Classifications

• • H—ELECTRICITY
  • H03—ELECTRONIC CIRCUITRY
  • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
  • H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
  • H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L19/00—Speech 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
  • G10L19/02—Speech 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
  • G10L19/032—Quantisation or dequantisation of spectral components
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L19/00—Speech 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
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L19/00—Speech 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
  • G10L19/04—Speech 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
  • G10L19/16—Vocoder architecture
  • G10L19/18—Vocoders using multiple modes
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L19/00—Speech 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
  • G10L19/02—Speech 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
  • G10L19/0212—Speech 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

Definitions

• SIGNAL ENVELOPE METHOD FOR DECODING SIGNAL ENVELOPE, AND CODING AND DECODING MODULES
• the invention relates to a method for the binary coding of quantization indices describing an envelope of a signal. It also relates to a binary coding module for implementing said method. The invention further relates to a method and a module for decoding a coded envelope by means of the method and the binary coding module according to the invention.
• the invention finds a particularly advantageous application in the field of transmission and storage of digital signals, such as audio-frequency signals for speech, music, etc.
• the method and the coding module according to the invention are more specially adapted for transform coding of audio-frequency signals.
• waveform coding methods, such as MIC or ADPCM (PCM or ADPCM) coding
• CELP coding Code Excited Linear Prediction
• the invention is essentially concerned with transform coding techniques.
• the ITU-T G.722.1 standard (ITU-T G.722.1 Recommendation, Coding at 24 and 32 kbit / s for hand-free operation in systems with low frame loss, September 1999) discloses a transform encoder for compressing audio, speech or music signals in a bandwidth of 50-7000 Hz, referred to as an extended band, at a sampling rate of 16 kHz and at a rate of 24 or 32 kbit / s.
• Figure 1 gives the associated coding scheme, as provided by the above standard.
• the G.722.1 encoder is based on the MLT transform.
• the MLT transform or Malvar recovery modulated transformation, is a variant of the transformation method known as MDCT (Modified Discrete Cosine Transform).
• the value of the spectral envelope for the j-th subband is defined in the logarithmic domain as:
• the spectral envelope is then quantized in the following manner.
• the values obtained are bounded:
• range of quantization indices that can be represented by the binary coding
• range of quantization indices the differential index range is limited to the range [-
• the range of the G.722.1 encoder is said to be "sufficient" to encode the differences between rmsjndex (j) and rms_index (j-1) if
• the range of the G.722.1 encoder is called "insufficient".
• the quantization index rmsjndex ( ⁇ ) is transmitted in the 5-bit G.722.1 encoder.
• the bits remaining after the coding of the quantization indices of the spectral envelope are then distributed to encode the MDCT coefficients normalized by the quantized envelope. Bit allocation among the subbands is performed by a categorization process which is not related to the present invention and will not be detailed here. The rest of the G.722.1 encoding is also not detailed for the same reason.
• the encoding of the MDCT spectral envelope in the G.722.1 encoder has a number of disadvantages.
• variable length coding can lead to using a very large number of bits for the coding of the spectral envelope in the worst cases.
• the differential coding it has been pointed out above the risk of saturation for certain signals with high spectral disparity, such as isolated sinusoids, the differential coding not working because, in these cases, the range of +/- 36.12 dB can not represent all the dynamics of the differences between the rms values.
• a technical problem to be solved by the object of the present invention is to propose a binary coding method of quantization indices describing an envelope of a signal, comprising a variable length coding step, which would make it possible to minimize the coding length at a limited number of bits, even in the worst cases.
• Another problem to be solved by the invention concerns the management of saturation risks for signals having high effective values, such as sinusoids.
• the solution to this technical problem consists, according to the present invention, in that said method comprising a first variable length coding mode is such that the first coding mode includes envelope saturation detection, and in that said method also includes a second coding mode, performed in parallel with the first coding mode and a selection retaining one of the two coding modes according to a code length criterion and the result of the envelope saturation detection of the first coding mode. coding.
• the method according to the invention is based on the competition of two coding modes including at least one variable length, so as to choose the mode leading to the lowest number of coding bits, especially in the worst case, ie for the least probable rms values.
• the selection is such that the second coding mode is retained if at least one of the following conditions is satisfied: the code length of the second coding mode is shorter than the code length of the first mode coding;
• the invention also relates to a binary coding module of an envelope of a signal, comprising a coding module of a first variable-length mode, which is remarkable in that the coding module of a first mode integrates a signal detector. envelope saturation and in that said coding module also comprises a second coding module of a second mode, arranged in parallel with the coding module of the first mode, and a mode selector able to retain one of the two coding modes , based on a code length criterion and the result from the envelope saturation detector.
• said mode selector is able to generate an indicator of the selected coding mode, in order to indicate to the downstream decoder which decoding mode to apply.
