback, in particular related to problems of robustness of the classification a priori.

C'est pourquoi des techniques utilisant une décision a posteriori du mode de codage ont été proposées. Par exemple dans le document :

  • " Finite state CELP for variable rate speech coding" Vaseghi, S.V.; Acoustics, Speech, and Signal Processing, 1990, ICASSP-90, 1990 International Conference, 3-6 Apr 1990 Page(s) : 37 -40 vol.1 ,
  • le codeur peut commuter entre différents modes par optimisation d'une mesure de qualité objective, la décision se fait donc a posteriori en fonction des caractéristiques du signal d'entrée, du rapport visé débit/SQNR (pour "Signal to Quantization Noise Ratio") et de l'état courant du codeur. Un tel schéma de codage permet d'obtenir une amélioration de la qualité. Cependant, les différents codages étant réalisés en parallèle, la complexité résultante de ce type de système est prohibitive.
This is why techniques using a posteriori decision of the coding mode have been proposed. For example in the document:
  • " Finite state CELP for variable rate speech coding "Vaseghi, SV; Acoustics, Speech, and Signal Processing, 1990, ICASSP-90, 1990 International Conference, 3-6 Apr 1990 Page (s): 37 -40 vol.1 ,
  • the encoder can switch between different modes by optimizing an objective quality measurement, the decision is therefore a posteriori based on the characteristics of the input signal, the report referred flow / SQNR (for " Signal to Quantization Noise Ratio ") and the current state of the encoder. Such a coding scheme makes it possible to obtain an improvement in the quality. However, the different codings being made in parallel, the resulting complexity of this type of system is prohibitive.

D'autres techniques combinant une décision a priori et une amélioration en boucle fermée ont été proposées. Dans le document :

  • " Multimode variable bit rate speech coding: an efficient paradigm for high-quality low-rate representation of speech signal" Das, A.; DeJaco, A.; Manjunath, S.; Ananthapadmanabhan, A.; Huang, J.; Choy, E.; Acoustics, Speech, and Signal Processing, 1999. ICASSP '99 Proceedings, 1999 IEEE International Conference, Volume: 4, 15-19 Mar 1999 Page(s): 2307 -2310 vol.4 ,
le système proposé effectue une première sélection (sélection en boucle ouverte) du mode, en fonction des caractéristiques du signal. Cette décision peut être effectuée par classification. Ensuite, à partir d'une mesure d'erreur, si les performances du mode sélectionné ne sont pas satisfaisantes, un mode de débit plus élevé est appliqué et l'opération se répète (selon une décision recherchée en boucle fermée).Other techniques combining a priori decision and a closed-loop improvement have been proposed. In the document :
  • " Variable Multimode Bit Rate Speech Coding: an efficient paradigm for high-quality low-rate representation of speech signal "Das, A .; DeJaco, A. Manjunath, S; Ananthapadmanabhan, A; Huang, J .; Choy, E; Acoustics, Speech, and Signal Processing, 1999. ICASSP '99 Proceedings, 1999 IEEE International Conference, Volume: 4, 15-19 Mar 1999 Page (s): 2307 -2310 vol.4 ,
the proposed system makes a first selection (open-loop selection) of the mode, depending on the characteristics of the signal. This decision can be made by classification. Then, from an error measurement, if the performances of the selected mode are not satisfactory, a higher rate mode is applied and the operation is repeated (according to a decision sought in closed loop).

De même, dans les documents :

  • * " Variable rate speech coding for umts" Cellario, L.; Sereno, D.; Speech Coding for Telecommunications, 1993. Proceedings, IEEE Workshop, 1993 Page(s): 1 -2
  • " Phonetically-based vector excitation coding of speech at 3.6 kbps" Wang, S.; Gersho, A.; Acoustics, Speech, and Signal Processing, 1989. ICASSP-89., 1989 International Conference, 23-26 May 1989 Page(s): 49 -52 vol.1
  • * " A modified CS-ACELP algorithm for variable-rate speech coding robust in noisy environments" Beritelli, F.; IEEE Signal Processing Letters, Volume: 6 Issue: 2, Feb 1999 Page(s): 31 -34 ,
des techniques similaires ont été utilisées. Une première sélection en boucle ouverte est réalisée après classification du signal d'entrée (classification phonétique, ou voisé/non-voisé), ensuite une décision en boucle fermée est effectuée:
  • soit sur le codeur complet et, dans ce cas, tout le segment de parole est codé à nouveau;
  • soit sur une partie du codage, comme dans les références ci-avant précédées d'une étoile (*), pour lesquels le choix du dictionnaire à utiliser est effectué en boucle fermée.
Similarly, in the documents:
  • * " Variable rate speech coding for umts "Cellario, L. Sereno, D. Speech Coding for Telecommunications, 1993. Proceedings, IEEE Workshop, 1993 Page (s): 1 -2
  • " Phonetically-based vector excitation coding of speech at 3.6 kbps "Wang, S. Gersho, A. Acoustics, Speech, and Signal Processing, 1989. ICASSP-89., 1989 International Conference, 23-26 May 1989 Page (s) : 49 -52 vol.1
  • * " A modified CS-ACELP algorithm for robust variable-rate speech coding in noisy environments "Beritelli, F., IEEE Signal Processing Letters, Volume: 6 Issue: 2, Feb 1999 Page (s): 31 -34 ,
similar techniques were used. A first open-loop selection is performed after classification of the input signal (phonetic classification, or voiced / unvoiced), then a closed-loop decision is made:
  • either on the complete encoder and, in this case, the entire speech segment is coded again;
  • either on a part of the coding, as in the above references preceded by a star (*) , for which the choice of the dictionary to be used is done in a closed loop.

L'ensemble des études mentionnées ci-dessus tend à résoudre le problème de la complexité de la sélection optimale du mode par l'utilisation, totale ou partielle, d'une sélection ou pré-sélection a priori, qui évite le codage multiple ou diminue le nombre de codeurs à mettre en oeuvre en parallèle.All the studies mentioned above tend to solve the problem of the complexity of the optimal selection of the mode by the use, total or partial, of a selection or pre-selection a priori, which avoids the multiple coding or decreases the number of coders to be implemented in parallel.

US 6 581 032 divulgue un système de compression de la parole comportant quatre codecs activés sélectivement selon un signal à compresser et selon une classification des signaux. US 6,581,032 discloses a speech compression system comprising four codecs selectively activated according to a signal to be compressed and according to a classification of the signals.

US 6 141 638 divulgue un codeur utilisant différents dictionnaires de code selon des paramètres du signal à coder. US 6,141,638 discloses an encoder using different code dictionaries according to parameters of the signal to be encoded.

Toutefois, aucune technique de l'art antérieur permettant de réduire la complexité des codages réalisés en parallèle n'a été proposée.However, no technique of the prior art that makes it possible to reduce the complexity of the codings made in parallel has been proposed.

La présente invention vient améliorer la situation.The present invention improves the situation.

Elle propose à cet effet un procédé de codage multiple en compression tel que défini dans la revendication 1.For this purpose, it proposes a method of multiple coding in compression as defined in claim 1.

Dans une réalisation avantageuse, les étapes ci-avant sont mises en oeuvre par un produit programme d'ordinateur comportant des instructions de programme à cet effet. A ce titre, la présente invention vise aussi un tel produit programme d'ordinateur, destiné à être stocké dans une mémoire d'une unité de traitement, notamment d'un ordinateur ou d'un terminal mobile, ou sur un support mémoire amovible et destiné â coopérer avec un lecteur de l'unité de traitement.In an advantageous embodiment, the above steps are implemented by a computer program product comprising program instructions for this purpose. As such, the present invention also aims at such a computer program product, intended to be stored in a memory of a processing unit, in particular a computer or a mobile terminal, or on a removable memory medium and intended to cooperate with a drive of the processing unit.

La présente invention vise aussi un dispositif d'aide à un codage en compression, pour la mise en oeuvre du procédé selon l'invention, et comportant alors une mémoire propre à stocker des instructions d'un produit programme d'ordinateur du type précité.The present invention also aims at a device for aiding compression coding, for the implementation of the method according to the invention, and then including a memory adapted to store instructions of a computer program product of the aforementioned type.

D'autres caractéristiques et avantages de l'invention apparaîtront à l'examen de la description détaillée ci-après, et des dessins annexés sur lesquels:

  • la figure 1a illustre schématiquement le contexte d'application de la présente invention, avec une pluralité de codeurs mis en parallèle,
  • la figure 1b illustre schématiquement l'application de l'invention, avec la mise en partage de blocs fonctionnels entre plusieurs codeurs mis en parallèle,
  • la figure 1c illustre schématiquement l'application de l'invention, avec la mise en partage de blocs fonctionnels en codage multi-modes,
  • la figure 1d illustre schématiquement l'application de l'invention, en codage multi-modes en treillis,
  • la figure 2 représente schématiquement les blocs fonctionnels principaux d'un codeur fréquentiel perceptuel,
  • la figure 3 représente schématiquement les blocs fonctionnels principaux d'un codeur à analyse par synthèse.
  • la figure 4a représente schématiquement les blocs fonctionnels principaux d'un codeur TDAC,
  • la figure 4b représente schématiquement le format du flux binaire codé par le codeur de la figure 4a,
  • la figure 5 représente schématiquement l'application de l'invention à une pluralité de codeurs TDAC en parallèle, selon une réalisation avantageuse,
  • la figure 6a représente schématiquement les blocs fonctionnels principaux d'un codeur MPEG-1 (layer I est II),
  • la figure 6b représente schématiquement le format du flux binaire codé par le codeur de la figure 6a,
  • la figure 7 représente schématiquement l'application de l'invention à une pluralité de codeurs MPEG-1 (layer I et II) mis en parallèle, selon une réalisation avantageuse,
  • et la figure 8 représente plus en détails les blocs fonctionnels d'un codeur à analyse par synthèse, ici de type NB-AMR selon la norme 3GPP.
Other features and advantages of the invention will appear on examining the detailed description below, and the attached drawings in which:
  • the figure 1a schematically illustrates the context of application of the present invention, with a plurality of coders in parallel,
  • the figure 1b schematically illustrates the application of the invention, with the sharing of functional blocks between several coders in parallel,
  • the figure 1c schematically illustrates the application of the invention, with the sharing of functional blocks in multi-mode coding,
  • the figure 1d schematically illustrates the application of the invention, in trellis multi-mode coding,
  • the figure 2 schematically represents the main functional blocks of a perceptual frequency coder,
  • the figure 3 schematically represents the main functional blocks of a synthesis analysis coder.
  • the figure 4a schematically represents the main functional blocks of a TDAC coder,
  • the figure 4b schematically represents the format of the bitstream coded by the coder of the figure 4a ,
  • the figure 5 schematically represents the application of the invention to a plurality of TDAC coders in parallel, according to an advantageous embodiment,
  • the figure 6a schematically represents the main functional blocks of an MPEG-1 coder (layer I and II),
  • the figure 6b schematically represents the format of the bitstream coded by the coder of the figure 6a ,
  • the figure 7 schematically represents the application of the invention to a plurality of MPEG-1 coders (layer I and II) connected in parallel, according to an advantageous embodiment,
  • and the figure 8 shows in more detail the functional blocks of a synthesis analysis coder, here of type NB-AMR according to the 3GPP standard.

On se réfère tout d'abord à la figure 1a sur laquelle on a représenté une pluralité de codeurs C0, C1, ..., CN, en parallèle et recevant chacun un signal d'entrée s0. Chaque codeur comporte des blocs fonctionnels BF1 à BFn pour mettre en oeuvre des étapes successives de codage et délivrer finalement un flux binaire codé BS0, BS1, ..., BSN. On indique en outre que dans une application en codage multi-modes, les sorties des codeurs C0 à CN sont reliées à un module MM de sélection du mode optimal et le flux binaire BS du codeur optimal est transmis (flèches en traits pointillés de la figure 1a).We first refer to the figure 1a on which there is shown a plurality of encoders C0, C1, ..., CN, in parallel and each receiving an input signal s 0 . Each encoder comprises functional blocks BF1 to BFn for implementing successive coding steps and ultimately outputting a coded bitstream BS0, BS1, ..., BSN. It is furthermore indicated that in an application in multi-mode coding, the outputs of the coders C0 to CN are connected to an optimal mode selection module MM and the bit stream BS of the optimal coder is transmitted (dotted line arrows of the figure 1a ).

Pour une raison de simplicité, tous les codeurs de l'exemple de la figure 1a ont le même nombre de blocs fonctionnels, mais bien entendu tous ces blocs fonctionnels ne sont pas nécessairement prévus dans tous les codeurs, en pratique.For the sake of simplicity, all the coders in the example of the figure 1a have the same number of functional blocks, but of course all these functional blocks are not necessarily provided in all the coders, in practice.

Certains blocs fonctionnels BFi sont parfois identiques d'un mode (ou d'un codeur) à l'autre, d'autres ne diffèrent qu'au niveau des quantificateurs. Des relations exploitables existent aussi lorsque l'on utilise des codeurs issus d'une même famille de codage, utilisant des modèles similaires ou calculant des paramètres liés physiquement au signal.Some BFi function blocks are sometimes identical from one mode (or encoder) to another, while others differ only in quantizer level. Usable relationships also exist when encoders from the same coding family are used, using similar models or computing parameters physically related to the signal.

Ce sont ces relations que la présente invention propose d'exploiter, afin de réduire la complexité des opérations de codage multiple.It is these relationships that the present invention proposes to exploit, in order to reduce the complexity of multiple coding operations.

Dans un premier temps, l'invention propose d'identifier les blocs fonctionnels composant chacun des codeurs. On exploite ensuite les similarités techniques entre les codeurs en considérant les blocs fonctionnels dont les fonctions sont équivalentes ou voisines. Pour chacun de ces blocs, l'invention propose :

  • d'une part de définir des opérations dites "communes", et de les effectuer une seule fois pour l'ensemble des codeurs;
  • d'autre part, de mettre en oeuvre des méthodes de calcul spécifiques à chaque codeur et utilisant notamment les résultats de ces calculs communs. Ces méthodes de calcul produisent un résultat éventuellement différent de celui produit par un codage complet. L'objectif est alors en fait d'accélérer les traitements en exploitant les informations disponibles et fournies notamment par les calculs communs. De tels procédés permettant d'accélérer les calculs sont par exemple mis en oeuvre dans des techniques destinées à réduire la complexité des opérations de transcodage (dites techniques de "transcodage intelligent").
In a first step, the invention proposes to identify the functional blocks that make up each of the coders. The technical similarities between the coders are then exploited by considering the functional blocks whose functions are equivalent or similar. For each of these blocks, the invention proposes:
  • on the one hand to define so-called " common " operations, and to perform them only once for all the coders;
  • on the other hand, to implement calculation methods specific to each encoder and using in particular the results of these common calculations. These methods of calculation produce a result possibly different from that produced by a complete coding. The objective is then to accelerate the processing by exploiting the information available and provided in particular by the common calculations. Such methods making it possible to accelerate the calculations are for example used in techniques designed to reduce the complexity of transcoding operations (so-called " intelligent transcoding " techniques) .

La figure 1b illustre la solution proposée. Dans l'exemple représenté, les opérations "communes" précitées sont effectuées une seule fois pour une partie au moins des codeurs et, préférentiellement, pour l'ensemble des codeurs, dans un module indépendant MI qui redistribuera les résultats obtenus à une partie au moins des codeurs, ou préférentiellement à tous ces codeurs. II s'agit ainsi d'une mise en partage entre une partie au moins de tous les codeurs C0 à CN (ou "mutualisation" ci-après) des résultats obtenus. Un tel module indépendant MI peut faire partie d'un dispositif d'aide à un codage multiple en compression tel que défini ci-avant.The figure 1b illustrates the proposed solution. In the example shown, the aforementioned " common " operations are performed once for at least a portion of the coders and, preferably, for all the coders, in an independent module MI which will redistribute the results. obtained at least part of the coders, or preferably all these coders. It is thus a sharing between at least part of all coders C0 to CN (or " pooling " below) the results obtained. Such an independent module MI may be part of a device for a multiple compression coding as defined above.

Dans une variante avantageuse, plutôt que d'avoir recours à un module de calcul externe MI, on utilise le ou les blocs fonctionnels existants BF1 à BFn d'un même ou de plusieurs codeurs distincts, ce ou ces codeurs étant choisis selon des critères qui seront décrits plus loin.In an advantageous variant, rather than having recourse to an external calculation module MI, the existing functional block or blocks BF1 to BFn of the same or more different coders is used, this or these coders being chosen according to criteria which will be described later.

La présente invention peut mettre en oeuvre plusieurs stratégies qui, bien entendu, peuvent différer selon le rôle du bloc fonctionnel considéré.The present invention can implement several strategies which, of course, may differ depending on the role of the functional block considered.

Une première stratégie consiste à utiliser les paramètres du codeur dont le débit est le plus faible pour focaliser la recherche des paramètres pour tous les autres modes.A first strategy is to use the parameters of the encoder whose bit rate is the lowest to focus the search parameters for all other modes.

A l'inverse, une deuxième stratégie consiste à utiliser les paramètres du codeur dont le débit est le plus élevé, puis de "dégrader" progressivement jusqu'au codeur dont le débit est le plus faible.Conversely, a second strategy is to use the parameters of the encoder whose rate is the highest, then to " degrade " progressively to the encoder whose bit rate is the lowest.

