CN102237878B - A Huffman decoding method - Google Patents

Abstract

The invention relates to a method for performing Huffman decoding on coded data on a transmission stream, which comprises the following steps: processing code words in a code book to generate at least one auxiliary comparison table, and storing each generated auxiliary comparison table into a storage device, wherein the code book is a preset code book or is extracted from the transmission stream; and searching a corresponding decoding value of a target code word in the encoded data according to the encoding book and at least the auxiliary comparison table stored in the storage device. By using the auxiliary lookup table for decoding, the search speed can be accelerated and less memory space is required, so that the Huffman decoding method of the invention can be efficiently used in most embedded systems.

Description

A Huffman decoding method

technical field

The present invention relates to a Huffman decoding method, more particularly, to a Huffman decoding method with higher search speed and less memory resources.

Background technique

Huffman codes are a common compression technique used in various standards, such as Moving Picture Experts Group Advanced Audio Coding (MPEGAAC) and Joint Photographic Experts Group (JPEG), And can greatly reduce the size of the audio or video media transport stream. However, because Huffman codes are currently one of the most computationally intensive algorithms for audio and video decoders, Huffman codes will cause a lot of computation for encoders, especially decoders. Big burden. In addition, although many embedded systems have dedicated decoders, the available resources and clock rate are very limited. Therefore, these embedded systems must use these limited resources (such as limited memory space) to process Huffman code.

Several different Huffman decoding methods have been proposed in the past, and these existing Huffman decoding techniques can be roughly divided into two categories: single-bit methods and multi-bit methods. The single-bit method is used in the Huffman tree (Huffman tree) search method and the Joint Image Experts Group standard; in addition, compared to the single-bit method using a bit by bit (bit by bit) search method, the multi-bit method usually uses a look-up table or other search techniques to search for Huffman codes.

Because the single-bit method uses a bit by bit (bit by bit) search method, it will have a very slow search speed. In addition, because the audio stream processed by the current embedded system or the joint image expert group The images have a high bit rate (bit rate) and resolution, plus most embedded systems usually only use low-speed clock signals to reduce power consumption, therefore, as mentioned above, because the single-bit method The search speed is too slow to efficiently process these audio streams or images, so it is not suitable for most embedded systems.

In addition, the multi-bit method was developed to speed up Huffman decoding. Most of the multi-bit methods, especially the look-up table approximation method, decode much faster than the single-bit method. However, the multi-bit method requires more memory space. Therefore, the multi-bit method is not perfect for embedded systems, because most embedded systems are designed with a limited memory size in order to reduce chip area.

Contents of the invention

The technical problem to be solved by the present invention is to provide a Huffman decoding method for the above defects of the prior art, which has a faster search speed and less memory space usage, and can be efficiently used in large Part of the embedded system to solve the above problems.

The technical solution adopted by the present invention to solve the technical problem is: to construct a method for carrying out Huffman decoding on a coded data on a transport stream, which includes: processing the codeword in a codebook to generating at least one auxiliary lookup table, and storing each generated auxiliary lookup table in a storage device, wherein the codebook is a predetermined codebook or extracted from the transport stream; and according to the codebook and At least the auxiliary look-up table stored in the storage device is used to search for a corresponding decoding value of a target codeword in the encoded data.

The method of the present invention, wherein the step of processing the codeword in the coding book includes:

Divide the codewords in the coding book into multiple codeword groups, and define an upper boundary and a lower boundary of each codeword group, wherein the multiple codeword groups correspond to a plurality of different bit forms respectively (bit pattern), and the codewords in the same codeword group all have a part of the same bit pattern; and

Set up the auxiliary comparison table, wherein the auxiliary comparison table records an upper boundary and a lower boundary of each code word group.

In the method of the present invention, the step of searching for the corresponding decoding value of the target codeword includes:

selecting a specific upper boundary and a specific lower boundary from the auxiliary comparison table according to the coded data;

searching for the target codeword from one or more codewords in the codebook with reference to the specified upper boundary and the specified lower boundary; and

The decoded value corresponding to the target codeword and recorded on the codebook is output.

In the method of the present invention, the step of searching for the target codeword from one or more codewords in the codebook with reference to the specific upper boundary and the specific lower boundary includes:

When there is only a specific codeword between the specific upper boundary and the specific lower boundary, the specific codeword in the coding book is used as the target codeword.

In the method of the present invention, the step of searching for the target codeword from one or more codewords in the codebook with reference to the specific upper boundary and the specific lower boundary includes:

When there are multiple specific codewords between the specific upper boundary and the specific lower boundary, a binary search is performed on the multiple specific codewords to find the target codeword.

In the method of the present invention, wherein the codewords in the coding book are divided into these multiple codeword groups, and the steps of defining an upper boundary and a lower boundary of each codeword group include:

For each codeword in the codebook, calculate the number of consecutive bits that are located after a marker bit and have the same value as the marker bit to obtain a first count value, wherein the codewords in the same codeword group have the same first count value; and

determining an upper boundary and a lower boundary for each group of codewords based on a plurality of first count values of the codewords in the codebook; and

The step of selecting the specific upper boundary and the specific lower boundary from the auxiliary look-up table according to the encoded data includes:

calculating the number of consecutive bits in the coded data which are located after a mark bit and have the same value as the mark bit to obtain a second count value; and

The specific upper boundary and the specific lower boundary corresponding to the second count value are selected from the auxiliary lookup table.

In the method of the present invention, wherein the codewords in the coding book are divided into these multiple codeword groups, and the steps of defining an upper boundary and a lower boundary of each codeword group include:

For each codeword in the codebook, calculate the number of consecutive bits that are located after a marker bit and have the same value as the marker bit to obtain a first count value, wherein the codewords in the same codeword group have the same first count value; and

An upper boundary and a lower boundary of each codeword group are determined according to a plurality of first count values of the codewords in the codebook.

The method of the present invention, wherein the step of processing the codeword in the coding book includes:

processing the codewords in the coding book to obtain a first auxiliary comparison table, and processing the codewords in the coding book and the first auxiliary comparison table to obtain a second auxiliary comparison table; and

The step of searching out the corresponding decoding value of the target codeword in the encoded data includes:

The decoded value corresponding to the target codeword in the coded data is searched according to the code book and the first and second auxiliary comparison tables stored in the storage device.