• the invention furthermore relates to a method of decoding an envelope of a signal, said envelope being coded by means of the binary coding method according to the invention, remarkable in that said decoding method comprises a step of detecting said indicator. the coding mode retained and a decoding step according to the coding retained.
• the invention also relates to a module for decoding an envelope of a signal, said envelope being coded by means of the binary coding module according to the invention, said decoding module comprising a decoding module of a first mode with a length variable, remarkable in that said decoding module also comprises a second decoding module of a second mode, arranged in parallel with said decoding module of the first variable-length mode, and a mode detector able to detect said mode of operation indicator. encoding and activate the decoding module corresponding to the detected indicator.
• the invention relates to a program comprising instructions recorded on a computer-readable medium for implementing the steps of the method according to the invention.
• Figure 1 is a diagram of an encoder according to G.722.1.
• Figure 2 shows the schematic of a DMCT type transformation.
• Fig. 3 is a table of the minimum (Min) and maximum (Max) bit length of the codes in each sub-band in a Huffman coding for the encoder of Fig. 1.
• Fig. 4 is a diagram of an encoder hierarchical audio including an MDCT encoder embodying the invention.
• Figure 5 is a detailed diagram of the MDCT encoder of Figure 4.
• FIG. 6 is a diagram of the spectral envelope coding module of the MDCT encoder of FIG. 5.
• FIG. 7 gives a table (a) defining the spectrum cut-out
• MDCT in 18 subbands and a table (b) giving the size of the subbands.
• Fig. 8 is a table giving an example of Huffman codes for representing the differential indices.
• FIG. 9 is a diagram of a hierarchical audio decoder including an MDCT decoder embodying the invention.
• Figure 10 is a detailed diagram of the MDCT decoder of Figure 9.
• Fig. 11 is a schematic diagram of the spectral envelope decoding module of the MDCT decoder of Fig. 10.
• the invention will now be described in the context of a hierarchical audio coder of 8 to 32 kbit / s of a particular type.
• the binary spectral envelope coding methods and modules according to the invention are not limited to this type of encoder and that they can be applied to any binary coding of spectral envelope describing the energy in subbands of a signal.
• the input signal of the wideband hierarchical encoder sampled at 16 kHz, is first broken down into two subbands by Quadrature Mirror Filter (QMF).
• QMF Quadrature Mirror Filter
• the low band from 0 to 4000 Hz is obtained by low-pass filtering 300 and decimation 301, and the high band from 4000 to 8000 Hz by high-pass filtering 302 and decimation 303.
• the filters 300 and 302 are of length 64 and are consistent with those described in the J. Johnston article, ICASSP. flight. 5, pp. 291-294, 1980.
• the low band is pre-processed by a high-pass filter 304 eliminating the components below 50 Hz before CELP 305 coding in 50-4000 Hz narrow band.
• the high-pass filtering takes account of the fact that the enlarged band is defined as the band 50-7000 Hz.
• the CELP 305 narrow-band coding used corresponds to a cascaded CELP coding comprising as a first stage a modified G.729 coding (ITU-T G.729 Recommendation, Coding of Speech at 8 kbps using Conjugate Structure Algebraic Code Excited Linear Prediction (CS-ACELP), March 1996) without pre-processing filter, and, as a second stage, an additional fixed dictionary.
• the error signal of the CELP coding is calculated by the subtractor 306 and perceptually weighted by a filter W NB (Z) 307 to obtain the signal .v / ( , This signal is analyzed by discrete modified cosine transform MDCT 308 to obtain the discrete transformed spectrum Xy.
• the high band is first unfolded spectrally 309 to compensate for the folding due to the QMF filter H 302, and then pretreated by a low pass filter 310 eliminating the components between 7000 and 8000 Hz in the original signal.
• the resulting signal ⁇ h is transformed by MDCT 311 to obtain the discrete transformed spectrum X hi .
• a band extension 312 is made from x hi e ⁇ X hi .
• the MDCT transformation is implemented by means of the algorithm described in the article by P. Duhamel, Y. Mahieux, JP Petit, A fast algorithm for the implementation of filter banks based on the lime domain aliasing cancellation, ICASSP, vol. 3, pp.2209-2212, 1991.
• the MDCT transforms in low band and high band Xy and X M are encoded in the MDCT encoder 313.
• the invention relates more specifically to this encoder.
• the different bitstreams generated by the coding modules 305, 312 and 313 are multiplexed and structured into a hierarchical binary train in the multiplexer 314.
• the coding is done in blocks of samples, or frames, of
• the encoding rate is 8, 12, or between 14 and 32 kbit / s in 2 kbit / s steps.
• the MDCT encoder 313 will be described in detail with reference to FIG.
• the MDCT transforms of the low and high bands are first combined in the merge block 400.
• the coefficients are first combined in the merge block 400.
• the X (O), ..., X (L-I) coefficients of X coming from the MDCT are grouped into K subbands.
• the first subband then comprises the coefficients X (tabis (0)) to X (tabis (I) -I), while the second subband includes the coefficients X (tabis (1)) to X (tabis (2) ) -1), etc.
• K 18, the associated cut is specified in Table (a) of Figure 7.
• the amplitude spectral envelope log_rm.s describing the energy distribution by subbands is calculated 401 then coded by the spectral envelope encoder 402 to obtain the indices rmsjndex.
• the bits are allocated to each sub-band 403 and a spherical vector quantization 404 is applied to the X spectrum.
• the bit allocation corresponds to the method set forth in the article by Y. Mahieux, JP Petit , Transform coding of audio signed at 64 kbit / s, IEEE GLOBECOM, pp. 518-522, vol.1, 1990, and spherical vector quantization is performed as described in International Application PCT / FR04 / 00219.
• the bits resulting from the coding of the spectral envelope and the vector quantization of the MDCT coefficients are processed by the multiplexer 314. We will now more particularly describe the calculation and the spectral envelope coding.
• the spectral envelope log_rm, s in the logarithmic domain is defined for the j-th subband as:
• nb_coeff (j) tabis (j + l) -tabis (j) is the number of coefficients in the j-th subband.
• ⁇ serves to avoid Iog 2 (0).