Bien entendu, si l'on souhaite privilégier un codeur particulier, il est possible de coder un segment de signal en utilisant ce codeur, puis, en appliquant les deux stratégies ci-avant, d'atteindre les codeurs de débit supérieur et inférieur.Of course, if it is desired to favor a particular encoder, it is possible to encode a signal segment using this encoder, then, by applying the two strategies above, to reach the upper and lower rate encoders.

Bien entendu, d'autres critères que le débit peuvent être utilisés pour piloter la recherche. On peut par exemple, pour certains blocs fonctionnels, favoriser le codeur dont les paramètres se prêtent le mieux à une extraction (ou une analyse) et/ou à un codage efficaces des paramètres similaires des autres codeurs, l'efficacité pouvant être jugée selon la complexité, la qualité ou un compromis des deux.Of course, other criteria than flow can be used to drive the search. For some functional blocks, for example, it is possible to favor the encoder whose parameters lend themselves best to efficient extraction (or analysis) and / or coding of similar parameters from the others. coders, the efficiency being judged according to the complexity, the quality or a compromise of the two.

II peut être prévu aussi de créer un module de codage indépendant, non présent dans les codeurs, mais permettant un codage plus efficace des paramètres du bloc fonctionnel considéré, pour l'ensemble des codeurs.It may also be envisaged to create an independent coding module, not present in the coders, but allowing a more efficient coding of the parameters of the functional block considered, for all the coders.

Ces diverses stratégies de mise en oeuvre sont particulièrement intéressantes dans le cas du codage multi-modes. Dans ce contexte illustré à la figure 1c, la présente invention permet de réduire la complexité des calculs préliminaires à la sélection a posteriori d'un codeur effectuée en dernière étape, par exemple par le dernier module MM avant la transmission du flux binaire BS.These various implementation strategies are particularly interesting in the case of multi-mode coding. In this context illustrated in figure 1c the present invention makes it possible to reduce the complexity of the calculations preliminary to the a posteriori selection of an encoder performed in the last step, for example by the last module MM before the transmission of the bit stream BS.

Dans ce cas particulier du codage multi-modes, une variante de la présente invention, représentée dans l'exemple de la figure 1c, propose d'introduire un module de sélection partielle MSPi (avec i = 1, 2,..., N) après chaque étape de codage (donc après les blocs fonctionnels BFi1 à BFiN1 mis en compétition et dont le résultat du ou des blocs sélectionnés BFicc va être utilisé par la suite). Ainsi, les similitudes entre les différents modes sont exploitées pour accélérer le calcul de chaque bloc fonctionnel. Tous les schémas de codage ne seront alors pas obligatoirement évalués.In this particular case of multi-mode coding, a variant of the present invention, represented in the example of the figure 1c proposes to introduce a partial selection module MSPi (with i = 1, 2, ..., N) after each coding step (thus after the functional blocks BFi1 to BFiN 1 put in competition and whose result of the selected blocks BFicc will be used later). Thus, the similarities between the different modes are exploited to speed up the calculation of each functional block. All coding schemes will not necessarily be evaluated.

Une variante plus sophistiquée de la structure multi-modes reposant sur la découpe en blocs fonctionnels décrite ci-avant est maintenant proposée, en référence à la figure 1d. La structure multi-modes de la figure 1d est dite "en treillis", avec plusieurs chemins possibles dans le treillis. En fait, sur la figure 1d, on a représenté tous les chemins possibles du treillis de sorte qu'il se présente sous une forme arborescente. On indique en particulier que chaque chemin du treillis est défini par une combinaison de modes de fonctionnement des blocs fonctionnels, chaque bloc fonctionnel alimentant plusieurs variantes possibles du bloc fonctionnel suivant.A more sophisticated variant of the multi-mode structure based on the functional block cutting described above is now proposed, with reference to the figure 1d . The multi-mode structure of the figure 1d is called "trellis", with several possible paths in the trellis. In fact, on the figure 1d all the possible paths of the lattice are represented so that it is in a tree form. In particular, it is indicated that each path of the trellis is defined by a combination of operating modes of the functional blocks, each functional block supplying several possible variants of the next functional block.

Ainsi, chaque mode de codage est issu de la combinaison de modes de fonctionnement des blocs fonctionnels : le bloc fonctionnel 1 possède N1 modes de fonctionnement, le bloc fonctionnel 2 en possède N2, et ainsi de suite jusqu'au bloc P. L'ensemble des NN= N1 x N2 x ... x Np combinaisons possibles est donc représenté par un treillis de NN branches décrivant, bout-à-bout, un codeur multi-modes complet à NN modes. On peut éventuellement supprimer a priori certaines branches du treillis et définir ainsi une arborescence comportant un nombre réduit de branches. Une première particularité de cette structure est qu'elle prévoit, pour un bloc fonctionnel donné, un module de calculs communs par sortie du bloc fonctionnel précédent. Ces modules de calculs communs effectuent les mêmes opérations, mais sur la base de signaux différents puisqu'ils sont issus de blocs antérieurs différents. Avantageusement, les modules de calculs communs d'un même niveau sont mutualisés : les résultats d'un module donné exploitables par les modules suivants sont fournis à ces modules suivants. D'autre part, une sélection partielle, effectuée à l'issue du traitement de chaque bloc fonctionnel, permet avantageusement de supprimer les branches les moins performantes selon le critère choisi. On peut donc réduire le nombre de branches du treillis à évaluer.Thus, each coding mode is derived from the combination of operating modes of the functional blocks: the functional block 1 has N 1 operating modes, the functional block 2 has N 2 , and so on up to the block P. L The set of NN = N 1 x N 2 x ... x Np possible combinations is therefore represented by a lattice of NN branches describing, end-to-end, a complete multi-mode encoder with NN modes. Some branches of the lattice may be removed a priori and thus define a tree with a reduced number of branches. A first feature of this structure is that it provides, for a given functional block, a common calculation module per output of the previous functional block. These common calculation modules perform the same operations, but on the basis of different signals since they come from different previous blocks. Advantageously, the common calculation modules of the same level are pooled: the results of a given module usable by the following modules are provided to these following modules. On the other hand, a partial selection, made at the end of the processing of each functional block, advantageously makes it possible to eliminate the less efficient branches according to the chosen criterion. It is therefore possible to reduce the number of branches of the trellis to be evaluated.

Une application avantageuse de cette structure multi-modes en treillis est la suivante.
Lorsque les blocs fonctionnels sont susceptibles d'opérer à des débits respectifs différents et en utilisant des paramètres respectifs propres auxdits débits, pour un bloc fonctionnel donné, le chemin du treillis choisi est celui traversant le bloc fonctionnel de débit le plus faible, ou encore le bloc fonctionnel de débit le plus élevé selon le contexte de codage, et les résultats obtenus du bloc fonctionnel de débit le plus faible (ou le plus élevé) sont adaptés aux débits d'une partie au moins des autres blocs fonctionnels par une recherche focalisée de paramètres pour une partie au moins de tous les autres blocs fonctionnels, jusqu'au bloc fonctionnel de débit le plus élevé (ou respectivement le plus faible).
An advantageous application of this multi-mode lattice structure is as follows.
When the functional blocks are capable of operating at different respective flow rates and using respective parameters specific to said flow rates, for a given functional block, the chosen trellis path is the one passing through the lowest flow functional block, or the highest rate functional block according to the coding context, and the results obtained from the lowest (or highest) bit rate functional block are adapted to the bit rates of at least a portion of the other functional blocks by a focused search of parameters for at least part of all other functional blocks, up to the highest (or lowest) rate functional block.

En variante, on choisit un bloc fonctionnel de débit donné et on adapte progressivement au moins une partie des paramètres propres à ce bloc fonctionnel:

  • jusqu'au bloc fonctionnel capable d'opérer au débit le plus faible, par recherche focalisée, et
  • jusqu'au bloc fonctionnel capable d'opérer au débit le plus élevé, par recherche focalisée.
    De manière générale, on réduit ainsi la complexité associée au codage multiple.
As a variant, a given flow function block is chosen and at least a portion of the parameters specific to this functional block are progressively adapted:
  • to the functional block capable of operating at the lowest rate, by focused search, and
  • to the functional block capable of operating at the highest rate, by focussed search.
    In general, this reduces the complexity associated with multiple coding.

L'invention s'applique à tout schéma de compression mettant en oeuvre le codage multiple d'un contenu multimédia. Trois exemples de réalisation sont présentés dans ce qui suit, dans le domaine de la compression audio (parole et son). Les deux premiers exemples de réalisation se situent dans le contexte de la famille des codeurs par transformée, dont on peut donner le document suivant à titre de référence :

  • " Perceptual Coding of Digital Audio", Painter, T.; Spanias, A.; Proceedings of the IEEE, Vol. 88, No 4, April 2000 .
The invention applies to any compression scheme implementing the multiple encoding of a multimedia content. Three embodiments are presented in the following, in the field of audio compression (speech and sound). The first two exemplary embodiments are in the context of the family of transform coders, the following document of which can be given for reference:
  • " Perceptual Coding of Digital Audio ", Painter, T. Spanias, A. Proceedings of the IEEE, Vol 88, No 4, April 2000 .

Le troisième exemple de réalisation se situe dans le contexte des codeurs CELP, dont on peut donner le document suivant à titre de référence :
" Code Excited Linear Prediction (CELP): High quality speech at very low bit rates" Schroeder M.R.; Atal B.S.; Acoustics, Speech, and Signal Processing, 1985. Proceedings. 1985 IEEE International Conference, Page(s): 937-940 .
The third exemplary embodiment is in the context of the CELP coders, the following document of which can be given for reference:
" Code Excited Linear Prediction (CELP): Schroeder MR; Atal BS; Acoustics, Speech, and Signal Processing, 1985. Proceedings. 1985 IEEE International Conference, Page (s): 937-940 .

Un rappel des principales caractéristiques de ces deux familles de codage est tout d'abord présenté dans ce qui suit.A reminder of the main characteristics of these two families of coding is first presented in the following.

* Les codeurs par transformée ou en sous bandes* Transformers or subband coders

II s'agit de codeurs en compression par transformée ou en sous bandes basés sur des critères psychoacoustiques. Ce type de codeur procède par transformation sur des blocs du signal temporel pour obtenir un ensemble de coefficients. Les transformations sont du type temps-fréquence, l'une des transformations les plus utilisées étant la transformée en cosinus discrète modifiée (dit "MDCT", de l'anglais "Modified Discrete Cosine Transform"). Avant la quantification de ces coefficients, un algorithme procède à l'allocation des bits de façon à ce que le bruit de quantification soit le moins audible possible. L'allocation binaire et la quantification des coefficients mettent en oeuvre une courbe de masquage, obtenue à l'aide d'un modèle psychoacoustique permettant d'évaluer, pour chaque raie spectrale considérée, un seuil de masquage représentatif de l'amplitude nécessaire pour qu'un son à cette fréquence soit audible. La figure 2 donne le schéma de principe d'un codeur fréquentiel. On remarquera que la structure sous forme de blocs fonctionnels est bien représentée. En se référant à la figure 2, les blocs fonctionnels principaux sont :

  • un bloc 21 de transformation temps/fréquence du signal audionumérique d'entrée s0,
  • un bloc 22 de détermination d'un modèle perceptuel à partir du signal transformé,
  • un bloc 23 de quantification et codage, à partir du modèle perceptuel,
  • et un bloc 24 de formatage du flux binaire pour obtenir une trame audio codée stc.
These are transform or subband compressors based on psychoacoustic criteria. This type of encoder proceeds by transforming blocks of the time signal to obtain a set of coefficients. The transformations are of the time-frequency type, one of the most used transformations being the modified discrete cosine transform ( "MDCT") of the " Modified Discrete Cosine Transform ". Before the quantization of these coefficients, an algorithm proceeds to the allocation of the bits so that the quantization noise is the least audible possible. The binary allocation and the quantification of the coefficients implements a masking curve, obtained by means of a psychoacoustic model making it possible to evaluate, for each spectral line considered, a masking threshold representative of the amplitude necessary for qu a sound at this frequency is audible. The figure 2 gives the schematic diagram of a frequency encoder. It will be noted that the structure in the form of functional blocks is well represented. Referring to the figure 2 , the main functional blocks are:
  • a time / frequency transformation block 21 of the input digital audio signal s 0 ,
  • a block 22 for determining a perceptual model from the transformed signal,
  • a quantization and coding block 23, from the perceptual model,
  • and a block 24 for formatting the bitstream to obtain an encoded audio frame s tc .

* Les codeurs à analyse par synthèse (codage CELP)* Synthetic analysis coders (CELP coding)

Dans les codeurs à analyse par synthèse, le modèle de synthèse du signal reconstruit est utilisé au codeur pour extraire les paramètres modélisant les signaux à coder. Ces signaux peuvent être échantillonnés à la fréquence de 8 kHz (bande téléphonique 300-3400 Hz) ou à une fréquence plus élevée, par exemple à 16 kHz pour le codage en bande élargie (bande passante de 50Hz à 7 kHz). Selon l'application et la qualité désirée, le taux de compression varie de 1 à 16. Ces codeurs fonctionnent à des débits de 2 à 16 kbit/s en bande téléphonique, et à des débits de 6 à 32 kbit/s en bande élargie. Le dispositif de codage numérique de type CELP, codeur à analyse par synthèse le plus utilité actuellement, est présenté à la figure 3 sous forme de blocs fonctionnels principaux. Le signal de parole s0 est échantillonné et converti en une suite de trames d'un nombre L d'échantillons. Chaque trame est synthétisée en filtrant une forme d'onde extraite d'un répertoire (appelé "dictionnaire"), multipliée par un gain, à travers deux filtres variant dans le temps. Le dictionnaire d'excitation fixe est un ensemble fini de formes d'ondes des L échantillons. Le premier filtre est un filtre de prédiction à long terme. Une analyse "LTP" (pour "Long Term Prediction") permet d'évaluer les paramètres de ce prédicteur à long terme qui exploite la périodicité des sons voisés, cette composante harmonique étant modélisée sous la forme d'un dictionnaire adaptatif (bloc 32). Le second filtre est un filtre de prédiction à court terme. Les méthodes d'analyse "LPC" (pour "Linear Prediction Coding") permettent d'obtenir ces paramètres de prédiction à court terme, représentatifs de la fonction de transfert du conduit vocal et caractéristiques de l'enveloppe du spectre du signal. Le procédé utilisé pour déterminer la séquence d'innovation est la méthode d'analyse par synthèse qui se résume comme suit. Au codeur, un grand nombre de séquences d'innovation du dictionnaire d'excitation fixes sont filtrées par le filtre LPC (filtre de synthèse du bloc fonctionnel 34 de la figure 3). Au préalable, l'excitation adaptative a été obtenue de façon similaire. La forme d'onde sélectionnée est celle produisant le signal synthétique le plus proche du signal original (minimisation de l'erreur au niveau du bloc fonctionnel 35), selon un critère de pondération perceptuelle (bloc fonctionnel 36) qui est connu en général sous le nom de critère "CELP".In the synthesis analysis coders, the reconstructed signal synthesis model is used at the encoder to extract the parameters modeling the signals to be coded. These signals can be sampled at the frequency of 8 kHz (telephone band 300-3400 Hz) or at a higher frequency, for example at 16 kHz for wideband coding (bandwidth 50 Hz to 7 kHz). Depending on the application and the desired quality, the compression ratio varies from 1 to 16. These encoders operate at rates of 2 to 16 kbit / s in the telephone band, and at speeds of 6 to 32 kbit / s in the extended band. . The CELP type digital coding device, currently the most useful synthesis analysis coder, is presented to the figure 3 in the form of main functional blocks. The speech signal s 0 is sampled and converted into a sequence of frames of a number L of samples. Each frame is synthesized by filtering a waveform extracted from a directory (called " dictionary "), multiplied by a gain, through two filters varying in time. The fixed excitation dictionary is a finite set of waveforms of the L samples. The first filter is a long-term prediction filter. A " LTP " ( Long Term Prediction ) analysis makes it possible to evaluate the parameters of this long-term predictor which exploits the periodicity of the voiced sounds, this harmonic component being modeled in the form of an adaptive dictionary (block 32) . The second filter is a short-term prediction filter. The " LPC " ( Linear Prediction Coding ) analysis methods make it possible to obtain these short-term prediction parameters, which are representative of the vocal tract transfer function and characteristics of the envelope of the signal spectrum. The method used to determine the innovation sequence is the synthetic analysis method which is summarized as follows. At the encoder, a large number of innovation sequences of the fixed excitation dictionary are filtered by the LPC filter (synthesis filter of the function block 34 of the figure 3 ). Beforehand, the adaptive excitation was obtained in a similar way. The selected waveform is that producing the synthetic signal closest to the original signal (error minimization at function block 35), according to a perceptual weighting criterion (function block 36) which is generally known as criterion name " CELP ".

Dans le schéma de principe du codeur CELP donné à la figure 3, l'extraction de la fréquence fondamentale des sons voisés (ou "pitch"), appliquée sur le signal résultant de l'analyse LPC du bloc 31, permet ensuite d'en extraire la corrélation à long terme au niveau du bloc 32, appelée composante harmonique ou excitation adaptative (E.A.). Le signal résiduel est enfin modélisé classiquement par quelques impulsions, dont l'ensemble des positions est prédéfini dans un répertoire, appelé répertoire d'excitation fixe (E.F) dans le bloc 33.In the block diagram of the CELP coder given to the figure 3 , the extraction of the fundamental frequency of the voiced sounds (or " pitch ") , applied to the signal resulting from the LPC analysis of the block 31, then makes it possible to extract the long-term correlation at the level of the block 32, called harmonic component or adaptive excitation (EA). The residual signal is finally modeled conventionally by a few pulses, all of the positions of which are predefined in a directory called fixed excitation directory (EF) in block 33.