The method of the present invention, wherein the step of processing the codeword in the coding book to obtain the first auxiliary comparison table includes:

The codewords in the codebook are processed to divide the codewords in the codebook into multiple first codeword groups, wherein the multiple first codeword groups correspond to multiple different codeword lengths (codewordlength ), and the codewords in the same first codeword group all have the same codeword length;

Establish the first auxiliary comparison table, wherein the first auxiliary comparison table records a sequence number of the first code word group to which each code word belongs in the coding book, and records a start code of each first code word group in addition word, the index value of each initial codeword in the coding book, and the codeword length of each first codeword group, wherein the codewords in the same first codeword group all have the same sequence number.

The method of the present invention, wherein the step of processing the codeword in the coding book and the first auxiliary comparison table to obtain the second auxiliary comparison table includes:

Referring to the first auxiliary comparison table, the codewords in the coding book are divided into a plurality of second codeword groups, and the upper and lower boundaries of each second codeword group are defined, wherein each second codeword group The upper and lower boundaries are respectively the upper and lower boundaries of the sequence numbers of one or more first codeword groups, and the one or more first codeword groups have at least one identical codeword with the second codeword group; A plurality of second codeword groups correspond to a plurality of different bit patterns respectively, and the codewords in the same codeword group all have a part of the same bit pattern; and

Establish the second auxiliary comparison table, wherein the second auxiliary comparison table records an upper boundary and a lower boundary of each second code word group.

In the method of the present invention, the step of searching for the decoding value corresponding to the target codeword in the encoded data according to the code book and the first and second auxiliary comparison tables stored in the storage device includes: :

selecting a specific upper boundary and a specific lower boundary from the second auxiliary comparison table according to the coded data;

referring to the specific upper boundary and the specific lower boundary to search for a target initial codeword from one or more initial codewords in the first auxiliary look-up table; and

The decoded value corresponding to the target codeword and recorded on the coding book is output according to the target initial codeword.

The method of the present invention, wherein the step of outputting the decoded value corresponding to the target codeword and recorded on the coding book according to the target initial codeword comprises:

Obtain an offset value according to the difference between the encoded data and the target initial codeword; and

The codebook is searched according to an index value of the target initial codeword and the offset value to obtain a Huffman value.

The method of the present invention, wherein referring to the specific upper boundary and the specific lower boundary to search for an initial codeword from one or more initial codewords in the first auxiliary look-up table as the target initial codeword The steps include:

When there is only a specific initial codeword between the specific upper boundary and the specific lower boundary, the specific initial codeword in the first auxiliary look-up table is used as the target initial codeword.

The method of the present invention, wherein referring to the specific upper boundary and the specific lower boundary to search for an initial codeword from one or more initial codewords in the first auxiliary look-up table as the target initial codeword The steps include:

When there are a plurality of specific start code words stored in the first auxiliary look-up table between the specific upper boundary and the specific lower boundary, a binary search is performed on the specific start code words to find the target start code. Start code word.

In the method of the present invention, wherein referring to the first auxiliary comparison table, the codewords in the coding book are divided into a plurality of second codeword groups, and the step of defining the upper and lower boundaries of each second codeword group Contains:

For each codeword in the codebook, calculate the number of consecutive bits that are located after a marker bit and have the same value as the marker bit to obtain a first count value, wherein the same second codeword group the codewords have the same first count value; and

determining an upper boundary and a lower boundary of each second codeword group according to a plurality of first count values of the codewords in the codebook and the serial number of the first codeword group to which each codeword belongs; and

The step of selecting the specific upper boundary and the specific lower boundary from the second auxiliary comparison table according to the encoded data includes:

calculating the number of consecutive bits in the coded data which are located after a mark bit and have the same value as the mark bit to obtain a second count value; and

The specific upper boundary and the specific lower boundary corresponding to the second count value are selected from the second auxiliary look-up table.

In the method of the present invention, wherein referring to the first auxiliary comparison table, the codewords in the coding book are divided into a plurality of second codeword groups, and the step of defining the upper and lower boundaries of each second codeword group Contains:

For each codeword in the codebook, calculate the number of consecutive bits that are located after a marker bit and have the same value as the marker bit to obtain a first count value, wherein the same second codeword group the codewords have the same first count value; and

The upper boundary and the lower boundary of each second codeword group are determined according to a plurality of first count values of the codewords in the coding book and the serial number of the first codeword group to which each codeword belongs.

In the method of the present invention, the transport stream complies with the specifications of Moving Picture Experts Group Advanced Audio Coding (MPEG AAC) or Joint Photographic Experts Group (JPEG).

Implementing the Huffman decoding method of the present invention has the following beneficial effects: by using the auxiliary look-up table for decoding, the search speed can be accelerated and only less memory space is required, so the Huffman decoding method of the present invention can be efficient used in most embedded systems.

Description of drawings

The present invention will be further described below in conjunction with accompanying drawing and embodiment, in the accompanying drawing:

FIG. 1 is a schematic diagram of an advanced audio codec according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of a joint video expert group decoder according to another embodiment of the present invention;

3 is a Huffman decoding method according to a first embodiment of the present invention, wherein the Huffman decoding method is used to decode encoded data in a stream;

FIG. 4 is a schematic diagram of a comparison table according to the Huffman decoding method shown in FIG. 3;

Fig. 5A, Fig. 5B are the detailed flow charts that generate auxiliary comparison table in the process shown in Fig. 3;

Fig. 6 is a detailed flow chart of searching for Huffman values using the auxiliary comparison table generated in Fig. 5A and Fig. 5B;

7 is a Huffman decoding method according to a second embodiment of the present invention, wherein the Huffman decoding method is used to decode encoded data in a stream;

FIG. 8 is a schematic diagram of a comparison table according to the Huffman decoding method shown in FIG. 7;

Fig. 9 is a detailed flowchart of establishing the first auxiliary comparison table, and the first auxiliary comparison table includes an intermediate comparison table MI and three sub-comparison tables MC, BA, and ML;

10A and 10B are detailed flowcharts for generating the second auxiliary comparison table.