• the spectral envelope corresponds to the rms in dB of the j-th subband; it is therefore an envelope of amplitude.
• the size of the subbands nb_coeff (j) is given in Table (b) of Figure 7.
• ⁇ 2 "24 , which implies log_rms (j) ⁇ -12.
• the resulting rmsjndex vector contains integer indices between -11 and +20 (32 possible values).
• the low band envelope rms_index_bb is binary form by two coding modules 502 and 503 put into competition, namely a variable length coding module 502 and a fixed length coding module, said "equiprobable", 503.
• the module 502 is a differential Huffman coding module
• the module 503 is a natural binary coding module.
• the differential Huffman coding module 502 comprises two coding steps which are detailed below:
• the indicator satur_bb thus makes it possible to detect spectral envelope saturations by differential Huffman coding of the low band.
• the quantization index rmsjndex ( ⁇ ) has an integer value between -11 and +20. It is directly encoded in 5-bit fixed length binary.
• the Huffman table used is specified in the table in Figure 8.
• bit_cntl_bb the total number of bits, bit_cntl_bb, resulting from this binary conversion of rmsjndex ( ⁇ ) and Huffman coding of the diffjndex quantization indices (j) is variable.
• the mode selection is specified below:
• the differential Huffman mode is chosen.
• the mode selector 504 generates a bit that indicates the selected mode between the differential Huffman and equiprobable modes with the following convention: 0 for differential Huffman mode, 1 for equiprobable mode. This bit is multiplexed to the other bits generated by the envelope encoding.
• the mode selector 504 actuates a flip-flop 505 which makes it possible to multiplex in the multiplexer 314 the bits of the chosen coding mode.
• the high band envelope rms_index_bh undergoes a treatment identical to the processing of rms_index_bb: uniform coding of the first log_rms (0) index on 5 bits by the equiprobable coding module 507 and Huffman coding of the differential indices by the coding module 506.
• the table the Huffman used in the module 506 is identical to that of the module 502.
• the equiprobable encoding 507 is identical to that of the low band 503.
• the mode selector 508 generates a bit which indicates the mode selected between the differential Huffman codings. and equiprobable, and this bit is multiplexed with the bits coming from the flip-flop 509 in the multiplexer 314.
• the hierarchical audio decoder associated with the encoder which has just been described is represented in FIG. 9.
• the bits describing each frame of 20 ms are demultiplexed in the demultiplexer 600.
• the decoding which operates from 8 to 32 kbit / s is presented.
• the bit stream could be truncated at 8, 12, 14 or between 14 and 32 kbit / s in steps of 2 kbit / s.
• the bit stream of the 8 and 12 kbit / s layers is used by the decoder
• CELP 601 to generate a first narrow-band synthesis (0-4000 Hz).
• the portion of the bit stream associated with the 14 kbit / s layer is decoded by the band extension module 602; the signal obtained in the high band (4000-7000 Hz) is transformed by MDCT 603 into a transformed signal X h1 .
• MDCT decoding 604 is explained in Figure 10 and discussed below. It generates from the bit stream associated with the bit rates of 14 to 32 kbit / s a reconstructed spectrum X 10 in low band and a reconstructed spectrum X h ⁇ in high band. These spectra are brought back to time signals x 10 and x h 1 by inverse MDCT in blocks 605 and 606.
• the signal ⁇ ; o is added to the CELP synthesis 608 after inverse perceptual filtering 607, the result is then post-filtered 609.
• the extended band output signal sampled at 16 kHz, is obtained through the synthesis QMF filter bank which includes the 610 and 612 oversampling, low pass and high pass 611 and 613 and addition 614.
• the MDCT decoder 604 is now described with reference to FIG.
• the bits associated with this module are demultiplexed in the demultiplexer 600.
• the spectral envelope is first decoded 701 to obtain the indices rmsjndex as well as the spectral envelope reconstructed in linear scale m ⁇ s_q.
• the decoding module 701 is explained in FIG. 11 and detailed below. In the absence of a binary error and if all the bits describing the spectral envelope are well received, the indices rmsjndex correspond exactly to those which are calculated with the encoder; this property is essential because the allocation of the bits 702 must have the same information to the encoder and the decoder so that the encoder and decoder are compatible. Normalized MDCT coefficients are decoded in block 703.
• the sub-bands that are not received or not coded, because they are too low in energy, are replaced by those of the X h ⁇ spectrum in the substitution module 704.
• the module 705 applies the envelope of amplitude by subbands to the coefficients provided at the module output 704, and the reconstructed spectrum X is separated 706 into a reconstructed spectrum X lo in a low band (0-4000 Hz) and a reconstructed spectrum X 10 in a high band (4000-7000 Hz).
• Figure 11 shows the operation of the decoding of the spectral envelope.
• the bits associated with the spectral envelope are demultiplexed by the demultiplexer 600.
• the decoding begins with the reading in the mode selector 801 of the value of the mode selection bit received from the encoder, differential Huffman mode or equiprobable mode.
• the selector 801 follows the same convention as the coding, ie: 0 for the differential Huffman mode, 1 for the equiprobable mode.
• the value of the bit makes it possible to actuate the flip-flops 802 and 805.
• the mode selection bit is 1
• the decoding indicates to the decoder MDCT that an error has been detected. occurred.
• the bits associated with the low band are decoded in the same way as those of the high band.
• the vector rmsjndex represents the spectral envelope reconstructed on a logarithmic scale in base 2 ; the spectral envelope is converted on a linear scale by the conversion module 812, which performs the following operation:
• the envelope coded by the invention may correspond to the temporal envelope describing the rms value per subframes of a signal instead of a spectral envelope describing the rms value per subframe.
• the fixed-length coding step competing with the differential Huffman coding may be replaced by a variable-length coding step, for example a Hufman coding of the quantization indices instead of that of the differential indices.
• Huffman coding can also be replaced by any other form of lossless coding, such as arithmetic coding, Tunstall coding, and so on.