Le décodage est, quant à lui, beaucoup moins complexe que le codage. Le flux binaire généré par le codeur permet au décodeur, après démultiplexage, d'obtenir l'index de quantification de chaque paramètre. Le décodage des paramètres et l'application du modèle de synthèse permettent alors de reconstruire le signal.Decoding is, for its part, much less complex than coding. The bitstream generated by the coder enables the decoder, after demultiplexing, to obtain the quantization index of each parameter. The decoding of the parameters and the application of the synthesis model then make it possible to reconstruct the signal.

On décrit ci-après les trois exemples de réalisation précités, en commençant tout d'abord par un codeur par transformée du type représenté sur la figure 2.The three aforementioned embodiments are described below, starting first with a transform coder of the type shown in FIG. figure 2 .

* Premier exemple de réalisation : application à un codeur "TDAC"* First example of implementation: application to a coder " TDAC "

Le premier exemple de réalisation concerne le codeur fréquentiel perceptuel dit "TDAC" et décrit notamment dans le document publié US-2001/027393 . Ce codeur TDAC est utilisé pour coder des signaux audio numériques échantillonnés à 16 kHz (bande élargie). La figure 4a illustre les blocs fonctionnels principaux de ce codeur. Un signal audio x(n) limité en bande à 7 kHz et échantillonné à 16 kHz est découpé en trames de 320 échantillons (20 ms). Une transformée en cosinus discrète modifiée (ou "MDCT") est appliquée (bloc fonctionnel 41) sur des trames du signal d'entrée de 640 échantillons avec un recouvrement de 50 %, donc avec un rafraîchissement de l'analyse MDCT toutes les 20 ms. On limite le spectre à 7225 Hz en mettant à zéro les 31 derniers coefficients (seuls les 289 premiers coefficients sont différents de 0). Une courbe de masquage (bloc 42) est déterminée à partir de ce spectre et tous les coefficients masqués sont mis à zéro. Le spectre est divisé en 32 bandes de largeurs inégales. Les éventuelles bandes masquées sont déterminées en fonction des coefficients transformés des signaux. Pour chaque bande du spectre, l'énergie des coefficients MDCT est calculée (pour obtenir des facteurs d'échelle). Les 32 facteurs d'échelle constituent l'enveloppe spectrale du signal qui est ensuite quantifiée puis codée par un codage entropique (bloc fonctionnel 43), et enfin transmise dans la trame codée sc.The first embodiment relates to the perceptual frequency coder called " TDAC " and described in particular in the published document. US 2001 / 027,393 . This TDAC encoder is used to encode digital audio signals sampled at 16 kHz (wide band). The figure 4a illustrates the main functional blocks of this encoder. An audio signal x (n) limited in band at 7 kHz and sampled at 16 kHz is cut into frames of 320 samples (20 ms). A Modified Discrete Cosine Transform (or " MDCT ") is applied (function block 41) on 640 sample input signal frames with 50% overlap, thus with a refresh of MDCT analysis every 20 ms . The spectrum is limited to 7225 Hz by setting the last 31 coefficients to zero (only the first 289 coefficients are different of 0). A masking curve (block 42) is determined from this spectrum and all masked coefficients are set to zero. The spectrum is divided into 32 bands of unequal widths. Any masked bands are determined according to the transformed coefficients of the signals. For each band of the spectrum, the energy of the MDCT coefficients is calculated (to obtain scale factors). The 32 scale factors constitute the spectral envelope of the signal which is then quantized and coded by entropy encoding (function block 43), and finally transmitted in the coded frame s c .

L'allocation dynamique des bits (bloc fonctionnel 44) se base sur une courbe de masquage par bande (bloc fonctionnel 42) calculée à partir de la version décodée et déquantifiée de l'enveloppe spectrale. Cette mesure permet d'avoir une compatibilité entre l'allocation binaire du codeur et du décodeur. Les coefficients MDCT normalisés dans chaque bande sont ensuite quantifiés (bloc fonctionnel 45) par des quantificateurs vectoriels utilisant des dictionnaires imbriqués en taille, les dictionnaires étant composés d'une union de codes à permutation de type II. Finalement, en se référant à la figure 4b, les informations sur la tonalité (codées ici sur un bit B1) et le voisement (codées ici sur un bit B0), ainsi que l'enveloppe spectrale eq(i) et les coefficients codés yq(j) sont multiplexés (bloc 46 de la figure 4a) et transmis en trames.The dynamic allocation of the bits (functional block 44) is based on a band masking curve (functional block 42) calculated from the decoded and dequantized version of the spectral envelope. This measurement makes it possible to have compatibility between the bit allocation of the encoder and the decoder. The normalized MDCT coefficients in each band are then quantized (function block 45) by vector quantizers using size-nested dictionaries, the dictionaries being composed of a type II permutation code union. Finally, referring to the figure 4b , the information on the tone (coded here on a bit B 1 ) and the voicing (coded here on a bit B 0 ), as well as the spectral envelope eq (i) and the coefficients coded y q (j) are multiplexed ( block 46 of the figure 4a ) and transmitted in frames.

Ce codeur pouvant fonctionner à plusieurs débits, on se propose de réaliser un codeur multi-débits par exemple à 16, 24 et 32 kbit/s. Dans ce schéma de codage, les blocs fonctionnels suivants peuvent être mis en commun entre les différents modes:

  • Transformée MDCT (bloc 41),
  • Détection de voisement (bloc fonctionnel 47 de la figure 4a) et détection de tonalité (bloc fonctionnel 48 de la figure 4a),
  • Calcul, quantification et codage entropique de l'enveloppe spectrale (bloc 43),
  • Calcul d'une courbe de masquage, coefficient par coefficient, et d'une courbe de masquage par bande (bloc 42).
Since this encoder can operate at several rates, it is proposed to make a multi-rate encoder for example at 16, 24 and 32 kbit / s. In this coding scheme, the following functional blocks can be shared between the different modes:
  • MDCT Transform (block 41),
  • Voicing detection (functional block 47 of the figure 4a ) and tone detection (function block 48 of the figure 4a )
  • Calculation, quantification and entropic coding of the spectral envelope (block 43),
  • Calculation of a masking curve, coefficient by coefficient, and a band masking curve (block 42).

Ces différents blocs constituent 61,5% de la complexité du traitement dans le processus de codage. Leur factorisation est donc d'un intérêt important pour réduire cette complexité lors de la génération de plusieurs flux binaires correspondants à des débits différents.These different blocks make up 61.5% of the processing complexity in the coding process. Their factorization is therefore of great interest to reduce this complexity when generating several bit streams corresponding to different rates.

Les résultats de ces blocs fonctionnels permettent déjà d'obtenir une première partie commune à tous les flux binaires de sortie qui contient les bits d'information sur le voisement, la tonalité et l'enveloppe spectrale codée.The results of these functional blocks already make it possible to obtain a first portion common to all the output bit streams which contains the information bits on the voicing, the tone and the coded spectral envelope.

Dans une première variante de cet exemple de réalisation, il est possible de réaliser les opérations d'allocation des bits et de quantification pour chacun des flux binaires de sortie correspondant à chacun des débits binaires considérés. Ces deux opérations sont effectuées exactement de la même manière qu'habituellement dans un codeur TDAC.In a first variant of this exemplary embodiment, it is possible to carry out the bit allocation and quantization operations for each of the output bit streams corresponding to each of the bit rates considered. These two operations are performed in exactly the same way as usual in a TDAC encoder.

Dans une seconde variante plus avancée telle qu'illustrée sur la figure 5, on peut mettre en oeuvre des techniques de transcodage "intelligent" (comme décrit dans le document publié US-2001/027393 cité ci-avant) pour réduire davantage la complexité et mutualiser certaines opérations, notamment :

  • l'allocation de bits (bloc fonctionnel 44),
  • et aussi la quantification des coefficients (blocs fonctionnels 45_i), comme on le verra ci-après.
In a second more advanced variant as illustrated on the figure 5 , we can implement "intelligent " transcoding techniques (as described in the published document US 2001 / 027,393 cited above) to further reduce complexity and pool certain operations, including:
  • bit allocation (function block 44),
  • and also the quantization of the coefficients (functional blocks 45_i), as will be seen below.

Sur la figure 5, les blocs fonctionnels mis en partage entre les codeurs (ou "mutualisés") portent la même référence que ceux d'un seul codeur TDAC tel que représenté sur la figure 4a. II s'agit des blocs 41, 42, 47, 48, 43 et 44. En particulier, le bloc 44 d'allocation des bits est utilisé en plusieurs passes, et le nombre de bits alloués est ajusté pour la transquantification qu'effectue chaque codeur (blocs 45_1, ... , 45_(K-2), 45_(K-1)), comme on le verra ci-après. On remarque en outre que ces transquantifications utilisent les résultats obtenus par le bloc fonctionnel 45_0 de quantification pour un codeur choisi, d'indice 0 (le codeur de débit le plus faible dans l'exemple décrit). Finalement, les seuls blocs fonctionnels des codeurs qui agissent sans interaction réelle sont les blocs de multiplexage 46_0, 46_1,..., 46_(K-2), 46_(K-1), bien qu'ils utilisent tous les mêmes informations de voisement et de tonalité, ainsi que la même enveloppe spectrale codée. A ce titre, on indique simplement qu'une mutualisation partielle du multiplexage peut être menée, là encore.On the figure 5 , the functional blocks shared by the coders (or " pooled ") bear the same reference as those of a single TDAC coder as represented in FIG. figure 4a . These are blocks 41, 42, 47, 48, 43 and 44. In in particular, the bit allocation block 44 is used in several passes, and the number of bits allocated is adjusted for the transquantification performed by each coder (blocks 45_1,..., 45_ (K-2), 45_ (K -1)), as will be seen below. Note further that these transquantifications use the results obtained by the quantization function block 45_0 for a chosen encoder, index 0 (the lowest rate encoder in the example described). Finally, the only functional blocks of the encoders which act without real interaction are the multiplexing blocks 46_0, 46_1, ..., 46_ (K-2), 46_ (K-1), although they all use the same information of voicing and tone, as well as the same coded spectral envelope. As such, it is simply stated that a partial pooling of the multiplexing can be conducted, again.

Pour les deux blocs fonctionnels d'allocation de bits et de quantification, la stratégie employée consiste à exploiter les résultats des deux blocs fonctionnels d'allocation des bits et de quantification réalisés pour le flux binaire (0), au débit le plus bas D 0, pour accélérer les opérations des deux blocs fonctionnels correspondants pour les K-1 autres flux binaire (k) (1 ≤ k < K). On peut aussi considérer le schéma de codage multi-débits qui utilise un bloc fonctionnel d'allocation de bits par flux binaire (sans factorisation prévue pour ce bloc) mais mutualise une partie des opérations de quantification ensuite.For the two bit allocation and quantization functional blocks, the strategy employed is to exploit the results of the two bit allocation and quantization functional blocks made for the bit stream (0), at the lowest bit rate D 0 , to speed up the operations of the two corresponding function blocks for the K-1 other bitstream (k) (1 ≤ k < K ). It is also possible to consider the multi-rate coding scheme which uses a bit-allocation functional block per bit stream (without factorization provided for this block) but mutualizes a part of the quantization operations thereafter.

Les techniques de codage multiple présentées ci-après se basent avantageusement sur un transcodage intelligent utilisé pour la réduction du débit binaire de flux audio codé, généralement situé dans un noeud du réseau.The multiple coding techniques presented below are advantageously based on intelligent transcoding used for the reduction of the coded audio stream bit rate, generally located in a node of the network.

Dans la suite, les flux binaires k , 0 ≤ k < K, sont classés suivant un ordre croissant de débits (D0 < D1 <... < DK-1 ). Ainsi, le flux binaire 0 correspond au débit binaire le plus bas.In the following, the bit streams k , 0 ≤ k < K , are classified in an increasing order of rates ( D 0 < D 1 <... < D K-1 ) . Thus, bit stream 0 corresponds to the lowest bit rate.

* Allocation de bits * Bit allocation

L'allocation de bits dans le codeur TDAC se réalise en deux phases. D'abord un premier calcul du nombre de bits à allouer à chaque bande est effectué de préférence suivant la formule suivante : b opt i = 1 2 log 2 e q 2 i S b j + C , 0 i M - 1 ,

Figure imgb0001
C = B M - 1 2 M l = 0 M - 1 log 2 e q 2 l / S b l
Figure imgb0002
est une constante,
B est le nombre total de bits disponibles,
M est le nombre de bandes,
eq (i) est la valeur décodée et déquantifiée de l'enveloppe spectrale sur la bande i,
et Sb (i) est le seuil de masquage pour cette bande.Bit allocation in the TDAC encoder is done in two phases. Firstly, a first calculation of the number of bits to be allocated to each band is preferably carried out according to the following formula: b Opt i = 1 2 log 2 e q 2 i S b j + VS , 0 i M - 1 ,
Figure imgb0001
or VS = B M - 1 2 M Σ l = 0 M - 1 log 2 e q 2 l / S b l
Figure imgb0002
is a constant,
B is the total number of bits available,
M is the number of bands,
e q ( i ) is the decoded and dequantized value of the spectral envelope on the band i ,
and S b ( i ) is the masking threshold for this band.

Chacune des valeurs obtenues est arrondie à l'entier naturel le plus proche. Si le débit total alloué n'est pas exactement égal à celui disponible, une seconde phase est utilisée pour réaliser le réajustement. Cette étape se fait préférentiellement par une succession d'opérations itératives basées sur un critère perceptuel qui ajoute ou retire des bits aux bandes.Each of the values obtained is rounded to the nearest natural integer. If the total flow allocated is not exactly equal to that available, a second phase is used to perform the readjustment. This step is preferably done by a succession of iterative operations based on a perceptual criterion that adds or removes bits from the bands.

Ainsi, si le nombre total de bits distribués est inférieur à celui disponible, l'ajout de bits se fait aux bandes où l'amélioration perceptuelle est la plus importante. Cette amélioration perceptuelle est mesurée par la variation du rapport bruit à masque entre l'allocation initiale et finale des bandes. Le débit est augmenté pour la bande où cette variation est la plus grande. Dans le cas contraire où le nombre total de bits distribués est supérieur à celui disponible, l'extraction de bits sur les bandes se fait de manière duale à cette dernière procédure.Thus, if the total number of bits distributed is less than that available, the bits are added to the bands where the perceptual improvement is the most important. This perceptual improvement is measured by the variation of the noise to mask ratio between the initial and final allocation of the bands. The rate is increased for the band where this variation is greatest. In the opposite case where the total number of distributed bits is greater than that available, the extraction of bits on the bands is dual to the latter procedure.

Dans le schéma de codage multi-débits correspondant au codeur TDAC, il est possible de factoriser certaines opérations pour l'allocation de bits. Ainsi, la première phase de détermination par la formule ci-avant peut se faire une seule fois en se basant sur le débit binaire D 0 le plus bas. La phase de réajustement en ajoutant des bits peut se faire ensuite de manière continue. Une fois que le nombre total de bit distribué atteint le nombre correspondant à un débit binaire d'un flux binaire k, k=1,2...,K-1, la distribution courante est considérée comme celle qui est utilisée pour la quantification des vecteurs de coefficients normalisés par bande de ce flux binaire.In the multi-rate coding scheme corresponding to the TDAC coder, it is possible to factorize certain operations for bit allocation. So, the first determination step by the above formula can be done once based on the lowest bit rate D 0 . The readjustment phase by adding bits can then be done continuously. Once the total number of bits distributed reaches the number corresponding to a bit rate of a bit stream k, k = 1.2 ..., K-1, the current distribution is considered as that which is used for the quantization band-normalized coefficient vectors of this bit stream.

* Quantification des coefficients * Quantification of the coefficients

Pour ce qui concerne la quantification des coefficients, le codeur TDAC utilise une quantification vectorielle utilisant des dictionnaires imbriqués en taille, les dictionnaires étant composés d'une union de codes à permutation de type II. Ce type de quantification s'applique sur chacun des vecteurs des coefficients MDCT sur une bande. Un tel vecteur est normalisé au préalable en utilisant la valeur déquantifiée de l'enveloppe spectrale sur cette bande. On note :

  • C(bi,di ) le dictionnaire correspondant au nombre de bits bi et à la dimension di ,
  • N(bi,di ) le nombre d'éléments dans ce dictionnaire,
  • CL(bi,di ) l'ensemble de ses leaders, et
  • NL(bi, di ) le nombre de leaders.
For quantization of coefficients, the TDAC encoder uses vector quantization using size-nested dictionaries, the dictionaries being composed of a type II permutation code union. This type of quantization applies to each of the vectors of the MDCT coefficients on a band. Such a vector is normalized beforehand by using the dequantized value of the spectral envelope on this band. We notice :
  • C ( b i , d i ) the dictionary corresponding to the number of bits b i and to the dimension d i ,
  • N ( b i , d i ) the number of elements in this dictionary,
  • CL ( b i , d i ) all of its leaders, and
  • NL ( b i , d i ) the number of leaders.