[Description of main component symbols]

Detailed ways

FIG. 1 is a schematic diagram of an Advanced Audio Code (AAC) decoder 100 according to an embodiment of the present invention, wherein the decoder 100 can be implemented by hardware or software. The decoder 100 includes a stream deformattor 101, a Huffman decoding module 102, a variable value group calculation and inverse quantization (tuple calculation and dequantization) module 104, a stereo processing module 106 and an inverse filter Module (inverse filtering) 108 , wherein the Huffman decoding module 102 performs a decoding operation according to a Huffman lookup table 112 and an auxiliary lookup table 114 stored in a storage device 110 . In the operation of the decoder 100, after the serial inverse format converter 101 obtains the required bits from an ACOD stream, the obtained bits are sent to the decoder directly related to the ACOD stream. modules (such as the Huffman decoding module 102, the stereo processing module 106, and the inverse filtering module 108). In addition, the Huffman comparison table 112 includes eleven predetermined series of codebooks (codebook) and one scale factor codebook, and the content of the auxiliary comparison table 114 is derived from at least one codebook in the Huffman comparison table 112 . In addition, in the Huffman decoding module 102, many traditional decoding methods, such as tree traverse, can be used to decode Huffman symbols; in the variable value group calculation and inverse quantization module 104, using decoding The final Huffman symbols are used to generate two or four variable value groups, and then the generated variable value groups are dequantized; The single-channel dequantization result of the single-channel to create a two-channel output; finally, the inverse filtering module 108 converts the two-channel output into a time-domain audio output, and the time-domain audio output can be sent to a speaker system for playback.

FIG. 2 is a schematic diagram of a joint video expert group decoder 200 according to another embodiment of the present invention, wherein the decoder 200 can be implemented by hardware or software. The decoder 200 includes a stream deformattor 201, a Huffman decoding module 202, a receiving and expanding module 204, an inverse quantization (dequantization) module 206 and an inverse discrete cosine transform (inverse discrete cosine) transform) module 208, wherein the Huffman decoding module 202 performs a decoding operation according to a Huffman look-up table 212 and an auxiliary look-up table 214 stored in a storage device 210 . The functions of the serial inverse format converter 201 and the Huffman decoding module 202 are the same as those of the serial inverse format converter 101 and the Huffman decoding module 102 shown in FIG. It is not pre-set but carried on a transport stream, so the Huffman lookup table 212 is established through a setting module (not shown). The JEVS codebook has the same characteristics as the AIC codebook, using the methods described in the JEVS standard, so for both the JVG and AIC specifications, Huo The Fman decoding module 202 can use the same decoding method. Afterwards, the receiving and expanding module 204 uses the Huffman symbols decoded by the Huffman decoding module 202 to obtain multiple quantization values from the JPEG stream. In addition, similar to the Huffman comparison table and the Huffman coding book, The JPEG quantization table is also contained in the JPEG stream, and needs to be analyzed and established using the configuration module. Afterwards, the inverse quantization module 206 processes the output of the receiving and expanding module 204 to generate a frequency-domain image block, and finally, the inverse discrete cosine transform module 208 converts the frequency-domain image block into a time-domain image block (time-domain image block).

Please refer to FIG. 3. FIG. 3 is a Huffman decoding method according to a first embodiment of the present invention. An encoded data is decoded. First, the Huffman comparison table 112 or 212 including at least one codebook is extracted from the stream and stored in the storage device 110 or 210. It should be noted that in this embodiment, All codewords in the coding book of Huffman comparison table 112 or 212 are 16 bits, and as shown in the first three columns of Figure 4, according to the highest bit (Most-Significant Bit, MSB) of these codewords to sort, and the sorted codewords are stored in a comparison table CW in the storage device 110 or 210 . The code book shown in Figure 4 contains information such as Huffman value HV (Huffman value), code word length CL (code length), code word CW, sign bit (sign bit), and the first count value. It should be noted that Yes, the codebook shown in Figure 4 is only an example and is not intended to be a limitation of the present invention. The decoding process of the present invention is as follows:

In step 300, the decoding job starts. Next, in step 302, the codewords in the coding book are divided into multiple codeword groups, and the upper and lower boundaries of each codeword group are defined, wherein the multiple codeword groups correspond to multiple different bit pattern, and the codewords in the same codeword group all have a part of the same bit pattern as shown in FIG. 4 . For example, the codewords in the codeword group 2 shown in FIG. 4 have a part of the same bit pattern "001", the codewords in the codeword group 4 have a part of the same bit pattern "10", and The codewords in the codeword group 6 have a part of the same bit pattern "1110"... etc. Therefore, the codewords contained in each codeword group are those that contain its upper and lower boundaries and are located between the upper and lower boundaries all codewords of . In addition, in this embodiment, the most significant bit (MSB) of each codeword in the codebook is regarded as a mark bit, and the continuous The number of bits is calculated as a first count value, and the first count value is used to identify different bit patterns, in other words, each codeword in the same codeword group has the same first count value A count value. For example, the first count value of the code word "00110" in code word group 2 is "1", and the first count value of the code word "11101111" in code word group 6 is "2".

Then, in step 304, set up the auxiliary comparison table 114 or 214, and the auxiliary comparison table records an upper boundary and a lower boundary of each codeword group, in addition, in the present embodiment, each codeword group's The upper and lower boundaries are determined according to the flag bit of the codeword in the codebook and the first count value. For example, for codeword group 2, the auxiliary look-up table stores information of an upper boundary (corresponding to CW=0010) and a lower boundary (corresponding to CW=00111); and for codeword group 4, the auxiliary lookup table stores Information of an upper boundary (corresponding to CW=100000) and a lower boundary (corresponding to CW=101111).

In step 306, according to an encoded data, a specific upper boundary and a specific lower boundary are selected from the auxiliary look-up table. In detail, the Huffman decoding module 102 or 202 receives the encoded data from the serial inverse format converter 101 or 201, and determines which codeword group the encoded data belongs to. In addition, in this embodiment, the Huffman The Mann decoding module 102 or 202 regards the highest bit (MSB) of the encoded data as a mark bit, and calculates the number of consecutive bits that are positioned after the mark bit and have the same value as the mark bit to obtain As a second count value; then, the Huffman decoding module 102 or 202 determines which codeword group the coded data belongs to according to the second count value of the coded data and the flag bit of the coded data. For example, if the second count value of the coded data is "1" and the flag bit is greater than "0", then the coded data belongs to codeword group 2, and the upper and lower boundaries of codeword group 2 serve as the A specific upper boundary and the specific lower boundary; if the second count value of the coded data is "2" and the flag bit is less than "0", then the coded data belongs to codeword group 6, and the codeword group 6 The upper and lower boundaries serve as the specific upper boundary and the specific lower boundary, respectively.