Landscapes

• Engineering & Computer Science (AREA)
• Physics & Mathematics (AREA)
• Spectroscopy & Molecular Physics (AREA)
• Computational Linguistics (AREA)
• Signal Processing (AREA)
• Health & Medical Sciences (AREA)
• Audiology, Speech & Language Pathology (AREA)
• Human Computer Interaction (AREA)
• Acoustics & Sound (AREA)
• Multimedia (AREA)
• Theoretical Computer Science (AREA)
• Compression, Expansion, Code Conversion, And Decoders (AREA)

Applications Claiming Priority (2)

Publications (1)

Family

ID=37192313

Family Applications (1)

Country Status (9)

Cited By (2)

Families Citing this family (25)

Citations (2)

Family Cites Families (20)

• 2007
  • 2007-02-13 US US12/280,644 patent/US8315880B2/en not_active Expired - Fee Related
  • 2007-02-13 RU RU2008137987/09A patent/RU2420816C2/ru not_active IP Right Cessation
  • 2007-02-13 KR KR1020087023295A patent/KR101364979B1/ko not_active IP Right Cessation
  • 2007-02-13 JP JP2008555850A patent/JP5235684B2/ja not_active Expired - Fee Related
  • 2007-02-13 CN CN2007800066863A patent/CN101390158B/zh not_active Expired - Fee Related
  • 2007-02-13 BR BRPI0708267-3A patent/BRPI0708267A2/pt not_active IP Right Cessation
  • 2007-02-13 WO PCT/FR2007/050781 patent/WO2007096551A2/fr active Application Filing
  • 2007-02-13 EP EP07731606A patent/EP1989707A2/fr not_active Withdrawn
  • 2007-02-13 MX MX2008010836A patent/MX2008010836A/es active IP Right Grant

Patent Citations (2)

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events