Le résultat de quantification pour chaque bande i de la trame est un mot de code mi transmis dans le flux binaire. II représente l'index du vecteur quantifié dans le dictionnaire et calculé à partir des informations suivantes :

  • le numéro Li , dans l'ensemble CL(bi,di ) des leaders du dictionnaire C(bi,di ), du vecteur leader quantifié q(i) plus proche voisin d'un leader courant (i),
  • le rang ri de Yq (i) dans la classe du leader q (i),
  • et la combinaison de signes signq (i) à appliquer à Yq (i) (ou à q (i)), où l'on précise les notations suivantes :
  • Y(i) est le vecteur des valeurs absolues des coefficients normalisés de la bande i,
  • sign(i) est le vecteur des signes des coefficients normalisés de la bande i,
  • (i) est le vecteur leader du vecteur Y(i) précité, obtenu par ordonnancement décroissant de ses composantes (la permutation correspondante est notée perm(i)),
  • et Yq (i) est le vecteur quantifié de Y(i) (ou "le plus proche voisin" de Y(i) dans le dictionnaire C(bi,di )).
The quantization result for each band i of the frame is a code word m i transmitted in the bit stream. It represents the index of the quantized vector in the dictionary and calculated from the following information:
  • the number L i , in the set CL ( b i , d i ) of the leaders of the dictionary C ( b i , d i ), of the quantized leader vector q ( i ) nearest neighbor of a current leader ( i )
  • the rank r i of Y q ( i ) in the class of the leader q ( i ),
  • and the combination of signs sign q ( i ) to be applied to Y q ( i ) (or q ( i )) , where the following notations are specified:
  • Y ( i ) is the vector of the absolute values of the normalized coefficients of the band i,
  • sign ( i ) is the vector of the signs of the normalized coefficients of the band i,
  • ( i ) is the vector leader of the vector Y ( i ) above, obtained by descending ordering of its components (the corresponding permutation is denoted perm ( i )),
  • and Y q ( i ) is the quantized vector of Y ( i ) (or " nearest neighbor " of Y ( i ) in the dictionary C ( b i , d i )).

Dans la suite, la notation α(k), avec un exposant k, indique le paramètre utilisé dans le traitement effectué pour obtenir le flux binaire du codeur k. Les paramètres sans cet exposant étant calculés une seule fois pour toutes pour le flux binaire 0. Ils sont indépendants du débit (ou du mode) considéré.In the following, the notation α ( k ) , with an exponent k , indicates the parameter used in the processing performed to obtain the bitstream of the encoder k. The parameters without this exponent being calculated once and for all for the bit stream 0. They are independent of the flow (or mode) considered.

La propriété "d'imbrication" des dictionnaires précitée s'exprime selon la relation : C b i 0 d i C b i k - 1 d i C b i k d i C b i K - 1 d i

Figure imgb0003

avec aussi: CL b i 0 d i CL b i k - 1 d i CL b i k d i CL b i K - 1 d i
Figure imgb0004
The " nesting " property of the aforementioned dictionaries is expressed according to the relation: VS b i 0 d i ... VS b i k - 1 d i VS b i k d i ... VS b i K - 1 d i
Figure imgb0003

with also: CL b i 0 d i ... CL b i k - 1 d i CL b i k d i ... CL b i K - 1 d i
Figure imgb0004

On note CL b i k d i \ CL b i k - 1 d i

Figure imgb0005
le complémentaire de CL b i k - 1 d i
Figure imgb0006
dans CL b i k d i .
Figure imgb0007
Son cardinal est égal à NL b i k d i - NL b i k - 1 d i .
Figure imgb0008
We notice CL b i k d i \ CL b i k - 1 d i
Figure imgb0005
the complementary of CL b i k - 1 d i
Figure imgb0006
in CL b i k d i .
Figure imgb0007
His cardinal is equal to NL b i k d i - NL b i k - 1 d i .
Figure imgb0008

L'obtention des mots de code m i k

Figure imgb0009
(avec 0 ≤ k < K), résultats de la quantification du vecteur des coefficients de la bande i pour chacun des flux binaires k, se fait comme suit.

  • Pour le flux binaire k = 0, l'opération de quantification se fait de manière classique comme habituellement dans le codeur TDAC. Elle permet d'obtenir les paramètres, signq (0)(i), L i 0
    Figure imgb0010
    et ri (0) qui permettent de construire le mot de code m i 0 .
    Figure imgb0011
    On détermine d'ailleurs dans cette même étape les vecteurs (i) et sign(i). Ils sont stockés en mémoire, ainsi que la permutation correspondante perm(i), pour être utilisés, le cas échéant, dans les étapes suivantes relatives aux autres flux binaires.
  • Pour les flux binaires 1 ≤ k < K, on procède de manière incrémentale, de k = 1 jusqu'à k = K -1, en utilisant préférentiellement les étapes suivantes :
    • Si b i k = b i k - 1
      Figure imgb0012
      alors :
      1. 1. le mot de code, sur la bande i, de la trame du flux binaire k est le même que celui de la trame du flux binaire (k-1) : et m i k = m i k - 1
        Figure imgb0013
    • Sinon, i.e. b i k > b i k - 1
      Figure imgb0014
      :
      • 2. On recherche parmi les NL b i k d i - NL b i k - 1 d i
        Figure imgb0015
        leaders de CL b i k d i \ CL b i k - 1 d i
        Figure imgb0016
        le plus proche voisin de (i),
      • 3. Avec le résultat de l'étape 2 et connaissant le plus proche voisin de (i) dans CL b i k - 1 d i ,
        Figure imgb0017
        on teste si le plus proche voisin de (i) dans CL b i k d i
        Figure imgb0018
        est dans CL b i k - 1 d i
        Figure imgb0019
        (cas "Flag=0" ci-après) ou CL b i k d i \ CL b i k - 1 d i
        Figure imgb0020
        (cas "Flag=1" ci-après),
      • 4. Si Flag=0 ((le leader le plus proche de (i) dans CL b i k - 1 d i
        Figure imgb0021
        est aussi son plus proche voisin dans CL b i k d i )
        Figure imgb0022
        alors : m i k = m i k - 1
        Figure imgb0023

        Si Flag=1 (le leader le plus proche de (i) dans CL b i k d i \ CL b i k - 1 d i
        Figure imgb0024
        trouvé à l'étape 2 est aussi son plus proche voisin dans CL b i k d i ) ,
        Figure imgb0025
        soit L i k
        Figure imgb0026
        son numéro (avec L i k NL b i k - 1 d i ) ,
        Figure imgb0027
        alors on effectue les étapes ci-après :
        1. a. Recherche du rang ri (k) de Yq (k)(i) (nouveau vecteur quantifié de Y(i) dans la classe du leader q (k)(i)) par exemple par l'algorithme de Schalkwijk en utilisant perm(i),
        2. b. Détermination de sign q k i
          Figure imgb0028
          en utilisant sign(i) et perm(i),
        3. c. Détermination du mot de code m i k .
          Figure imgb0029
          à partir de L i k ,
          Figure imgb0030
          ri (k) et sign q k i .
          Figure imgb0031
Obtaining code words m i k
Figure imgb0009
(with 0 ≤ k < K ), results of the quantization of the vector of the coefficients of the band i for each of the bit streams k, is as follows.
  • For the bit stream k = 0, the quantization operation is done conventionally as usual in the TDAC coder. It allows to obtain the parameters, sign q (0) ( i ), The i 0
    Figure imgb0010
    and r i (0) which make it possible to construct the code word m i 0 .
    Figure imgb0011
    Moreover, the vectors ( i ) and sign ( i ) are determined in this same step. They are stored in memory, as well as the perm permutation perm ( i ), to be used, if necessary, in the following steps relating to other bitstreams.
  • For the bit streams 1 ≤ k < K , we proceed incrementally, from k = 1 to k = K -1, preferably using the following steps:
    • Yes b i k = b i k - 1
      Figure imgb0012
      so :
      1. 1. the code word, on the band i, of the frame of the bit stream k is the same as that of the frame of the bit stream ( k -1): and m i k = m i k - 1
        Figure imgb0013
    • Otherwise, ie b i k > b i k - 1
      Figure imgb0014
      :
      • 2. We are looking for NL b i k d i - NL b i k - 1 d i
        Figure imgb0015
        leaders of CL b i k d i \ CL b i k - 1 d i
        Figure imgb0016
        the nearest neighbor of ( i ),
      • 3. With the result of step 2 and knowing the nearest neighbor of ( i ) in CL b i k - 1 d i ,
        Figure imgb0017
        we test whether the nearest neighbor of ( i ) in CL b i k d i
        Figure imgb0018
        is in CL b i k - 1 d i
        Figure imgb0019
        ("Flag = 0" case below) or CL b i k d i \ CL b i k - 1 d i
        Figure imgb0020
        ("Flag = 1" case below),
      • 4. If Flag = 0 ((the nearest leader of ( i ) in CL b i k - 1 d i
        Figure imgb0021
        is also his closest neighbor in CL b i k d i )
        Figure imgb0022
        so : m i k = m i k - 1
        Figure imgb0023

        If Flag = 1 (the closest leader to ( i ) in CL b i k d i \ CL b i k - 1 d i
        Figure imgb0024
        found in step 2 is also his closest neighbor in CL b i k d i ) ,
        Figure imgb0025
        is The i k
        Figure imgb0026
        his number (with The i k NL b i k - 1 d i ) ,
        Figure imgb0027
        then we carry out the following steps:
        1. at. Search for the rank r i ( k ) of Y q ( k ) ( i ) (new quantized vector of Y ( i ) in the class of the leader q ( k ) ( i )) for example by the Schalkwijk algorithm using perm ( i ),
        2. b. Determination of sign q k i
          Figure imgb0028
          using sign ( i ) and perm (i),
        3. vs. Determining the codeword m i k .
          Figure imgb0029
          from The i k ,
          Figure imgb0030
          r i ( k ) and sign q k i .
          Figure imgb0031

* Deuxième exemple de réalisation : application à un codeur par transformée de type MPEG-1 Layer I&IISecond Example: Application to an MPEG-1 Layer I & II Transform Encoder

Le codeur MPEG-1 Layer I&II, présenté à la figure 6a, utilise un banc de filtres à 32 sous-bandes uniformes (bloc 61 de la figure 6a) pour réaliser la transformation temps/fréquence du signal audio d'entrée s0. Les échantillons de sortie de chaque sous-bande sont regroupés, puis normalisés par un facteur d'échelle commun (déterminé par le bloc fonctionnel 67) avant d'être quantifiés (bloc 62). Le nombre de niveaux du quantificateur scalaire uniforme utilisé pour chaque sous-bande résulte d'une procédure d'allocation dynamique des bits (réalisée par le bloc 63). Cette procédure utilise un modèle psychoacoustique (bloc 64) pour déterminer la répartition des bits qui rend le bruit de quantification le moins perceptible possible. Les modèles d'audition proposés dans la norme se basent sur l'estimation du spectre obtenu par une transformée de Fourier rapide (FFT) du signal temporel d'entrée (réalisée par le bloc 65). En se référant à la figure 6b, la trame sc, multiplexée par le bloc 66 de la figure 6a et qui est finalement transmise, contient, après un champ d'entête HD, l'ensemble des échantillons des sous-bandes quantifiés ESB, qui représentent l'information principale, et une information complémentaire utilisée pour l'opération de décodage constituée par les facteurs d'échelle FE et l'allocation de bits Ai.The MPEG-1 Layer I & II coder presented to the figure 6a , uses a filter bank with 32 uniform sub-bands (block 61 of the figure 6a ) to perform the time / frequency transformation of the input audio signal s 0. The output samples of each subband are grouped and then normalized by a common scale factor (determined by the function block 67) before being quantified (block 62). The number of levels of the uniform scalar quantizer used for each subband results from a dynamic bit allocation procedure (performed by block 63). This procedure uses a psychoacoustic model (block 64) to determine the bit distribution that makes the quantization noise as noticeable as possible. The hearing models proposed in the standard are based on the estimation of the spectrum obtained by a fast Fourier transform (FFT) of the input temporal signal (made by block 65). Referring to the figure 6b , the frame s c , multiplexed by the block 66 of the figure 6a and which is finally transmitted, contains, after a header field H D , the set of samples of the quantized sub-bands E SB , which represent the main information, and complementary information used for the decoding operation constituted by the scaling factors F E and the bit allocation A i .

A partir de ce schéma de codage, la construction d'un codeur multi-débits, dans une application de l'invention, peut être réalisée en mettant en commun les blocs fonctionnels suivants, en se référant à la figure 7 :

  • Banc de filtres d'analyse 61
  • Détermination des facteurs d'échelle 67
  • Calcul 65 de la transformée de Fourier FFT
  • Détermination des seuils de masquage suivant un modèle psychoacoustique 64.
From this coding scheme, the construction of a multi-rate encoder, in one application of the invention, can be achieved by pooling the following functional blocks, with reference to the figure 7 :
  • Bench of Analysis Filters 61
  • Determination of scale factors 67
  • Calculation 65 of the FFT Fourier Transform
  • Determination of masking thresholds according to a psychoacoustic model 64.

Les deux blocs 64 et 65 fournissent déjà les rapports signal à masque (flèches SMR des figures 6a et 7), utilisés pour la procédure d'allocation de bits (bloc 70 de la figure 7).The two blocks 64 and 65 already provide the signal to mask ratios (SMR arrows Figures 6a and 7 ), used for the bit allocation procedure (block 70 of the figure 7 ).

Dans cet exemple de réalisation tel que représenté sur la figure 7, il est possible de tirer profit de la procédure utilisée pour l'allocation de bits pour la mettre aussi en commun, mais en ajoutant toutefois quelques modifications à l'allocation (bloc 70 d'allocation des bits de la figure 7). Seul le bloc fonctionnel de quantification 62_0 à 62_(K-1) est donc spécifique à chaque flux binaire correspondant à un débit Dk , 0 ≤ kK -1. Il en va de même pour le bloc de multiplexage 66_0 à 66_(K-1).In this embodiment, as shown in the figure 7 , it is possible to take advantage of the procedure used for the allocation of bits to put it also in common, but adding however some modifications to the allocation (block 70 of allocation of the bits of the figure 7 ). Only the quantization function block 62_0 to 62_ (K-1) is therefore specific to each bit stream corresponding to a rate D k , 0 ≤ kK - 1 . The same is true for the multiplexing block 66_0 to 66_ (K-1).

* Allocation des bits * Bit allocation

Dans le codeur MPEG-1 Layer I&II, l'allocation se fait préférentiellement par une succession d'étapes itératives comme suit.

  • Etape 0 : Initialisation à zéro du nombres de bits bi de chacune des sous-bandes i, 0 ≤ i < M.
  • Etape 1 : Mise à jour de la fonction de distorsion NMR(i) (appelée "rapport bruit à masque", de l'anglais "Noise to Mask Ratio") sur chacune des sous-bandes : NMR i = SMR i - SNR b i ,
    Figure imgb0032

    SNR(bi ) est le rapport signal à bruit correspondant au quantificateur ayant un nombre de bits bi ,
    et SMR(i) le rapport signal à masque fourni par le modèle psychoacoustique.
  • Etape 2 : Incrémentation du nombre de bits b i 0 de la sous-bande i 0 où cette distorsion est maximale: b i 0 = b i 0 + ε , i 0 = arg max i NMR i
    Figure imgb0033

    où ε est une valeur entière positive dépendant de la bande, en général prise égale à 1.
In the MPEG-1 Layer I & II coder, the allocation is preferentially done by a succession of iterative steps as follows.
  • Step 0: Zero initialization of the number of bits b i of each of the sub-bands i, 0 ≤ i < M.
  • Step 1: Update the distortion function NMR (i) (referred to as "noise ratio mask" English "Noise to Mask Ratio") on each of the sub-bands: NMR i = SMR i - SNR b i ,
    Figure imgb0032

    where SNR ( b i ) is the signal-to-noise ratio corresponding to the quantizer having a number of bits b i ,
    and SMR ( i ) the signal to mask ratio provided by the psychoacoustic model.
  • Step 2: Incrementing the number of bits b i 0 of the sub-band i 0 where this distortion is maximum: b i 0 = b i 0 + ε , i 0 = arg max i NMR i
    Figure imgb0033

    where ε is a positive integer value dependent on the band, usually taken equal to 1.

Les étapes 1 et 2 sont répétées de manière itérative jusqu'à ce que le nombre total de bits disponibles, correspondant au débit de fonctionnement, soit distribué. Le résultat est alors un vecteur de distribution de bits (b 0 ,b 1 ,...,b M-1). Steps 1 and 2 are repeated iteratively until the total number of available bits, corresponding to the operating rate, is distributed. The result is then a bit distribution vector ( b 0 , b 1 , ..., b M -1 ) .

Dans le schéma de codage multi-débits, ces étapes sont mises en commun avec quelques autres modifications, notamment :

  • le bloc fonctionnel ayant pour sortie K vecteurs de distributions de bits b 0 k b 1 k b M - 1 k
    Figure imgb0034
    (avec 0 ≤ kK -1), un vecteur b 0 k b 1 k b M - 1 k
    Figure imgb0035
    est obtenu lorsque le nombre total de bits disponibles correspondant au débit binaire Dk du flux binaire k est distribué, à l'itération des étapes 1 et 2.
  • L'arrêt de l'itération des étapes 1 et 2 se fait lorsque le nombre total de bits disponibles correspondant au débit binaire le plus élevé D K-1 est totalement distribué (on rappelle que les flux binaires sont ordonnés suivant un ordre croissant de débits).
In the multi-rate coding scheme, these steps are pooled along with a few other changes, including:
  • the function block having for output K bit distribution vectors b 0 k b 1 k ... b M - 1 k
    Figure imgb0034
    (with 0 ≤ kK -1), a vector b 0 k b 1 k ... b M - 1 k
    Figure imgb0035
    is obtained when the total number of available bits corresponding to the bit rate D k of the bit stream k is distributed, at the iteration of steps 1 and 2.
  • Stopping of the iteration of steps 1 and 2 is done when the total number of available bits corresponding to the highest bit rate D K -1 is totally distributed (it is recalled that the bit streams are ordered in an increasing order of bit rates ).