Then, in step 308, by referring to the specific upper boundary and the specific lower boundary, one or more codewords in the codebook are searched to determine a target codeword corresponding to the coded data. Particularly, in this embodiment, if there is only one specific codeword between the specific upper boundary and the specific lower boundary (such as the codeword group 1 shown in FIG. 4 ), then the specific codeword is taken as the target codeword; if there are a plurality of specific codewords (for example codeword group 2～7 shown in Figure 4) between this specific upper boundary and this specific lower boundary, then can utilize binary search method (binary search) to from this Finding a specific code word from a plurality of specific code words as the target code word, that is, determining the target code word by comparing the coded data with one or more specific code words in a code word group in the code book Codeword.

Finally, in step 310, output a Huffman value recorded in the coding book and corresponding to the target codeword as a decoded data. For example, when the target codeword is "100000", the Huffman value "12" is output; the target codeword is "11110000", then the Huffman value "53" is output as the decoded data.

The Huffman decoding method shown in Fig. 3 and Fig. 4 has the following advantages: when a Huffman value corresponding to the coded data needs to be searched and obtained, only the single codeword group shown in Fig. 4 is needed Groups can be searched, and which codeword group the coded data is located in can be efficiently determined through the second count value of the coded data. Compared with the conventional Huffman decoding method that uses all the codeword groups shown in FIG. 4 as the search range, the Huffman decoding method of the present invention can save memory space and speed up the search for Huffman values .

Please refer to FIG. 5A and FIG. 5B . FIG. 5A and FIG. 5B are detailed flowcharts of generating the auxiliary comparison table HT in step 304 of FIG. 3 . As shown in Fig. 5A and Fig. 5B, at step 500, the job starts. Afterwards, in the initialization process of step 501, the codeword index k and a variable pl are reset to "0", and a function 'sc' is used to calculate the codeword located after a marker bit in a codebook And have the number of consecutive bits with the same value as the mark bit, and generate a first count value accordingly, and determine a variable ps according to the first count value of a first code word CW[0], also That is, ps=sc(CW[0]), it should be noted that in many embedded processors, the function 'sc' can be implemented with a single instruction. In addition, a codebook length, that is, the number of codewords in the codebook, is stored as a value n, and a variable pff is set to "1".

The above process will process each codeword in the codebook to generate an auxiliary comparison table HT. For each codeword, in step 502, the function 'sc' will first operate on the codeword to generate a After the first count value s, in step 503, check the mark bit of code word (that is the most significant bit of code word CW[k]), if code word CW[k]＞=0, then flow process enters step 504 to judge the relationship between ps and s, if ps is equal to s, the numerical value of this codeword index increases by 1 (that is step 510, k=k+1) and judges whether the currently processed codeword is the codebook Last code word (step 518); If the code word processed at present is not the last code word in this coding book, flow process returns to step 502 to obtain the first count value of next code word; If ps is not equal to s, An index value i of the auxiliary lookup table HT is initialized as i=s+1 (step 505 ). Afterwards, in steps 506-508, the values HT[2*i] and HT[2*i+1] in the auxiliary comparison table are respectively set to pl and k until the index value i==ps; when the index value i＞ When ps, the process enters step 509 and the variables pl and ps are updated to pl=k and ps=s respectively. Next, increase the value of the codeword index k and determine whether the currently processed codeword is the last codeword in the codebook (steps 510 and 518).

If codeword CW[k]＜0, judge whether variable ppf is equal to 1 (step 511), if variable ppf is equal to 1, then the value of variable ppf is cleared, and index value i is initialized as s (step 514). Similar to the above steps 506-508, the values HT[2*i] and HT[2*i+1] in the auxiliary comparison table HT are set to pl and k respectively, and the index value i is gradually increased in the loop until i= = ps (steps 515-517). After the loop of the index value i ends (i.e. i>ps), the values of the variables pl and ps are updated, and the value of the codeword index k increases by 1 and it is judged whether k is less than n (steps 510 and 518); back to the steps 511, if the variable ppf is not equal to 1, judge whether the variable ps is equal to s (step 512), if the variable ps is not equal to s, then the value HT=[2*ps+32] and HT[2*ps+ 33] is set to pl and k (step 513) respectively, then, the value of variable pl and ps is set to k and s (step 509) respectively, and the value of codeword index k increases by 1 and judges whether k is less than n (steps 510 and 518). If the variable ps is equal to s, the process enters step 510 to increase the value of the codeword index k by 1, and judge whether k is less than n (step 518).

After all the codewords in the codebook have gone through the above process (ie codeword index k==n), the index value i will be reset to ps (step 519 ). If the index value i>=16, then enter step 523 to end the flow process of generating the auxiliary comparison table HT; if the index value i<16, then continue the loop of steps 520-522 until the index value i==16, in addition, loop here , the values HT=[2*i+32] and HT[2*i+33] in the auxiliary lookup table are set as pl and (pi+1) respectively (step 521).

Please refer to FIG. 6 . FIG. 6 is a detailed flow chart of searching for Huffman values in steps 306 - 310 of FIG. 3 , wherein the step of searching for Huffman values uses the auxiliary comparison table HT generated in FIGS. 5A and 5B . As shown in FIG. 6, in step 600, a job starts. In step 601, a 16-bit encoded data is extracted from the AAC or JPEG stream without moving the read pointer, and the 16-bit encoded data is set to an integer value x; then , in step 602, treat the 16-bit coded data as a signed integer (signed integer number), and use the function 'sc' to calculate the The number of consecutive bits with the same value is marked to generate a second count value, and the second count value is stored in a variable s. Afterwards, in step 603, it is judged whether the 16-bit encoded data x is less than 0: if the 16-bit encoded data x is less than 0, then the variable (that is, the second count value) s needs to be updated to (s+16) (step 604), and the updated variable s is used as an index value of the auxiliary comparison table HT in the following steps; if the 16-bit encoded data x is not less than 0, then the variable (that is, the second count value) s is directly used as an auxiliary The index value of the comparison table HT. Afterwards, in step 605 , the values HT=[2*s] and HT[2*s+1] in the auxiliary lookup table HT are retrieved and stored as variables bl and bh respectively.

Afterwards, in step 606, an index value i is calculated according to the formula i=(bh+bl)>>1, wherein ">>" is a bit shift operator (bit shift operator), and the above formula will Integer sum of bh and bl shifted 1 bit to the right. This index value i is used as the index value of three necessary Huffman decoding comparison tables, and these three Huffman decoding comparison tables are: codeword comparison table CW, used to store all codewords in a coding book The codeword length comparison table CL of the number of bits, and the Huffman value comparison table HV shown in the first three columns of FIG. 4 . Afterwards, in step 607, the difference between the variables bh and bl (that is, (bh-bl)) is compared with "1": if the difference between the variables bh and bl is greater than 1, the 16-bit encoded data x Compare with codeword CW[i] (step 608), about step 608, 16-bit encoded data x and codeword CW[i] should be regarded as unsigned integer (unsigned integer number) at this moment, and if x >CW[i], update variable bl to bl=i (step 609); if x<CW[i], update variable bh to bh=i (step 611), after updating variable bl or bh, decode The flow returns to step 606; if x==CW[i], that is, the codeword index that can be used to retrieve the Huffman decoding result has been found, so the flow enters step 612 to obtain a codeword length CL[i ], then in step 613, move the index and return a Huffman value HV[i] as the final decoding result. In addition, returning to the description of step 607, if the difference between the variables bh and bl is not greater than 1, then the current index value i is also the final index value to be searched, so the process will enter the final step, that is, step 612～ 613 to obtain the code word length CL[i], move the index and return the Huffman value HV[i].

Please refer to FIG. 7. FIG. 7 is a Huffman decoding method according to a second embodiment of the present invention, wherein the Huffman decoding method is used for a stream provided by the stream inverse format converter 101, 201 Encoded data is decoded. First, the Huffman comparison table 112 or 212 including at least one codebook is extracted from the stream and stored in the storage device 110 or 210. It should be noted that in this embodiment, All codewords in the coding book of Huffman comparison table 112 or 212 are 16 bits, and as shown in the first three columns of Fig. 8, according to the most significant bit (Most-Significant Bit, MSB) of these codewords The sorted codewords are stored in a comparison table CW in the storage device 110 or 210 . The code book shown in Figure 8 contains information such as Huffman value HV (Huffman value), code word length CL (code length), code word CW, sign bit (sign bit), and the first count value. It should be noted that Yes, the code book shown in FIG. 8 is just an example and not a limitation of the present invention. The decoding process of the present invention is as follows:

In step 700, a decoding job starts. Afterwards, in step 702, the codewords in the codebook are divided into a plurality of first codeword groups, wherein each first codeword group corresponds to a different codeword length, and the same first codeword group The codewords in all have the same codeword length as shown in Figure 8. For example, as shown in Figure 8, the first codeword group 3 includes a plurality of codewords with a codeword length of 5, and the first codeword group 4 includes a plurality of codewords with a codeword length of 6 .

Then, in step 704, set up a first auxiliary comparison table, wherein the first auxiliary comparison table records the serial number of the first code word group to which each code word belongs, wherein, the code in the same first code word group The words all have the same serial number. In addition, the first auxiliary comparison table additionally records a start code word of each first code word group, the index value of the start code word in the coding book, and the first code word The codeword length of the word group. For example, the initial codeword in the first codeword group 3 is "00110" and its sequence number is 2.

Next, in step 706, referring to the first auxiliary comparison table, the codewords in the coding book are divided into a plurality of second codeword groups, and the upper and lower boundaries of each second codeword group are defined, wherein each The upper and lower boundaries of the second codeword group are respectively the upper and lower boundaries of the sequence numbers of one or more first codeword groups, wherein the plurality of second codeword groups correspond to a plurality of different bit patterns (bit pattern) respectively. ), and the codewords in the same codeword group all have a part of the same bit pattern as shown in FIG. 8 . For example, the codewords in the codeword group 2 shown in FIG. 8 have a part of the same bit pattern "001", the codewords in the codeword group 3 have a part of the same bit pattern "01", and Codewords in codeword group 5 have a portion of the same bit pattern "110" . . . and so on. In addition, in this embodiment, the most significant bit (MSB) of each codeword in the coding book is regarded as a mark bit, and the consecutive bits that are located after the mark bit and have the same value as the mark bit The number of bits is calculated as a first count value, and the first count value is used to identify different bit patterns, in other words, each codeword in the same second codeword group has the same first count value.

Then, in step 708, a second auxiliary comparison table is established, and the second auxiliary comparison table records an upper boundary and a lower boundary of each second codeword group. In addition, in this embodiment, each second The upper and lower boundaries of the codeword group are determined according to the first count value of the codeword in the codebook. In addition, the second auxiliary comparison table records which first codeword group the codewords in each second codeword group are located in. For example, the codewords "01000" to "01100" are located in the second codeword group group 3, and is also located in the first code word group 3; Therefore, the second auxiliary lookup table records that the second codeword group 3 is associated with the first codeword group 3 and 4, and the second auxiliary lookup table also records the start code of the associated first codeword group words (ie, the initial codeword "00110" of the first codeword group 3 and the initial codeword "011010" of the first codeword group 4). In practice, the second auxiliary comparison table records that the second codeword group 3 is associated with the first codeword group numbers 2 and 3 (corresponding to the first codeword group 3 and 4 respectively), in other words, In this embodiment, because the upper and lower boundaries of each second codeword group are respectively the upper and lower boundaries of the serial numbers of one or more first codeword groups, therefore, the upper boundary of the second codeword group 3 is the first The sequence number 2 of a codeword group 3 (with the start codeword "00110"), and the lower boundary is the sequence number 3 of the first codeword group 4 (with the start codeword "011010").

In step 710, a specific upper boundary and a specific lower boundary are selected from the second auxiliary look-up table according to a coded data. Specifically, the Huffman decoding module 102 or 202 receives the encoded data from the serial inverse format converter 101 or 201, and determines which second codeword group the encoded data belongs to. In addition, in this embodiment, The Huffman decoding module 102 or 202 regards the highest bit (MSB) of the coded data as a mark bit, and calculates the number of consecutive bits that are located after the mark bit and have the same value as the mark bit Count as a second count value, and then, the Huffman decoding module 102 or 202 determines which second codeword the coded data belongs to according to the second count value of the coded data and the flag bit of the coded data group, and accordingly select a specific upper boundary and a specific lower boundary from the second auxiliary comparison table. The detailed operation of step 710 is similar to step 306, therefore, the description is omitted here.