On notera que les vecteurs de distribution de bits sont obtenus successivement à partir de k = 0 jusqu'à k = K-1. Les K sorties de ce bloc d'allocation de bits alimentent alors les blocs de quantification pour chacun des flux binaires au débit donné.It will be noted that the bit distribution vectors are successively obtained from k = 0 up to k = K -1 . The K outputs of this bit allocation block then feed the quantization blocks for each of the bit streams at the given bit rate.

* Troisième exemple de réalisation : application à un codeur de type CELPThird example embodiment: application to a CELP type encoder

Le dernier exemple de réalisation concerne le codage de la parole multi-modes à décision a posteriori à partir du codeur 3GPP NB-AMR (pour "Narrow-Band Adaptive Multi-Rate") qui est un codeur de parole en bande téléphonique multi-débits adaptatif, selon une norme 3GPP. Ce codeur qui appartient à la famille bien connue des codeurs CELP dont le principe a été décrit brièvement plus haut, comporte huit modes (ou débits) allant de 12,2 kbit/s à 4,75 kbit/s, tous basés sur la technique ACELP (pour "Algebraic Code Excited Linear Prediction"). La figure 8 donne le schéma de codage en blocs fonctionnels de ce codeur. Cette structure a été exploitée afin de réaliser un codeur multi-modes à décision a posteriori, basé sur 4 modes du codeur NB-AMR (7,4; 6,7; 5,9; 5,15).The last exemplary embodiment relates to the coding of the multi-mode speech with a posteriori decision from the 3GPP NB-AMR (" Narrow-Band Adaptive Multi-Rate ") coder which is a multi-rate bandband speech encoder. adaptive, according to a 3GPP standard. This encoder, which belongs to the well-known family of CELP coders whose principle was briefly described above, has eight modes (or bit rates) ranging from 12.2 kbit / s to 4.75 kbit / s, all based on the technique ACELP (for " Algebraic Code Excited Linear Prediction "). The figure 8 gives the coding scheme in functional blocks of this encoder. This structure was exploited in order to realize a post-decision multi-mode encoder, based on 4 modes of the NB-AMR encoder (7.4, 6.7, 5.9, 5.15).

Dans une première variante, seule la mutualisation des blocs fonctionnels identiques est exploitée (les résultats des quatre codages sont alors identiques à ceux des quatre codages en parallèle).In a first variant, only the sharing of the identical functional blocks is exploited (the results of the four codings are then identical to those of the four parallel codings).

Dans une deuxième variante, la complexité est encore plus réduite. Les calculs de blocs fonctionnels non identiques pour certains modes sont accélérés en exploitant ceux d'un autre mode ou d'un module de traitement commun, comme on le verra ci-après. Les résultats des quatre codages ainsi mutualisés sont alors différents de ceux des quatre codages en parallèle.In a second variant, the complexity is even smaller. The non-identical functional block calculations for some modes are accelerated by exploiting those of another mode or a common processing module, as will be seen below. The results of the four encodings thus shared are then different from those of the four codings in parallel.

Dans une autre variante encore, les blocs fonctionnels de ces quatre modes sont utilisés pour un codage multi-modes en treillis, comme on l'a vu ci-avant en référence à la figure 1d.In yet another variant, the functional blocks of these four modes are used for trellis multi-mode coding, as has been seen above with reference to FIG. figure 1d .

On rappelle brièvement ci-après les quatre modes (7,4; 6,7; 5,9; 5,15) du codeur 3GPP NB-AMR.The four modes (7.4, 6.7, 5.9, 5.15) of the 3GPP NB-AMR encoder are briefly described below.

Le codeur 3GPP NB-AMR travaille sur un signal de parole limité en bande à 3,4 kHz et échantillonné à 8 kHz découpé en trames de 20 ms (160 échantillons). Chaque trame comporte 4 sous-trames de 5 ms (40 échantillons) regroupées 2 par 2 dans des "super sous-trames" de 10 ms (80 échantillons). Pour tous les modes, les mêmes types de paramètres sont extraits du signal mais avec des variantes de modélisation et/ou de quantification de ces paramètres. Dans le codeur NB-AMR, cinq types de paramètres sont à analyser et à coder. Les paramètres LSP (pour "Line Spectral Pairs") sont traités une fois par trame pour tous les modes, sauf pour le mode 12,2 (donc une fois par super sous-trame). Les autres paramètres (notamment le retard LTP, le gain de l'excitation adaptative, l'excitation fixe, le gain de l'excitation fixe) sont traités une fois par sous-trame.The 3GPP NB-AMR coder is working on a 3.4 kHz band-limited speech signal sampled at 8 kHz cut into 20 ms frames (160 samples). Each frame has 4 subframes of 5 ms (40 samples) grouped 2 by 2 in " super subframes " of 10 ms (80 samples). For all modes, the same types of parameters are extracted from the signal but with variants of modeling and / or quantification of these parameters. In the NB-AMR encoder, five types of parameters are to be analyzed and coded. Line Spectral Pairs (LSP) parameters are processed once per frame for all modes, except for 12.2 mode (so once per super subframe). The other parameters (in particular the LTP delay, the gain of the adaptive excitation, the fixed excitation, the gain of the fixed excitation) are processed once per subframe.

Les quatre modes considérés ici (7,4; 6,7; 5,9; 5,15) se distinguent essentiellement par les quantifications de leurs paramètres. L'allocation binaire de ces 4 modes est résumée dans le tableau 1 ci-après. Tableau 1: Allocation binaire des 4 modes (7,4; 6,7; 5,9; 5,15) du codeur 3GPP NB-AMR Mode (kbit/s) 7,4 6,7 5,9 5,15 LSP 26=(8+9+9) 26=(8+9+9) 26=(8+9+9) 23=(8+8+7) Retards LTP 8 / 5 / 8 / 5 8 / 4 / 8 /4 8 / 4 / 8 / 4 8 / 4 / 4 / 4 Excitation fixe 17/17/ 17/17 14/14/ 14/14 11/11/ 11/11 9 / 9 / 9 / 9 Gains des excitations fixe et adaptative 7 / 7 / / 7 / 7 7- / 7 / 7 / 7 6 / 6 / 6 / 6 6 / 6 / 6 / 6 Total par trame 148 134 118 103 The four modes considered here (7.4, 6.7, 5.9, 5.15) are distinguished essentially by the quantifications of their parameters. The binary allocation of these 4 modes is summarized in Table 1 below. <b> Table 1: </ b> Bit allocation of the 4 modes (7.4, 6.7, 5.9, 5.15) of the 3GPP NB-AMR encoder Mode (kbit / s) 7.4 6.7 5.9 5.15 LSP 26 = (8 + 9 + 9) 26 = (8 + 9 + 9) 26 = (8 + 9 + 9) 23 = (8 + 8 + 7) LTP delays 8/5/8/5 8/4/8/4 8/4/8/4 8/4/4/4 Fixed excitation 17/17/17/17 14/14 / 14/14 11/11/11/11 9/9/9/9 Gains of fixed and adaptive excitations 7/7 / / 7/7 7- / 7/7/7 6/6/6/6 6/6/6/6 Total per frame 148 134 118 103

Ces 4 modes du codeur NB-AMR (7,4; 6,7; 5,9; 5,15) possèdent des modules identiques comme par exemple le pré-traitement, l'analyse des coefficients de prédiction linéaire, le calcul de signal pondéré. Le pré-traitement du signal est un filtrage passe-haut de fréquence de coupure 80 Hz pour supprimer les composantes continues combiné à une division par deux des signaux d'entrée pour éviter des débordements. L'analyse LPC comprend des sous-modules de fenêtrage, de calcul des autocorrélations, de mise en oeuvre de l'algorithme de Levinson-Durbin, de transformation A(z)→LSP, de calcul des paramètres LSPi non quantifiées pour chaque sous-trame (i=0,...,3) par interpolation entre les LSP de la trame passée et ceux de la trame courante, et de transformation inverse (LSPi→ Ai(z)).These 4 modes of the NB-AMR encoder (7.4, 6.7, 5.9, 5.15) have identical modules such as preprocessing, analysis of linear prediction coefficients, signal calculation weighted. The signal preprocessing is 80 Hz high-pass cut-off filtering to suppress the continuous components combined with division of the input signals to avoid overflows. The LPC analysis includes sub-modules of windowing, autocorrelation calculation, implementation of the Levinson-Durbin algorithm, transformation A (z) → LSP, calculation of unquantized LSP i parameters for each sub-module. -frame (i = 0, ..., 3) by interpolation between the LSPs of the past frame and those of the current frame, and of inverse transformation (LSP i → A i (z)).

Le calcul du signal de parole pondéré réside en un filtrage par le filtre de pondération perceptuelle (Wi(z)=Ai(z/γ1)/Ai(z/γ2) où Ai(z) est le filtre non quantifié de la sous-trame d'indice i avec γ1 =0,94 et γ2=0,6).The calculation of the weighted speech signal resides in a filtering by the perceptual weighting filter (W i (z) = A i (z / γ 1 ) / A i (z / γ 2 ) where A i (z) is the filter unquantized subframe of index i with γ 1 = 0.94 and γ 2 = 0.6).

D'autres blocs fonctionnels ne sont identiques que pour trois de ces modes (7,4; 6,7; 5,9). Par exemple, la recherche du retard LTP en boucle ouverte effectuée sur le signal pondéré une fois par super sous-trame pour ces trois modes. Pour le mode à 5,15, elle n'est effectuée en revanche qu'une fois par trame.Other functional blocks are identical for only three of these modes (7.4, 6.7, 5.9). For example, searching for LTP delay in open loop performed on the weighted signal once per super subframe for these three modes. For the 5.15 mode, however, it is performed only once per frame.

De même, si les quatre modes utilisent une quantification vectorielle pondérée prédictive MA (pour "Moving Average") d'ordre 1 à moyenne supprimée et par produit cartésien des paramètres LSP dans le domaine fréquentiel normalisé, la quantification des paramètres LSP du mode à 5,15 kbit/s se fait sur 23 bits, celle des trois autres modes sur 26 bits. Après transformation dans le domaine fréquentiel normalisé, la quantification vectorielle par produit cartésien (dite "split VQ") des paramètres LSP scinde les 10 paramètres LSP en 3 sous-vecteurs, de dimension 3, 3 et 4. Le premier sous-vecteur composé des 3 premiers LSP est quantifié sur 8 bits par le même dictionnaire pour les quatre modes. Le deuxième sous-vecteur composé des 3 LSP suivants est quantifié pour les 3 modes haut débit par un dictionnaire de taille 512 (9 bits) et pour le mode à 5,15 par la moitié de ce dictionnaire (un vecteur sur 2). Le troisième et dernier sous-vecteur composé des 4 derniers LSP est quantifié pour les 3 modes haut débit par un dictionnaire de taille 512 (9 bits) et pour le mode de plus faible débit par un dictionnaire de taille 128 (7 bits). La transformation dans le domaine fréquentiel normalisé, le calcul des poids du critère d'erreur quadratique et la prédiction MA (pour "Moving Average") du résidu LSP à quantifier sont identiques pour les 4 modes. Les trois modes haut débit utilisant les même dictionnaires pour quantifier les LSP, ils peuvent partager, en plus du même module de quantification vectorielle, la transformation inverse (pour revenir du domaine fréquentiel normalisé vers le domaine en cosinus), ainsi que le calcul des LSPQ i quantifiées pour chaque sous-trame (i=0,...,3) par interpolation entre les LSP quantifiés de la trame passée et ceux de la trame courante, et enfin la transformation inverse LSPQ i → AQ i(z).Likewise, if the four modes use an average 1-to-average predefined weighted vector prediction MA (for "Moving Average") quantization and LSP parameters in the normalized frequency domain, the quantization of the LSP parameters from , 15 kbit / s is done on 23 bits, that of the other three modes on 26 bits. After transformation into the normalized frequency domain, the Cartesian product vector quantization (so-called " split VQ ") of the LSP parameters divides the LSP parameters into 3 sub-vectors, of size 3, 3 and 4. The first sub-vector composed of The first 3 LSP is quantized on 8 bits by the same dictionary for the four modes. The second sub-vector composed of the following 3 LSPs is quantized for the 3 high-speed modes by a dictionary of size 512 (9 bits) and for the mode with 5,15 by half of this dictionary (one vector out of 2). The third and last sub-vector composed of the last 4 LSPs is quantized for the 3 high-speed modes by a dictionary of size 512 (9 bits) and for the mode of lower bit rate by a dictionary of size 128 (7 bits). The transformation in the normalized frequency domain, the calculation of the squared error criterion weights and the MA prediction (for " Moving Average ") of the LSP residue to be quantized are identical for the 4 modes. Since the three broadband modes use the same dictionaries to quantify the LSPs, they can share, in addition to the same vector quantization module, the inverse transformation (to return from the normalized frequency domain to the cosine domain), as well as the calculation of the LSPs. Q i quantized for each sub-frame (i = 0, ..., 3) by interpolation between the quantized LSPs of the past frame and those of the current frame, and finally the inverse transformation LSP Q i → A Q i (z ).

Les recherches en boucle fermée des excitations adaptative et fixe sont faites séquentiellement et nécessitent au préalable le calcul de la réponse impulsionnelle du filtre de synthèse pondéré, puis de signaux-cible. La réponse impulsionnelle du filtre de synthèse pondéré (Ai(z/γ1)/[AQ i(z)Ai(z/γ2)]) est identique pour les 3 modes haut débit (7,4; 6,7; 5,9). Pour chaque sous-trame, le calcul du signal-cible pour l'excitation adaptative dépend du signal pondéré (indépendamment du mode); du filtre quantifié AQ i(z) (identique pour 3 des modes) et du passé de la sous-trame (différent pour chaque sous-trame autre que la première sous-trame). Pour chaque sous-trame, le signal-cible pour l'excitation fixe est obtenu en retirant au signal-cible précédent la contribution de l'excitation adaptative filtrée de cette sous-trame (qui est différente d'un mode à l'autre sauf pour la première sous-trame des 3 premiers modes).The closed-loop searches of the adaptive and fixed excitations are done sequentially and require the calculation of the impulse response of the weighted synthesis filter and then of the target signals beforehand. The impulse response of the weighted synthesis filter (A i (z / γ 1 ) / [A Q i (z) A i (z / γ 2 )]) is identical for the 3 high-speed modes (7.4, 6, 7; 5,9). For each subframe, the calculation of the target signal for the adaptive excitation depends on the weighted signal (regardless of the mode); the quantized filter A Q i (z) (identical for 3 modes) and the past subframe (different for each subframe other than the first subframe). For each subframe, the target signal for the fixed excitation is obtained by removing from the previous target signal the contribution of the filtered adaptive excitation of this subframe (which is different from one mode to another except for the first subframe of the first 3 modes).

Trois dictionnaires adaptatifs sont utilisés. Le premier dictionnaire, pour les sous-trames paires (i=0 et 2) des modes (7,4; 6,7; 5,9) et pour la première sous-trame du mode à 5,15, comporte 256 retards absolus fractionnaires, de résolution 1/3 dans l'intervalle [19 + 1/3,84 + 2/3] et de résolution entière dans l'intervalle [85,143]. La recherche dans ce dictionnaire de retards absolus est focalisée autour du retard trouvé en boucle ouverte (intervalle de ±5 pour le mode à 5,15, de ±3 pour les autres modes). Pour la première sous-trame des modes (7,4; 6,7; 5,9), le signal-cible et le retard en boucle ouverte étant identique, le résultat de cette recherche en boucle fermée l'est aussi. Les deux autres dictionnaires sont de type différentiel et permettent de coder la différence entre le retard courant et le retard entier Ti-1 le plus proche du retard fractionnaire de la sous-trame précédente. Le premier dictionnaire différentiel sur 5 bits, utilisé pour les sous-trames impaires du mode à 7,4, est de résolution 1/3 autour du retard entier Ti-1 dans l'intervalle [Ti-1-5 +2/3, Ti-1+4 +2/3]. Le deuxième dictionnaire différentiel sur 4 bits, inclus dans le premier, est utilisé pour les sous-trames impaires des modes à 6,7 et 5,9 ainsi que pour les trois dernières sous-trames du mode à 5,15. Ce deuxième dictionnaire est de résolution entière autour du retard entier Ti-1 dans l'intervalle [Ti-1-5, Ti-1+4] plus une résolution de 1/3 dans l'intervalle [Ti-1-1 + 2/3, Ti-1 + 2/3].Three adaptive dictionaries are used. The first dictionary, for the even subframes (i = 0 and 2) of the modes (7.4, 6.7, 5.9) and for the first subframe of the 5.15 mode, comprises 256 absolute delays. fractional, 1/3 resolution in the range [19 + 1 / 3.84 + 2/3] and full resolution in the range [85,143]. The search in this dictionary of absolute delays is focused around the delay found in open loop (range of ± 5 for the 5.15 mode, ± 3 for the other modes). For the first sub-frame of the modes (7.4, 6.7, 5.9), the target signal and the open-loop delay being identical, the result of this closed-loop search is also identical. The other two dictionaries are of differential type and make it possible to code the difference between the current delay and the integer delay T i-1 closest to the fractional delay of the preceding sub-frame. The first 5-bit differential dictionary, used for the odd subframes of the 7.4 mode, is 1/3 resolution around the entire delay T i-1 in the interval [T i-1 -5 + 2 / 3, T i-1 +4 + 2/3]. The second 4-bit differential dictionary, included in the first one, is used for the odd subframes of the modes at 6.7 and 5.9 as well as for the last three subframes of the 5.15 mode. This second dictionary is of integer resolution around the integer delay T i-1 in the interval [T i-1 -5, T i-1 +4] plus a resolution of 1/3 in the interval [T i-1 -1 + 2/3, T i-1 + 2/3].