Next, in step 712, by referring to the specific upper boundary and the specific lower boundary, one or more initial codewords in the first auxiliary look-up table are searched to determine a target initial codeword. Especially, in this embodiment, if there is only one specific start codeword (such as the second codeword group 1 shown in FIG. 8 ) between the specific upper boundary and the specific lower boundary, then the specific start codeword The codeword is used as the target initial codeword; if there are multiple specific initial codewords (such as the second codeword group 3-5 shown in Figure 8) between the specific upper boundary and the specific lower boundary, then A binary search method (binary search) can be used to find a specific start code word from the plurality of specific start code words as the target start code word. In this embodiment, because the encoded data requires a ratio of The number of specific starting codewords for the pair is very small, so the time spent by the binary search method can be greatly reduced. For example, as shown in FIG. 8 , assuming that the upper and lower boundaries of the second codeword group 3 are respectively used as the specific upper boundary and the specific lower boundary, since the specific upper boundary and the specific lower boundary are respectively the first auxiliary comparison Two initial codewords in the table (that is, the initial codeword CW=00110 of the first codeword group 3 and the initial codeword CW=011010 of the first codeword group 4), then the Two start codewords are searched to determine which start codeword is the target start codeword. For example, assuming that the coded data is "011111", the selected target start codeword must be the closest to the coded data and smaller than the coded data among the multiple start codewords in the first auxiliary comparison table. (that is, the initial codeword CW=011010 of the first codeword group 4); for another example, assuming that the coded data is "01100", the selected target initial code The word is the start codeword CW=00110 of the first codeword group 3 .

Afterwards, in step 714, an offset value (offset value) is obtained according to the difference between the coded data and the target initial codeword, and the step of obtaining the offset value will be described in detail later. Finally, in step 716, the codebook is searched according to an index value of the target initial codeword and the offset value to obtain a Huffman value.

Please refer to FIG. 9 , which is a detailed flowchart of establishing the first auxiliary comparison table. The first auxiliary comparison table includes an intermediate comparison table MI and three sub-comparison tables MC, BA, and ML. MI[i] is the serial number of a codeword in a coding book corresponding to a first codeword group, and MC[ i] is an initial codeword of the first codeword group, BA[i] is the index value of the initial codeword in the coding book, and ML[i] is the codeword length of the first codeword group. As shown in FIG. 9, in step 900, a job starts. Afterwards, in step 901, the variables k, i, and pl are first initialized to "0", and a codebook length (that is, the number of codewords in the codebook) is stored as a value of n. Next, in step 902, it is judged whether the variable k (as a code word index at this time) is less than the codebook length n, if the variable k is not less than the codebook length n, then the process enters step 910 and ends the step of establishing the first auxiliary comparison table ; If the variable k is less than the codebook length n, enter step 903 to determine the codeword length CL[k], and store CL[k] as a value of m. Then, in step 904, compare code word length m and variable pl, if variable pl is equal to code word length m, then flow process enters step 908 so that MI[k] is set as (i-1); If variable pl is not equal to code word length m, the process enters step 905 to update three sub-comparison tables (MC[i]=CW[k], BA[i]=k, ML[i]=CL[k]) respectively. Afterwards, in steps 906 and 907, the variable pl is set to the codeword length m, and the index i is updated to (i+1). After the variables and the content of the sub-lookup table are updated (steps 905-907), the process goes to step 908 to set MI[k] to (i-1). Finally, in step 909 , the variable k is incremented by “1” (ie k=k+1), and then the process returns to step 902 .

Please refer to FIG. 10A and FIG. 10B . FIG. 10A and FIG. 10B are detailed flowcharts of generating the second auxiliary comparison table HT. The processes in Fig. 10A and Fig. 10B are similar to those in Fig. 5A and Fig. 5B, and the main difference between Fig. 5 and Fig. 10 is that the intermediate comparison table MI generated in the process shown in Fig. 9 is used to set the The second auxiliary comparison table HT generated by the process shown in 10 . In step 1000, the job starts. Next, in step 1001, the variable pl and the codeword index k are initialized to "0", and the variable ppf is set to "1", in addition, the variable ps is initialized to a first count value of the codeword CW[0] , and a codebook length (that is, the number of codewords in the codebook) is stored as an n value, where the first count value of the codeword CW[0] is located in a marker bit in the codeword CW[0] The number of consecutive bits that follow and have the same value as the flag bit.

In step 1002, use the function 'sc' to calculate the first count value of each codeword, and store the calculated first count value in a variable s. Then, in step 1003, it is judged whether the codeword CW[k] (at this moment, the codeword CW[k] is regarded as a 16-bit signed integer (signed integer)) is less than 0, if CW[k]＞= 0 and ps==s (step 1004), then the numerical value of codeword index k increases " 1 " (that is step 1010, k=k+1) and judges whether all have been processed (step 1018); If CW[k ]>=0 and ps! =s, the process goes to step 2005 to initialize a variable i to (s+1). After that, in step 1006, the values HT[2*i] and HT[2*i+1] in the second auxiliary look-up table are set as MI[pl] and MI[k]+1 respectively. Then, in steps 1007 and 1008, the value of variable i is increased by 1 and it is judged whether variable i is less than or equal to variable ps. If I<=ps, the process returns to step 1006; if i>ps, the process enters step 1009 to change variable p1 and ps are respectively set to k and s.

Then, if CW[k]<0 and ppf==1 (step 1011 ), then the flow goes to step 1014 to set the variable ppf to 0 and the variable i to s. Next, in step 1015, the values HT[2*i] and HT[2*i+1] in the second auxiliary lookup table are set as MI[pl] and MI[k]+1 respectively. Then, in steps 1016 and 1017, the value of the variable i is increased by 1, and the variable i is compared with the variable ps. If i<=ps, the process returns to step 1015; if i>ps, the process enters step 1009 to The variables pl and ps are set to k and s, respectively. In addition, if CW[k]<0 and ppf! =1, then the process enters step 1012 to determine whether the variable ps is not equal to variable s, if ps is not equal to s, then the process enters steps 1013 and 1009 to use the values HT[2*ps+32] and HT[2*ps+33] is set to MI[pl] and MI[k]+1, respectively, and the variables pl and ps are set to k and s, respectively. If it is determined in step 1012 that ps is equal to s, then the process proceeds to step 1010 to increase the value of the codeword index k by 1.

Afterwards, flow process enters step 1018 to judge whether all codewords have been processed, if not, flow process returns to step 1002; Otherwise, if all codewords have been processed, then flow process enters step 1019 to set variable i is the value of the variable ps. Then, in step 1020, it is judged whether the variable i is less than 16, if the variable i is not less than 16, then enter step 1023 to end the process of establishing the second auxiliary comparison table; if the variable i is less than 16, then the process enters step 1021 to set The values HT[2*i+32] and HT[2*i+33] in the second auxiliary lookup table are respectively set to MI[pl] and MI[pl]+1. Afterwards, in step 1022 , the variable i is incremented by 1, and the process returns to step 1020 . The output of the process shown in Figure 10 is the second auxiliary comparison table HT, and the second auxiliary comparison table HT is used together with the three sub-comparison tables MC, BA, and ML in the first auxiliary comparison table to decode the Huffman value .

The detailed process of searching for the Huffman value in steps 710-716 of FIG. 7 is described below, wherein the step of searching for the Huffman value uses the three sub-comparison tables MC, MC, BA, ML, and the second auxiliary comparison table generated in Fig. 10A and Fig. 10B. The specific process is as follows:

First the decoding job starts.

Then, a 16-bit encoded data is extracted from the AAC or JPEG stream without moving the read pointer, and the 16-bit encoded data is set to an integer value x.

Next, treat the 16-bit encoded data as a signed integer (signed integer number), and use the function 'sc' to calculate the 16-bit encoded data that is located after a mark bit and has the same The number of consecutive bits of the value is used to generate a second count value, and the second count value is stored in a variable s.

Next, judge whether x is less than 0: if x is less than 0, the variable s is updated to (s+16), and the values HT[2*s] and HT[2*s+1] in the second auxiliary comparison table are supplemented are extracted and stored as variables bl and bh, respectively.

Next, an index value i is calculated according to the formula i=(bh+bl)>>1, wherein ">>" is a bit shift operator.

Next, compare the difference between the variables bh and bl (ie (bh-bl)) with 1, there are two situations:

First, if the difference between the variable bh and bl is greater than 1, compare the variable x with the codeword MC[i], where MC[i] is an unsigned integer (unsigned integer number) generated in the flow process of Figure 9, If x>MC[i], set the variable bl to i; if x is not greater than MC[i], then judge whether x is equal to MC[i]; if x is not equal to MC[i], set the variable bh to i. After the variable bl or bh is updated, the process returns to the above-mentioned "calculate an index value i according to the formula i=(bh+bl)>>1". In addition, if x==MX[i], then (x-MC[i]) is shifted to the right by (16-ML[i]) bits to generate a variable off, where MC[i] is a code word, and ML[i] is the codeword length of the codeword MC[i], in other words, that is to calculate a binary difference (binary difference) between the coded data and the target initial codeword, and The binary difference (ie (x-MC[i])) is shifted right by (16-ML[i]) bits to generate the variable off. Next, an index value ii is calculated according to the formula ii=BA[i]+off, wherein BA[i] is the index value of the initial codeword generated in the process of FIG. 9 in the coding book. Finally, the index bit is shifted to ML[i] to obtain a Huffman value HV[ii] as a decoding result, and then the decoding process ends.

In the second type, if the difference between the variable bh and bl is not greater than 1, the process performed when x==MX[i] is the same, that is, (x-MC[i]) is shifted to the right by (16-ML [i]) bits to generate a variable off, where MC[i] is a codeword, and ML[i] is the codeword length of the codeword MC[i], in other words, the coded data is calculated A binary difference (binary difference) with the target start codeword, and shift the binary difference (ie (x-MC[i])) to the right by (16-ML[i]) bit to generate the variable off. Next, an index value ii is calculated according to the formula ii=BA[i]+off, wherein BA[i] is the index value of the initial codeword generated in the process of FIG. 9 in the coding book. Finally, the index bit is shifted to ML[i] to obtain a Huffman value HV[ii] as a decoding result, and then the decoding process ends.

In addition, in the AAC decoder 100 shown in FIG. 1 and the JPEG decoder 200 shown in FIG. 2, the Huffman lookup tables 112, 212 and the auxiliary lookup tables 114, 214 are stored in the same storage device 110, 210, However, in other embodiments of the present invention, the Huffman lookup table and the auxiliary lookup table may be respectively established in different storage devices, and such design changes shall fall within the scope of the present invention.

It should be noted that the Huffman decoding method shown in FIG. 3 and FIG. 7 is applied to the AAC decoder 100 shown in FIG. 1 and the JPEG decoder 200 shown in FIG. 2 , however, in other embodiments of the present invention Among them, the Huffman decoding method of the present invention can also be applied to other decoding devices that need to use Huffman decoding.

Summarizing the present invention briefly, the method for performing Huffman decoding on a coded data on a transport stream of the present invention includes: processing codewords in a codebook to generate at least one auxiliary look-up table, and converting the Each auxiliary look-up table generated is stored in a storage device, wherein the code book is retrieved from the transport stream; and the code is searched based on the code book and at least the auxiliary look-up table stored in the storage device The corresponding decoded value of a target codeword in the data. By using the auxiliary look-up table for decoding, the search speed can be accelerated and less memory space is required. Therefore, the Huffman decoding method of the present invention can be efficiently used in most embedded systems.

The above descriptions are only preferred embodiments of the present invention, and all equivalent changes and modifications made according to the scope of the patent application of the present invention shall fall within the scope of the present invention.

Claims (13)