Les dictionnaires fixes appartiennent à la famille bien connue des dictionnaires ACELP. La structure d'un répertoire ACELP est basée sur le concept ISPP (pour "Interleaved Single-Pulse Permutation") qui consiste à diviser l'ensemble des L positions en K pistes entrelacées, chacune des N impulsions étant localisée dans certaines pistes prédéfinies. Les 4 modes (7,4; 6,7; 5,9; 5,15) utilisent la même découpe des 40 échantillons d'une sous-trame en 5 pistes de longueur 8 entrelacées, comme le montre le tableau 2a. Le tableau 2b montre, quant à lui, pour les 3 modes (7,4; 6,7; 5,9) le débit du dictionnaire, le nombre d'impulsions et leur répartition dans les pistes. La répartition des 2 impulsions du dictionnaire ACELP à 9 bits du mode à 5,15 est encore plus contrainte. Tableau 2a: Découpe en pistes entrelacées des 40 positions d'une sous-trame du codeur 3GPP NB-AMR Piste Positions P0 0, 5, 10, 15, 20, 25, 30, 35 P1 1, 6, 11, 16, 21, 26, 31, 36 P2 2, 7, 12, 17, 22, 27, 32, 37 P3 3, 8, 13, 18, 23, 28, 33, 38 P4 4, 9, 14, 19, 24, 29, 34, 39 Tableau 2b: Répartition des impulsions dans les pistes pour les modes 7,4; 6,7; 5,9 du codeur 3GPP NB-AMR Mode (kbit/s) 7,4 6,7 5,9 Débit du dictionnaire ACELP
(positions+amplitudes)
17
(13+4)
14
(11+3)
11
(9+2)
Nombre d'impulsions 4 3 2 Pistes potentielles pour i0 p0 p0 p1, p3 Pistes potentielles pour i1 p1 p1, p3 p0, p1, p2, p4 Pistes potentielles pour i2 p2 p2, p4 - Pistes potentielles pour i3 p3, p4 - -
Fixed dictionaries belong to the well-known family of ACELP dictionaries. The structure of an ACELP directory is based on the ISPP ( Interleaved Single-Pulse Permutation ) concept, which consists of dividing all L positions into K interleaved tracks, each of the N pulses being located in certain predefined tracks. The four modes (7.4, 6.7, 5.9, 5.15) use the same slice of the 40 samples of a 5-track subframe of length 8 interleaved, as shown in Table 2a. Table 2b shows, for the 3 modes (7.4, 6.7, 5.9) the dictionary rate, the number of pulses and their distribution in the tracks. The distribution of the 2 pulses of the ACELP 9-bit dictionary of the 5.15 mode is even more constrained. <b> Table 2a: </ b> Interleaved Cutting of the 40 Positions of a Subframe of the 3GPP NB-AMR Encoder Track positions P 0 0, 5, 10, 15, 20, 25, 30, 35 P 1 1, 6, 11, 16, 21, 26, 31, 36 P 2 2, 7, 12, 17, 22, 27, 32, 37 P 3 3, 8, 13, 18, 23, 28, 33, 38 P 4 4, 9, 14, 19, 24, 29, 34, 39 Mode (kbit / s) 7.4 6.7 5.9 ACELP dictionary rate
(Positions + amplitudes)
17
(13 + 4)
14
(11 + 3)
11
(9 + 2)
Number of pulses 4 3 2 Potential tracks for i 0 p 0 p 0 p 1 , p 3 Potential tracks for i 1 p 1 p 1 , p 3 p 0 , p 1 , p 2 , p 4 Potential tracks for i 2 p 2 p 2 , p 4 - Potential tracks for i 3 p 3 , p 4 - -

Les gains des excitations adaptative et fixe sont quantifiés sur 7 ou 6 bits (avec une prédiction MA appliquée au gain de l'excitation fixe) par une quantification vectorielle conjointe minimisant le critère CELP.The gains of the adaptive and fixed excitations are quantified on 7 or 6 bits (with an MA prediction applied to the gain of the fixed excitation) by a joint vector quantization minimizing the CELP criterion.

* Codage multi-modes à décision a posteriori n'exploitant que la mutualisation des blocs fonctionnels identiques * Post-decision multi-mode coding exploiting only the sharing of identical functional blocks

A partir de ce schéma de codage, la construction d'un codeur multi-modes à décision a posteriori peut être réalisée en mettant en commun les blocs fonctionnels suivants.From this coding scheme, the construction of a post-decision multi-mode coder can be achieved by pooling the following functional blocks.

En se référant à la figure 8, pour les 4 modes, on effectue en commun :

  • le pré-traitement (bloc 81),
  • l'analyse des coefficients de prédiction linéaire (fenêtrage et calcul des autocorrélations 82, mise en oeuvre de l'algorithme de Levinson-Durbin 83, transformation A(z)→LSP 84, interpolation des LSP et transformation inverse 862),
  • le calcul du signal d'entrée pondéré 87,
  • la transformation des paramètres LSP dans le domaine fréquentiel normalisé, le calcul des poids du critère d'erreur quadratique pour la quantification vectorielle des LSP, la prédiction MA du résidu LSP, la quantification vectorielle des 3 premiers LSP (dans le bloc 85).
Referring to the figure 8 for the 4 modes, we perform in common:
  • pre-treatment (block 81),
  • analysis of linear prediction coefficients (windowing and calculation of autocorrelations 82, implementation of the Levinson-Durbin algorithm 83, transformation A (z) → LSP 84, interpolation of LSPs and inverse transformation 862),
  • the calculation of the weighted input signal 87,
  • the transformation of the LSP parameters in the standardized frequency domain, the calculation of the weights of the quadratic error criterion for the vector quantization of the LSPs, the MA prediction of the LSP residue, the vector quantization of the first 3 LSPs (in block 85).

Pour tous ces blocs, leur complexité cumulée est ainsi divisée par 4. Pour les 3 modes de plus haut débit (7,4; 6,7; 5,9), on effectue :

  • la quantification vectorielle des 7 derniers LSP (une fois par trame) (dans le bloc 85 de la figure 8),
  • la recherche du retard LTP en boucle ouverte (2 fois par trame) (bloc 88),
  • l'interpolation des LSP quantifiés (861) et la transformation inverse vers les filtres AQ i (pour chaque sous-trame),
  • le calcul de la réponse impulsionnelle 89 du filtre de synthèse pondéré (pour chaque sous-trame).
For all these blocks, their cumulative complexity is thus divided by 4. For the 3 modes of higher bitrate (7.4, 6.7, 5.9), we perform:
  • the vector quantization of the last 7 LSPs (once per frame) (in block 85 of the figure 8 )
  • the search for the LTP delay in open loop (2 times per frame) (block 88),
  • the interpolation of the quantized LSPs (861) and the inverse transformation to the filters A Q i (for each subframe),
  • calculating the impulse response 89 of the weighted synthesis filter (for each subframe).

Pour ces blocs, les calculs ne sont plus effectués 4 fois mais 2 fois, une fois pour les 3 modes à plus haut débit et une fois pour le mode à faible débit. Leur complexité est donc divisée par 2.For these blocks, the calculations are no longer performed 4 times but twice, once for the 3 higher rate modes and once for the low rate mode. Their complexity is therefore divided by 2.

On peut aussi, pour ces 3 modes de plus haut débit, mutualiser pour la première sous-trame le calcul des signaux-cible pour l'excitation fixe (bloc 91 sur la figure 8) et adaptative (bloc 90), ainsi que la recherche LTP en boucle fermée (bloc 881). II faut noter que la mutualisation de ces opérations pour la première sous-trame ne produit des résultats identiques que dans le contexte du codage multiple de type multi-modes à décision à posteriori. Dans le contexte général de codage multiple, le passé de la première sous-trame est, comme pour les 3 autres sous-trames, différent selon les débits, ces opérations conduisent généralement alors à des résultats différents.It is also possible, for these 3 higher rate modes, to pool for the first sub-frame the calculation of the target signals for the fixed excitation (block 91 on the figure 8 ) and adaptive (block 90), as well as the closed-loop LTP search (block 881). It should be noted that the pooling of these operations for the first sub-frame produces identical results only in the context of multiple coding of multi-mode type with a posteriori decision. In the general context of multiple coding, the past of the first sub-frame is, as for the other 3 sub-frames, different according to the flow rates, these operations generally lead to different results.

* Codage multi-modes à décision a posteriori avancée* Multi-mode coding with advanced posterior decision

Des blocs fonctionnels non identiques peuvent être accélérés en exploitant ceux d'un autre mode ou d'un module de traitement commun. Selon les contraintes de l'application (en termes de qualité et/ou de complexité), on peut utiliser différentes variantes. Quelques exemples sont décrits ci-après. Il est aussi possible de s'appuyer sur des techniques de transcodage intelligent entre codeurs CELP.Non-identical functional blocks can be accelerated by exploiting those of another mode or a common processing module. Depending on the constraints of the application (in terms of quality and / or complexity), different variants can be used. Some examples are described below. It is also possible to rely on intelligent transcoding techniques between CELP coders.

* La quantification vectorielle du deuxième sous-vecteur de LSP* The vector quantization of the second sub-vector of LSP

On peut, comme dans le cas du mode de réalisation pour le codeur TDAC, exploiter l'imbrication de certains dictionnaires pour accélérer les calculs. Ainsi, le dictionnaire du deuxième sous-vecteur de LSP du mode à 5,15 étant inclus dans celui des 3 autres modes, la quantification de ce sous-vecteur Y par les 4 modes peut être ainsi avantageusement combinée:

  • Etape 1: Chercher son plus proche voisin Yl dans le plus petit dictionnaire (correspondant à la moitié du grand dictionnaire)
    • ○ Yl quantifie Y pour le mode à 5,15
  • Etape 2: Chercher le plus proche voisin Yh dans le complémentaire dans le grand dictionnaire (soit l'autre moitié du dictionnaire)
  • Etape 3: Tester si le plus proche voisin de Y dans le dictionnaire à 9 bits est Yl (cas "Flag=0") ou Yh (cas "Flag=1")
    • ○ cas "Flag=0" : Yl quantifie aussi Y pour les modes à 7,4; 6,7 et 5,9
    • ○ sinon (cas "Flag=1"), Yh quantifie Y pour les modes à 7,4; 6,7 et 5,9
One can, as in the case of the embodiment for the TDAC encoder, exploit the nesting of certain dictionaries to accelerate the computations. Thus, the dictionary of the second LSP sub-vector of the 5.15 mode being included in that of the other 3 modes, the quantification of this sub-vector Y by the four modes can thus be advantageously combined:
  • Step 1: Find your nearest neighbor Y l in the smallest dictionary (corresponding to half of the big dictionary)
    • ○ Y l quantifies Y for the 5.15 mode
  • Step 2: Find the nearest neighbor Y h in the complement in the big dictionary (the other half of the dictionary)
  • Step 3: Test if Y's nearest neighbor in the 9-bit dictionary is Y l (case "Flag = 0") or Y h (case "Flag = 1")
    • ○ "Flag = 0" case: Y l also quantifies Y for 7.4 modes; 6.7 and 5.9
    • ○ else (case "Flag = 1"), Y h quantizes Y for modes at 7.4; 6.7 and 5.9

Cette mise en oeuvre donne un résultat identique à celui du codage multi-mode non optimisé. Si l'on désire réduire davantage la complexité de la quantification, on peut s'arrêter à l'étape 1 et prendre Yl comme vecteur quantifié pour les modes haut débit si ce vecteur est jugé suffisamment proche de Y. Cette simplification peut donc donner un résultat différent d'une recherche exhaustive.This implementation gives a result identical to that of the non-optimized multi-mode coding. If one wishes to further reduce the complexity of the quantization, one can stop at step 1 and take Y 1 as a quantized vector for the high-speed modes if this vector is considered sufficiently close to Y. This simplification can therefore give a different result from an exhaustive search.

* Accélération de la recherche LTP en boucle ouverte* Acceleration of LTP research in open loop

La recherche du retard LTP en boucle ouverte du mode à 5,15 peut exploiter les résultats de celle des autres modes. Si les deux retards en boucle ouverte trouvés sur les 2 super sous-trames sont suffisamment proches pour permettre un codage différentiel, la recherche en boucle ouverte du mode à 5,15 n'est pas effectuée. On utilise plutôt les résultats des modes supérieurs. Sinon, on peut:

  • effectuer la recherche classique,
  • ou focaliser la recherche en boucle ouverte sur toute la trame autour des deux retards en boucle ouverte trouvés par les modes supérieurs.
The search for the open-loop LTP delay of the 5.15 mode can exploit the results of that of the other modes. If the two open-loop delays found on the 2 super-frames are close enough to allow differential coding, the open-loop search of the 5.15 mode is not performed. Rather, the results of the higher modes are used. Otherwise, we can:
  • perform the classic search,
  • or focus the open-loop search on the entire frame around the two open-loop delays found by the higher modes.

A l'inverse, on peut aussi effectuer d'abord la recherche du retard en boucle ouverte sur le mode à 5,15 et focaliser les deux recherches du retard en boucle ouverte des modes supérieurs autour de la valeur déterminée par le mode à 5,15.Conversely, it is also possible to first carry out the search for the open-loop delay in the 5.15 mode and focus the two searches for the open-loop delay of the higher modes around the value determined by the 5-mode. 15.

Dans une troisième variante plus avancée, illustrée à la figure 1d, on se propose de réaliser un codeur multi-modes en treillis permettant plusieurs combinaisons de blocs fonctionnels, chaque bloc fonctionnel possédant au moins deux modes de fonctionnement (ou débits). On construit ce nouveau codeur à partir des quatre débits du codeur NB-AMR cités ci-avant (5,15; 5,90; 6,70; 7,40). Dans ce codeur, on distingue quatre blocs fonctionnels: le bloc LPC, le bloc LTP, le bloc excitation fixe et le bloc de gains. En se référant au tableau 1 présenté ci-avant, le tableau 3a ci-après récapitule pour chacun de ces blocs fonctionnels, son nombre de débits et ses débits. Tableau 3a: Nombre de débits et débits des blocs fonctionnels pour les quatre modes (5,15; 5,90; 6,70; 7,40) du codeur NB-AMR. Bloc fonctionnel Nombre de débits Débits des blocs fonctionnels LPC (LSP) 2 26 et 23 Retard LTP 3 26, 24 et 20 Excitation fixe 4 68, 56, 44 et 36 Gains 2 28 et 24 In a third, more advanced variant, illustrated in figure 1d , it is proposed to realize a multi-mode trellis coder for several combinations of functional blocks, each functional block having at least two modes of operation (or flows). This new encoder was constructed from the four NB-AMR encoder rates mentioned above (5.15, 5.90, 6.70, 7.40). In this encoder, there are four functional blocks: the LPC block, the LTP block, the fixed excitation block and the gain block. Referring to Table 1 presented above, Table 3a below summarizes for each of these functional blocks, its number of flow rates and its flow rates. <b> Table 3a: </ b> Number of flow rates and rates of the functional blocks for the four modes (5,15, 5,90, 6,70, 7,40) of the NB-AMR encoder. Functional block Number of flows Flow of the functional blocks LPC (LSP) 2 26 and 23 LTP delay 3 26, 24 and 20 Fixed excitation 4 68, 56, 44 and 36 Earnings 2 28 and 24

On a donc P=4 blocs fonctionnels et 2 x 3 x 4 x 2 = 48 combinaisons possibles. Dans l'exemple particulier de réalisation, on choisit de ne pas considérer le haut-débit du bloc fonctionnel 2 (LTP débit 26 bits/trame). Un autre choix est possible, bien entendu.We thus have P = 4 functional blocks and 2 x 3 x 4 x 2 = 48 possible combinations. In the particular embodiment, it is chosen not to consider the high bit rate of the functional block 2 (LTP bit rate 26 bits / frame). Another choice is possible, of course.

Le codeur multi-débits ainsi obtenu possède une grande granularité en débits, avec 32 modes possibles donnés dans le tableau 3b. Toutefois, on indique que le codeur ainsi obtenu n'est pas interopérable avec le codeur NB-AMR précité. Dans le tableau 3b, les modes correspondants aux trois débits du NB-AMR (5,15; 5,90; 6,70) sont présentés en gras, l'exclusion du débit le plus élevé du bloc fonctionnel LTP éliminant le débit de 7,40. Tableau 3b: Débit par bloc fonctionnel et global du codeur multi-modes en treillis Paramètres LSP Retard LTP Excitation fixe Gains des excitations fixe et adaptative Total Débit par trame 23 20 36 24 103 23 20 36 28 107 23 20 44 24 111 23 20 44 28 115 23 20 56 24 123 23 20 56 28 127 23 20 68 24 135 23 20 68 28 139 23 24 36 24 107 23 24 36 28 111 23 24 44 24 115 23 24 44 28 119 23 24 56 24 127 23 24 56 28 131 23 24 68 24 139 23 24 68 28 143 26 20 36 24 106 26 20 36 28 110 26 20 44 24 114 26 20 44 28 118 26 20 56 24 126 26 20 56 28 130 26 20 68 24 138 26 20 68 28 142 26 24 36 24 110 26 24 36 28 114 26 24 44 24 118 26 24 44 28 122 26 24 56 24 130 26 24 56 28 134 26 24 68 24 142 26 24 68 28 146 The multi-rate encoder thus obtained has a high granularity in rates, with 32 possible modes given in Table 3b. However, it is indicated that the encoder thus obtained is not interoperable with the aforementioned NB-AMR encoder. In Table 3b, the modes corresponding to the three flows of the NB-AMR (5.15, 5.90, 6.70) are shown in bold, the exclusion of the highest bit rate of the LTP functional block eliminating the flow of 7 40. <b> Table 3b: </ b> Flow rate per functional and global block of the multi-mode lattice encoder Settings LSP LTP delay Fixed excitation Gains of fixed and adaptive excitations Total Flow per frame 23 20 36 24 103 23 20 36 28 107 23 20 44 24 111 23 20 44 28 115 23 20 56 24 123 23 20 56 28 127 23 20 68 24 135 23 20 68 28 139 23 24 36 24 107 23 24 36 28 111 23 24 44 24 115 23 24 44 28 119 23 24 56 24 127 23 24 56 28 131 23 24 68 24 139 23 24 68 28 143 26 20 36 24 106 26 20 36 28 110 26 20 44 24 114 26 20 44 28 118 26 20 56 24 126 26 20 56 28 130 26 20 68 24 138 26 20 68 28 142 26 24 36 24 110 26 24 36 28 114 26 24 44 24 118 26 24 44 28 122 26 24 56 24 130 26 24 56 28 134 26 24 68 24 142 26 24 68 28 146

Ce codeur possédant 32 débits possibles, 5 bits sont nécessaires pour identifier le mode utilisé. Comme dans la variante précédente, la mutualisation de blocs fonctionnels est exploitée. On applique des stratégies de codage différentes pour les différents blocs fonctionnels.
Par exemple, pour le bloc fonctionnel 1 comprenant la quantification des LSP, on privilégie le bas débit comme mentionné ci-avant de la manière suivante :

  • Le premier sous-vecteur composé des 3 premiers LSP est quantifié sur 8 bits par le même dictionnaire pour les deux débits associés à ce bloc fonctionnel,
  • Le deuxième sous-vecteur composé des 3 LSP suivants est quantifié sur 8 bits par le dictionnaire du plus petit débit. Ce dictionnaire correspondant à la moitié du dictionnaire de plus haut débit, on n'effectue la recherche dans l'autre moitié du dictionnaire que si la distance entre les 3 LSP et l'élément choisi dans le dictionnaire dépasse un certain seuil.
  • Le troisième et dernier sous-vecteur composé des 4 derniers LSP est quantifié par un dictionnaire de taille 512 (9 bits) et par un dictionnaire de taille 128 (7 bits).
This encoder has 32 possible bit rates, 5 bits are needed to identify the mode used. As in the previous variant, the pooling of functional blocks is exploited. Different coding strategies are applied for the different functional blocks.
For example, for the functional block 1 comprising the quantification of the LSPs, the low bit rate is preferred as mentioned above in the following manner:
  • The first sub-vector composed of the first 3 LSPs is quantized on 8 bits by the same dictionary for the two data rates associated with this functional block,
  • The second sub-vector composed of the following 3 LSPs is quantized on 8 bits by the dictionary of the smallest bit rate. This dictionary corresponds to half of the dictionary of higher speed, one searches in the other half of the dictionary only if the distance between the 3 LSP and the element chosen in the dictionary exceeds a certain threshold.
  • The third and last compound sub-vector of the last 4 LSPs is quantized by a dictionary of size 512 (9 bits) and by a dictionary of size 128 (7 bits).

Par contre, comme mentionné ci-avant dans la deuxième variante (correspondant au codage multi-modes à décision a posteriori avancée), on choisit de privilégier le haut-débit pour le bloc fonctionnel 2 (retard LTP). Dans le codeur NB-AMR, la recherche du retard LTP en boucle ouverte est effectuée deux fois par trame pour le retard LTP de 24 bits et elle est effectuée une seule fois par trame pour celui de 20 bits. Pour ce bloc fonctionnel, on souhaite favoriser le haut débit. Donc, le calcul du retard LTP en boucle ouverte est réalisé de la manière suivante:

  • On calcule deux retards en boucle ouverte sur les 2 super sous-trames. S'ils sont suffisamment proches pour permettre un codage différentiel; la recherche en boucle ouverte sur la trame entière n'est pas effectuée. On utilise plutôt les résultats des deux super sous-trames.
  • Sinon, on effectue une recherche en boucle ouverte sur toute la trame en la focalisant autour des deux retards en boucle ouverte trouvés précédemment. Une variante réduisant la complexité retient uniquement le retard en boucle ouverte de ia première.
On the other hand, as mentioned above in the second variant (corresponding to the multi-mode coding with advanced posterior decision), one chooses to privilege the high bitrate for the functional block 2 (LTP delay). In NB-AMR coder, the search for the open-loop LTP delay is performed twice per frame for the 24-bit LTP delay and is performed once per frame for the 20-bit one. For this functional block, it is desired to promote broadband. Therefore, the computation of the LTP delay in open loop is carried out as follows:
  • Two open-loop delays are calculated on the 2 super-subframes. If they are close enough to allow differential coding; the open loop search on the entire frame is not performed. Instead, the results of the two super-frames are used.
  • Otherwise, an open-loop search is performed on the entire frame by focusing it around the two open-loop delays previously found. A complexity-reducing variant retains only the open-loop delay of the first one.

Après certains blocs fonctionnels, il est possible de réaliser une sélection partielle permettant de réduire le nombre de combinaisons à explorer. Par exemple, après le bloc fonctionnel 1 (LPC), on peut éliminer les combinaisons avec 26 bits pour ce bloc si la performance du débit de 23 bits est suffisamment proche ou inversement éliminer le mode à 23 bits si sa performance est trop dégradée par rapport au mode à 26 bits.After some functional blocks, it is possible to make a partial selection to reduce the number of combinations to explore. For example, after Function Block 1 (LPC), 26-bit combinations can be eliminated for this block if the performance of the 23-bit rate is close enough or conversely eliminate the 23-bit mode if its performance is too degraded compared to in 26-bit mode.

Ainsi, la présente invention permet de fournir une solution efficace au problème de la complexité des codages multiples, par la mutualisation et l'accélération des calculs mis en oeuvre par les différents codeurs. Les structures de codage peuvent donc être représentées à l'aide de blocs fonctionnels décrivant les différentes opérations effectuées au cours d'un traitement. Les blocs fonctionnels des différents codages mis en oeuvre dans un codage multiple possèdent des relations fortes qui sont exploitées au sens de la présente invention. Ces relations sont particulièrement fortes lorsque les différents codages correspondent à différents modes d'une même structure.Thus, the present invention makes it possible to provide an effective solution to the problem of the complexity of multiple codings by pooling and accelerating the calculations implemented by the various coders. The coding structures can therefore be represented using functional blocks describing the various operations performed during a treatment. The functional blocks of the different encodings implemented in multiple coding have strong relationships that are exploited within the meaning of the present invention. These relations are particularly strong when the different codings correspond to different modes of the same structure.

On indique enfin que la présente invention est flexible du point de vue de la complexité. II est possible en effet de décider a priori la complexité maximum du codage multiple et d'adapter le nombre de codeurs explorés en fonction de cette complexité.Finally, it is pointed out that the present invention is flexible from the point of view of complexity. It is indeed possible to decide a priori the maximum complexity of the multiple coding and to adapt the number of coders explored as a function of this complexity.

Claims (27)

  1. Multiple compression coding method, in which an input signal is intended to feed in parallel at least one first coder and one second coder, each of the first and second coders comprising a succession of functional units, with a view to a compression coding of said input signal by each of the first and second coders,
    at least some of said functional units carrying out calculations to deliver respective parameters for the coding of said input signal by each coder,
    the first and second coders comprising respectively at least a first and a second functional unit which are designed to perform common operations,
    characterized in that
    - calculations are carried out, a same step and in a single unit, to deliver a same set of parameters to the first unit and to the second unit, and
    - in the case where the first and/or the second coder operates at a different data rate from said single unit, said set of parameters is adapted to the data rate of the first and/or second coder so as to be used by the first and/or second unit, respectively.
  2. Method according to Claim 1, characterized in that said single unit consists of one or more units of one of the first and second coders.
  3. Method according to Claim 1, characterized in that it comprises the preparatory steps hereinafter:
    a) identifying the functional units forming each coder, as well as one or more functions carried out by each unit,
    b) tagging, among said functions, functions which are common from one coder to another, and
    c) executing said common functions, once and for all, for some at least of all the coders, within at least a same calculation module.
  4. Method according to Claim 3, characterized in that, for each function executed in step c), at least one functional unit of a coder chosen from among said plurality of coders is used, and in that the unit of said chosen coder is designed to deliver partial results to the other coders, for an efficient coding, at said other coders, satisfying an optimal criterion between the complexity and the quality of the coding.
  5. Method according to Claim 4, in which the coders may operate at different respective data rates, characterized in that the chosen coder is the coder of lowest data rate, and in that the results obtained, following the execution of the function in step c) with parameters specific to the chosen coder, are adapted to the data rates of some at least of the other coders by a focused search for parameters for some at least of all the other modes, until the coder of highest data rate.
  6. Method according to Claim 4, in which the coders may operate at different respective data rates, characterized in that the chosen coder is the coder of highest data rate, and in that the results obtained, following the execution of the function in step c) with parameters specific to the chosen coder, are adapted to the data rates of some at least of the other coders by a focused search for parameters for some at least of all the other modes, until the coder of lowest data rate.
  7. Method according to Claim 5, taken in combination with Claim 6, characterized in that, for a given data rate, the functional unit of a coder operating at said given data rate is used as the calculation module, and at least some of the parameters specific to this coder are progressively adapted:
    - until the coder of highest data rate, by focused search, and
    - until the coder of lowest data rate, by focused search.
  8. Method according to Claim 2, in which the functional units of the various coders are arranged in a trellis, with several possible paths in the trellis, characterized in that each path of the trellis is defined by a combination of operating modes of the functional units, each functional unit feeding several possible variants of the succeeding functional unit.
  9. Method according to Claim 8, characterized in that there is provided a module for partial selection, after each coding step conducted by one or more functional units, which is capable of selecting the results provided by one or more of these functional units, for succeeding coding steps.
  10. Method according to Claim 8, in which the functional units may operate at different respective data rates while using respective parameters specific to said data rates, characterized in that, for a given functional unit, the path of the trellis chosen is that passing through the functional unit of lowest data rate, and in that the results obtained from said functional unit of lowest data rate are adapted to the data rates of some at least of the other functional units by a focused search for parameters for some at least of all the other functional units, until the functional unit of highest data rate.
  11. Method according to Claim 8, in which the functional units may operate at different respective data rates while using respective parameters specific to said data rates, characterized in that, for a given functional unit, the path of the trellis chosen is that passing through the functional unit of highest data rate, and in that the results obtained from said functional unit of highest data rate are adapted to the data rates of some at least of the other functional units by a focused search for parameters for some at least of all the other functional units, until the functional unit of lowest data rate.
  12. Method according to Claim 10, taken in combination with Claim 11, characterized in that, for a given data rate associated with the parameters of a functional unit of a coder, the functional unit operating at said given data rate is used as the calculation module, and at least some of the parameters specific to this functional unit are progressively adapted:
    - until the functional unit capable of operating at the lowest data rate, by focused search, and
    - until the functional unit capable of operating at the highest data rate, by focused search.
  13. Method according to Claim 3, characterized in that said calculation module is a module independent of said coders, and designed to redistribute results obtained in step c) to all the coders.
  14. Method according to Claim 13, taken in combination with Claim 3, characterized in that the independent module and the unit or units of one at least of the coders are designed to mutually exchange results obtained in step c), and in that the calculation module is designed to perform an adaptation transcoding between functional units of different coders.
  15. Method according to one of Claims 13 and 14, characterized in that the independent module comprises a unit for at least partial coding and an adaptation transcoding unit.
  16. Method according to one of the preceding claims, in which the coders in parallel are designed to operate under multi-mode coding, characterized in that an a posteriori selection module is provided, capable of selecting a coder from among the coders.
  17. Method according to Claim 16, characterized in that there is provided a module for partial selection, after each coding step conducted by one or more functional units, which is independent of the coders and capable of selecting one or more coders.
  18. Method according to one of the preceding claims, in which the coders are of transform-based type, characterized in that the calculation module comprises a bit allocation unit, shared between all the coders, each allocation of bits performed for a coder being followed by an adaptation to this coder especially as a function of its data rate.
  19. Method according to Claim 18, characterized in that the method furthermore comprises a quantization step, the results of which are provided to all the coders.
  20. Method according to Claim 19, characterized in that it furthermore comprises steps common to all the coders from among:
    - a time-frequency transform (MDCT),
    - a detection of voicing in the input signal,
    - a detection of tonality,
    - the determination of a masking curve,
    - and a spectral envelope coding.
  21. Method according to Claim 18, in which the coders perform a sub-band coding (MPEG-1), characterized in that the method furthermore comprises steps common to all the coders from among:
    - the application of a bank of analysis filters,
    - a determination of scale factors,
    - a spectral transform calculation (FFT),
    - and the determination of masking thresholds in accordance with a psychoacoustic model.
  22. Method according to one of Claims 1 to 17, in which the coders are of the analysis by synthesis type (CELP), characterized in that the method comprises steps common to all the coders from among at least:
    - a preprocessing,
    - the analysis of linear-prediction coefficients,
    - a weighted input signal calculation,
    - and a quantization for at least some of the parameters.
  23. Method according to Claim 22, taken in combination with Claim 17, characterized in that the partial-selection module is implemented after a shared step of vector quantization for short-term parameters (LPC) .
  24. Method according to Claim 22, taken in combination with Claim 17, characterized in that the partial-selection module is implemented after a shared step of long-term parameter (LTP) open-loop search.
  25. Computer program product intended to be stored in a memory of a processing unit, especially of a computer or mobile terminal, or on a removable memory medium and intended to cooperate with a reader of the processing unit, characterized in that it comprises instructions for the implementation of the transcoding method according to one of the preceding claims.
  26. Device for aiding a multiple compression coding, in which coding an input signal is intended to feed in parallel a plurality of coders each comprising a succession of functional units, with a view to a compression coding of said signal by each coder, characterized in that it comprises a memory storing the instructions of a computer program product according to Claim 25.
  27. Device according to Claim 26, characterized in that it furthermore comprises an independent calculation module (MI) for the implementation of the method according to one of Claims 13 to 17 and 23,24.
EP04805538A 2003-12-10 2004-11-24 Optimized multiple coding method Not-in-force EP1692689B1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PL04805538T PL1692689T3 (en) 2003-12-10 2004-11-24 Optimized multiple coding method

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
FR0314490A FR2867649A1 (en) 2003-12-10 2003-12-10 OPTIMIZED MULTIPLE CODING METHOD
PCT/FR2004/003009 WO2005066938A1 (en) 2003-12-10 2004-11-24 Optimized multiple coding method

Publications (2)

Publication Number Publication Date
EP1692689A1 EP1692689A1 (en) 2006-08-23
EP1692689B1 true EP1692689B1 (en) 2009-09-09

Family

ID=34746281

Family Applications (1)

Application Number Title Priority Date Filing Date
EP04805538A Not-in-force EP1692689B1 (en) 2003-12-10 2004-11-24 Optimized multiple coding method

Country Status (12)

Country Link
US (1) US7792679B2 (en)
EP (1) EP1692689B1 (en)
JP (1) JP4879748B2 (en)
KR (1) KR101175651B1 (en)
CN (1) CN1890714B (en)
AT (1) ATE442646T1 (en)
DE (1) DE602004023115D1 (en)
ES (1) ES2333020T3 (en)
FR (1) FR2867649A1 (en)
PL (1) PL1692689T3 (en)
WO (1) WO2005066938A1 (en)
ZA (1) ZA200604623B (en)