1. A method for carrying out Huffman decoding to an encoded data on a transport stream, characterized in that it comprises: processing codewords in a codebook to generate at least one auxiliary lookup table, and storing each generated auxiliary lookup table in a storage device, wherein the codebook is a predetermined codebook or from the transport stream taken out of; and Searching for a corresponding decoding value of a target codeword in the encoded data according to the encoding book and at least the auxiliary comparison table stored in the storage device; Wherein the step of processing the codeword in the coding book includes: Divide the codewords in the coding book into multiple codeword groups, and define an upper boundary and a lower boundary of each codeword group, wherein the multiple codeword groups correspond to a plurality of different bit forms respectively , and the codewords in the same codeword group all have a part of the same bit pattern; and Set up the auxiliary comparison table, wherein the auxiliary comparison table records an upper boundary and a lower boundary of each code word group; Wherein the step of finding out the corresponding decoding value of the target codeword includes: selecting a specific upper boundary and a specific lower boundary from the auxiliary comparison table according to the coded data; searching for the target codeword from one or more codewords in the codebook with reference to the specified upper boundary and the specified lower boundary; and Outputting the decoded value corresponding to the target codeword and recorded on the coding book; Wherein the codewords in the coding book are divided into these multiple codeword groups, and the steps of defining an upper boundary and a lower boundary of each codeword group include: For each codeword in the codebook, calculate the number of consecutive bits that are located after a marker bit and have the same value as the marker bit to obtain a first count value, wherein the codewords in the same codeword group have the same first count value; and determining an upper boundary and a lower boundary for each group of codewords based on a plurality of first count values of the codewords in the codebook; and The step of selecting the specific upper boundary and the specific lower boundary from the auxiliary look-up table according to the encoded data includes: calculating the number of consecutive bits in the coded data which are located after a mark bit and have the same value as the mark bit to obtain a second count value; and The specific upper boundary and the specific lower boundary corresponding to the second count value are selected from the auxiliary lookup table. 2. The method according to claim 1, wherein the step of searching for the target codeword from one or more codewords in the codebook with reference to the specific upper boundary and the specific lower boundary comprises: When there is only a specific codeword between the specific upper boundary and the specific lower boundary, the specific codeword in the coding book is used as the target codeword. 3. The method according to claim 1, wherein the step of searching for the target codeword from one or more codewords in the codebook with reference to the specific upper boundary and the specific lower boundary comprises: When there are multiple specific codewords between the specific upper boundary and the specific lower boundary, a binary search is performed on the multiple specific codewords to find the target codeword. 4. The method according to claim 1, wherein the step of processing the codeword in the coding book comprises: processing the codewords in the coding book to obtain a first auxiliary comparison table, and processing the codewords in the coding book and the first auxiliary comparison table to obtain a second auxiliary comparison table; and The step of searching out the corresponding decoding value of the target codeword in the encoded data includes: The decoded value corresponding to the target codeword in the coded data is searched according to the code book and the first and second auxiliary comparison tables stored in the storage device. 5. The method according to claim 4, wherein the step of processing the codeword in the coding book to obtain the first auxiliary comparison table comprises: processing the codewords in the codebook to divide the codewords in the codebook into a plurality of first codeword groups, wherein the plurality of first codeword groups respectively correspond to a plurality of different codeword lengths, and The codewords in the same first codeword group all have the same codeword length; Establish the first auxiliary comparison table, wherein the first auxiliary comparison table records a sequence number of the first code word group to which each code word belongs in the coding book, and records a start code of each first code word group in addition word, the index value of each initial codeword in the coding book, and the codeword length of each first codeword group, wherein the codewords in the same first codeword group all have the same sequence number. 6. The method according to claim 5, wherein the step of processing the codeword in the coding book and the first auxiliary comparison table to obtain the second auxiliary comparison table includes: Referring to the first auxiliary comparison table, the codewords in the coding book are divided into a plurality of second codeword groups, and the upper and lower boundaries of each second codeword group are defined, wherein each second codeword group The upper and lower boundaries are respectively the upper and lower boundaries of the sequence numbers of one or more first codeword groups, and the one or more first codeword groups have at least one identical codeword with the second codeword group; A plurality of second codeword groups correspond to a plurality of different bit patterns, and the codewords in the same codeword group all have a part of the same bit pattern; and Establish the second auxiliary comparison table, wherein the second auxiliary comparison table records an upper boundary and a lower boundary of each second code word group. 7. The method according to claim 6, wherein the codeword corresponding to the target codeword in the coded data is searched according to the code book and the first and second auxiliary comparison tables stored in the storage device The steps to decode the value include: selecting a specific upper boundary and a specific lower boundary from the second auxiliary comparison table according to the coded data; referring to the specific upper boundary and the specific lower boundary to search for a target initial codeword from one or more initial codewords in the first auxiliary look-up table; and Outputting the decoded value corresponding to the target code word and recorded on the coding book according to the target initial code word. 8. The method according to claim 7, wherein the step of outputting the decoded value corresponding to the target codeword and recorded on the coding book according to the target initial codeword comprises: Obtain an offset value according to the difference between the encoded data and the target initial codeword; and The codebook is searched according to an index value of the target initial codeword and the offset value to obtain a Huffman value. 9. The method according to claim 8, wherein referring to the specific upper boundary and the specific lower boundary to search for an initial codeword from one or more initial codewords in the first auxiliary look-up table The steps of using as the target starting codeword include: When there is only a specific initial codeword between the specific upper boundary and the specific lower boundary, the specific initial codeword in the first auxiliary look-up table is used as the target initial codeword. 10. The method according to claim 7, wherein referring to the specific upper boundary and the specific lower boundary to search for an initial codeword from one or more initial codewords in the first auxiliary look-up table The steps of using as the target starting codeword include: When there are a plurality of specific start code words stored in the first auxiliary look-up table between the specific upper boundary and the specific lower boundary, a binary search is performed on the specific start code words to find the target start code. Start code word. 11. The method according to claim 7, wherein referring to the first auxiliary comparison table, the codewords in the coding book are divided into a plurality of second codeword groups, and each second codeword is defined The steps of the upper and lower boundaries of the word group include: For each codeword in the codebook, calculate the number of consecutive bits that are located after a marker bit and have the same value as the marker bit to obtain a first count value, wherein the same second codeword group the codewords have the same first count value; and Determine the upper boundary and lower boundary of each second codeword group according to a plurality of first count values of the codewords in the codebook and the sequence number of the first codeword group to which each codeword belongs; and The step of selecting the specific upper boundary and the specific lower boundary from the second auxiliary comparison table according to the encoded data includes: calculating the number of consecutive bits in the coded data which are located after a mark bit and have the same value as the mark bit to obtain a second count value; and The specific upper boundary and the specific lower boundary corresponding to the second count value are selected from the second auxiliary look-up table. 12. The method according to claim 6, wherein, referring to the first auxiliary comparison table, the codewords in the coding book are divided into a plurality of second codeword groups, and each second codeword is defined The steps of the upper and lower boundaries of the word group include: For each codeword in the codebook, calculate the number of consecutive bits that are located after a marker bit and have the same value as the marker bit to obtain a first count value, wherein the same second codeword group the codewords have the same first count value; and The upper boundary and the lower boundary of each second codeword group are determined according to a plurality of first count values of the codewords in the coding book and the serial number of the first codeword group to which each codeword belongs. 13. The method as claimed in claim 1, wherein the transport stream complies with the specifications of the Moving Video Experts Group Advanced Audio Coding or the Joint Video Experts Group.