Families Citing this family (27)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7987089B2 (en) * 2006-07-31 2011-07-26 Qualcomm Incorporated Systems and methods for modifying a zero pad region of a windowed frame of an audio signal
US8271554B2 (en) 2006-10-19 2012-09-18 Lg Electronics Encoding method and apparatus and decoding method and apparatus
KR101411900B1 (en) * 2007-05-08 2014-06-26 삼성전자주식회사 Method and apparatus for encoding and decoding audio signal
US9653088B2 (en) * 2007-06-13 2017-05-16 Qualcomm Incorporated Systems, methods, and apparatus for signal encoding using pitch-regularizing and non-pitch-regularizing coding
KR101403340B1 (en) * 2007-08-02 2014-06-09 삼성전자주식회사 Method and apparatus for transcoding
CA2729752C (en) * 2008-07-10 2018-06-05 Voiceage Corporation Multi-reference lpc filter quantization and inverse quantization device and method
FR2936898A1 (en) * 2008-10-08 2010-04-09 France Telecom CRITICAL SAMPLING CODING WITH PREDICTIVE ENCODER
MX2011011399A (en) * 2008-10-17 2012-06-27 Univ Friedrich Alexander Er Audio coding using downmix.
GB2466201B (en) * 2008-12-10 2012-07-11 Skype Ltd Regeneration of wideband speech
GB0822537D0 (en) 2008-12-10 2009-01-14 Skype Ltd Regeneration of wideband speech
US9947340B2 (en) * 2008-12-10 2018-04-17 Skype Regeneration of wideband speech
KR20110001130A (en) * 2009-06-29 2011-01-06 삼성전자주식회사 Apparatus and method for encoding and decoding audio signals using weighted linear prediction transform
KR101747917B1 (en) 2010-10-18 2017-06-15 삼성전자주식회사 Apparatus and method for determining weighting function having low complexity for lpc coefficients quantization
CN102394658A (en) * 2011-10-16 2012-03-28 西南科技大学 Composite compression method oriented to mechanical vibration signal
US9386267B1 (en) * 2012-02-14 2016-07-05 Arris Enterprises, Inc. Cooperative transcoding to multiple streams
JP2014123865A (en) * 2012-12-21 2014-07-03 Xacti Corp Image processing apparatus and imaging apparatus
US9549178B2 (en) 2012-12-26 2017-01-17 Verizon Patent And Licensing Inc. Segmenting and transcoding of video and/or audio data
KR101595397B1 (en) * 2013-07-26 2016-02-29 경희대학교 산학협력단 Method and apparatus for integrated encoding/decoding of different multilayer video codec
WO2015012514A1 (en) * 2013-07-26 2015-01-29 경희대학교 산학협력단 Method and apparatus for integrally encoding/decoding different multi-layer video codecs
CN104572751A (en) * 2013-10-24 2015-04-29 携程计算机技术(上海)有限公司 Compression storage method and system for calling center sound recording files
SE538512C2 (en) 2014-11-26 2016-08-30 Kelicomp Ab Improved compression and encryption of a file
SE544304C2 (en) * 2015-04-17 2022-03-29 URAEUS Communication Systems AB Improved compression and encryption of a file
US10872598B2 (en) * 2017-02-24 2020-12-22 Baidu Usa Llc Systems and methods for real-time neural text-to-speech
US10896669B2 (en) 2017-05-19 2021-01-19 Baidu Usa Llc Systems and methods for multi-speaker neural text-to-speech
US10872596B2 (en) 2017-10-19 2020-12-22 Baidu Usa Llc Systems and methods for parallel wave generation in end-to-end text-to-speech
WO2021248473A1 (en) 2020-06-12 2021-12-16 Baidu.Com Times Technology (Beijing) Co., Ltd. Personalized speech-to-video with three-dimensional (3d) skeleton regularization and expressive body poses
US11587548B2 (en) * 2020-06-12 2023-02-21 Baidu Usa Llc Text-driven video synthesis with phonetic dictionary

Family Cites Families (54)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH0398318A (en) * 1989-09-11 1991-04-23 Fujitsu Ltd Voice coding system
US5495555A (en) * 1992-06-01 1996-02-27 Hughes Aircraft Company High quality low bit rate celp-based speech codec
US5734789A (en) * 1992-06-01 1998-03-31 Hughes Electronics Voiced, unvoiced or noise modes in a CELP vocoder
JP3227291B2 (en) * 1993-12-16 2001-11-12 シャープ株式会社 Data encoding device
US5602961A (en) * 1994-05-31 1997-02-11 Alaris, Inc. Method and apparatus for speech compression using multi-mode code excited linear predictive coding
US5956674A (en) * 1995-12-01 1999-09-21 Digital Theater Systems, Inc. Multi-channel predictive subband audio coder using psychoacoustic adaptive bit allocation in frequency, time and over the multiple channels
US5987506A (en) * 1996-11-22 1999-11-16 Mangosoft Corporation Remote access and geographically distributed computers in a globally addressable storage environment
JP3134817B2 (en) * 1997-07-11 2001-02-13 日本電気株式会社 Audio encoding / decoding device
US6141638A (en) * 1998-05-28 2000-10-31 Motorola, Inc. Method and apparatus for coding an information signal
US6249758B1 (en) * 1998-06-30 2001-06-19 Nortel Networks Limited Apparatus and method for coding speech signals by making use of voice/unvoiced characteristics of the speech signals
US6173257B1 (en) * 1998-08-24 2001-01-09 Conexant Systems, Inc Completed fixed codebook for speech encoder
US6192335B1 (en) * 1998-09-01 2001-02-20 Telefonaktieboiaget Lm Ericsson (Publ) Adaptive combining of multi-mode coding for voiced speech and noise-like signals
JP3579309B2 (en) * 1998-09-09 2004-10-20 日本電信電話株式会社 Image quality adjusting method, video communication device using the method, and recording medium recording the method
SE521225C2 (en) * 1998-09-16 2003-10-14 Ericsson Telefon Ab L M Method and apparatus for CELP encoding / decoding
US7272556B1 (en) * 1998-09-23 2007-09-18 Lucent Technologies Inc. Scalable and embedded codec for speech and audio signals
US6691084B2 (en) * 1998-12-21 2004-02-10 Qualcomm Incorporated Multiple mode variable rate speech coding
US6260009B1 (en) * 1999-02-12 2001-07-10 Qualcomm Incorporated CELP-based to CELP-based vocoder packet translation
US6640209B1 (en) * 1999-02-26 2003-10-28 Qualcomm Incorporated Closed-loop multimode mixed-domain linear prediction (MDLP) speech coder
DE19911179C1 (en) * 1999-03-12 2000-11-02 Deutsche Telekom Mobil Method for adapting the operating mode of a multi-mode codec to changing radio conditions in a CDMA mobile radio network
JP2000287213A (en) * 1999-03-31 2000-10-13 Victor Co Of Japan Ltd Moving image encoder
US6532593B1 (en) * 1999-08-17 2003-03-11 General Instrument Corporation Transcoding for consumer set-top storage application
AU7486200A (en) * 1999-09-22 2001-04-24 Conexant Systems, Inc. Multimode speech encoder
US6581032B1 (en) * 1999-09-22 2003-06-17 Conexant Systems, Inc. Bitstream protocol for transmission of encoded voice signals
US6604070B1 (en) * 1999-09-22 2003-08-05 Conexant Systems, Inc. System of encoding and decoding speech signals
US6574593B1 (en) * 1999-09-22 2003-06-03 Conexant Systems, Inc. Codebook tables for encoding and decoding
AU1359601A (en) * 1999-11-03 2001-05-14 Tellabs Operations, Inc. Integrated voice processing system for packet networks
JP3549788B2 (en) * 1999-11-05 2004-08-04 三菱電機株式会社 Multi-stage encoding method, multi-stage decoding method, multi-stage encoding device, multi-stage decoding device, and information transmission system using these
FR2802329B1 (en) * 1999-12-08 2003-03-28 France Telecom PROCESS FOR PROCESSING AT LEAST ONE AUDIO CODE BINARY FLOW ORGANIZED IN THE FORM OF FRAMES
EP1164580B1 (en) * 2000-01-11 2015-10-28 Panasonic Intellectual Property Management Co., Ltd. Multi-mode voice encoding device and decoding device
SE519981C2 (en) * 2000-09-15 2003-05-06 Ericsson Telefon Ab L M Coding and decoding of signals from multiple channels
SE519976C2 (en) * 2000-09-15 2003-05-06 Ericsson Telefon Ab L M Coding and decoding of signals from multiple channels
US6615169B1 (en) * 2000-10-18 2003-09-02 Nokia Corporation High frequency enhancement layer coding in wideband speech codec
JP2002202799A (en) * 2000-10-30 2002-07-19 Fujitsu Ltd Voice code conversion apparatus
EP2627008A3 (en) * 2000-12-29 2013-09-11 Intel Mobile Communications GmbH Channel codec processor configurable for multiple wireless communications standards
US6614370B2 (en) * 2001-01-26 2003-09-02 Oded Gottesman Redundant compression techniques for transmitting data over degraded communication links and/or storing data on media subject to degradation
US6658383B2 (en) * 2001-06-26 2003-12-02 Microsoft Corporation Method for coding speech and music signals
US7200561B2 (en) * 2001-08-23 2007-04-03 Nippon Telegraph And Telephone Corporation Digital signal coding and decoding methods and apparatuses and programs therefor
JP2003125406A (en) * 2001-09-25 2003-04-25 Hewlett Packard Co <Hp> Method and system for optimizing mode selection for video coding based on oriented aperiodic graph
US7095343B2 (en) * 2001-10-09 2006-08-22 Trustees Of Princeton University code compression algorithms and architectures for embedded systems
JP2003195893A (en) * 2001-12-26 2003-07-09 Toshiba Corp Device and method for speech reproduction
US6829579B2 (en) * 2002-01-08 2004-12-07 Dilithium Networks, Inc. Transcoding method and system between CELP-based speech codes
US7254533B1 (en) * 2002-10-17 2007-08-07 Dilithium Networks Pty Ltd. Method and apparatus for a thin CELP voice codec
US7133521B2 (en) * 2002-10-25 2006-11-07 Dilithium Networks Pty Ltd. Method and apparatus for DTMF detection and voice mixing in the CELP parameter domain
US7023880B2 (en) * 2002-10-28 2006-04-04 Qualcomm Incorporated Re-formatting variable-rate vocoder frames for inter-system transmissions
JP2004208280A (en) * 2002-12-09 2004-07-22 Hitachi Ltd Encoding apparatus and encoding method
WO2004064041A1 (en) * 2003-01-09 2004-07-29 Dilithium Networks Pty Limited Method and apparatus for improved quality voice transcoding
KR100554164B1 (en) * 2003-07-11 2006-02-22 학교법인연세대학교 Transcoder between two speech codecs having difference CELP type and method thereof
US7469209B2 (en) * 2003-08-14 2008-12-23 Dilithium Networks Pty Ltd. Method and apparatus for frame classification and rate determination in voice transcoders for telecommunications
US7305055B1 (en) * 2003-08-18 2007-12-04 Qualcomm Incorporated Search-efficient MIMO trellis decoder
US7433815B2 (en) * 2003-09-10 2008-10-07 Dilithium Networks Pty Ltd. Method and apparatus for voice transcoding between variable rate coders
US7613606B2 (en) * 2003-10-02 2009-11-03 Nokia Corporation Speech codecs
US7170988B2 (en) * 2003-10-27 2007-01-30 Motorola, Inc. Method and apparatus for network communication
FR2867648A1 (en) * 2003-12-10 2005-09-16 France Telecom TRANSCODING BETWEEN INDICES OF MULTI-IMPULSE DICTIONARIES USED IN COMPRESSION CODING OF DIGITAL SIGNALS
US20050258983A1 (en) * 2004-05-11 2005-11-24 Dilithium Holdings Pty Ltd. (An Australian Corporation) Method and apparatus for voice trans-rating in multi-rate voice coders for telecommunications

Also Published As

Publication number Publication date
EP1692689A1 (en) 2006-08-23
US20070150271A1 (en) 2007-06-28
JP2007515677A (en) 2007-06-14
KR20060131782A (en) 2006-12-20
FR2867649A1 (en) 2005-09-16
ATE442646T1 (en) 2009-09-15
ZA200604623B (en) 2007-11-28
DE602004023115D1 (en) 2009-10-22
CN1890714A (en) 2007-01-03
PL1692689T3 (en) 2010-02-26
US7792679B2 (en) 2010-09-07
CN1890714B (en) 2010-12-29
WO2005066938A1 (en) 2005-07-21
JP4879748B2 (en) 2012-02-22
ES2333020T3 (en) 2010-02-16
KR101175651B1 (en) 2012-08-21

Similar Documents

Publication Publication Date Title
EP1692689B1 (en) Optimized multiple coding method
RU2326450C2 (en) Method and device for vector quantisation with reliable prediction of linear prediction parameters in voice coding at variable bit rate
EP2452337B1 (en) Allocation of bits in an enhancement coding/decoding for improving a hierarchical coding/decoding of digital audio signals
EP2254110B1 (en) Stereo signal encoding device, stereo signal decoding device and methods for them
WO2008104663A1 (en) Advanced encoding / decoding of audio digital signals
WO2009055493A1 (en) Scalable speech and audio encoding using combinatorial encoding of mdct spectrum
CA2766864C (en) Improved coding /decoding of digital audio signals
EP2128858B1 (en) Encoding device and encoding method
US7634402B2 (en) Apparatus for coding of variable bitrate wideband speech and audio signals, and a method thereof
EP1836699B1 (en) Method and device for carrying out optimized audio coding between two long-term prediction models
FR2784218A1 (en) LOW-SPEED SPEECH CODING METHOD
US6611797B1 (en) Speech coding/decoding method and apparatus
WO2023165946A1 (en) Optimised encoding and decoding of an audio signal using a neural network-based autoencoder
EP0616315A1 (en) Digital speech coding and decoding device, process for scanning a pseudo-logarithmic LTP codebook and process of LTP analysis
CA2567162A1 (en) Method for quantifying an ultra low-rate speech encoder
WO2011144863A1 (en) Encoding with noise shaping in a hierarchical encoder
WO2002029786A1 (en) Method and device for segmental coding of an audio signal
FR2791166A1 (en) METHODS OF ENCODING, DECODING AND TRANSCODING
FR2980620A1 (en) Method for processing decoded audio frequency signal, e.g. coded voice signal including music, involves performing spectral attenuation of residue, and combining residue and attenuated signal from spectrum of tonal components
FR2737360A1 (en) Audio digital signal coding method of successive sample blocks - using spectral analysis to select vector dictionary for each sample block and allocating vector and scalar quantisation bits
KR19980036961A (en) Speech encoding and decoding apparatus and method

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

17P Request for examination filed

Effective date: 20060526

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LI LU MC NL PL PT RO SE SI SK TR

DAX Request for extension of the european patent (deleted)
17Q First examination report despatched

Effective date: 20070621

GRAP Despatch of communication of intention to grant a patent

Free format text: ORIGINAL CODE: EPIDOSNIGR1

GRAS Grant fee paid

Free format text: ORIGINAL CODE: EPIDOSNIGR3

GRAA (expected) grant

Free format text: ORIGINAL CODE: 0009210

AK Designated contracting states

Kind code of ref document: B1

Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LI LU MC NL PL PT RO SE SI SK TR

REG Reference to a national code

Ref country code: GB

Ref legal event code: FG4D

Free format text: NOT ENGLISH

REG Reference to a national code

Ref country code: CH

Ref legal event code: EP

REG Reference to a national code

Ref country code: IE

Ref legal event code: FG4D

REF Corresponds to:

Ref document number: 602004023115

Country of ref document: DE

Date of ref document: 20091022

Kind code of ref document: P

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: SE

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090909

Ref country code: FI

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090909

NLV1 Nl: lapsed or annulled due to failure to fulfill the requirements of art. 29p and 29m of the patents act
REG Reference to a national code

Ref country code: ES

Ref legal event code: FG2A

Ref document number: 2333020

Country of ref document: ES

Kind code of ref document: T3

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: SI

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090909

Ref country code: NL

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090909

REG Reference to a national code

Ref country code: PL

Ref legal event code: T3

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: CY

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090909

REG Reference to a national code

Ref country code: IE

Ref legal event code: FD4D

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: IE

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090909

Ref country code: EE

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090909

Ref country code: RO

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090909

Ref country code: PT

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20100111

Ref country code: CZ

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090909

Ref country code: IS

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20100109

BERE Be: lapsed

Owner name: FRANCE TELECOM

Effective date: 20091130

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: SK

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090909

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: MC

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20091130

Ref country code: AT

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090909

REG Reference to a national code

Ref country code: CH

Ref legal event code: PL

PLBE No opposition filed within time limit

Free format text: ORIGINAL CODE: 0009261

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: NO OPPOSITION FILED WITHIN TIME LIMIT

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: DK

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090909

26N No opposition filed

Effective date: 20100610

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: LI

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20091130

Ref country code: GR

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20091210

Ref country code: CH

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20091130

Ref country code: BE

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20091130

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: BG

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20091130

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: LU

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20091124

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: HU

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20100310

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: TR

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090909

REG Reference to a national code

Ref country code: FR

Ref legal event code: PLFP

Year of fee payment: 12

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: GB

Payment date: 20151027

Year of fee payment: 12

Ref country code: DE

Payment date: 20151022

Year of fee payment: 12

Ref country code: IT

Payment date: 20151023

Year of fee payment: 12

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: ES

Payment date: 20151110

Year of fee payment: 12

Ref country code: PL

Payment date: 20151026

Year of fee payment: 12

Ref country code: FR

Payment date: 20151023

Year of fee payment: 12

REG Reference to a national code

Ref country code: DE

Ref legal event code: R119

Ref document number: 602004023115

Country of ref document: DE

GBPC Gb: european patent ceased through non-payment of renewal fee

Effective date: 20161124

REG Reference to a national code

Ref country code: FR

Ref legal event code: ST

Effective date: 20170731

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: IT

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20161124

Ref country code: FR

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20161130

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: DE

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20170601

Ref country code: GB

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20161124

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: PL

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20161124

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: ES

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20161125

REG Reference to a national code

Ref country code: ES

Ref legal event code: FD2A

Effective date: 20181120