Classifications

• • 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
  • G10L19/005—Correction of errors induced by the transmission channel, if related to the coding algorithm
• • 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
  • G10L19/24—Variable rate codecs, e.g. for generating different qualities using a scalable representation such as hierarchical encoding or layered encoding
• • 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
  • G10L19/26—Pre-filtering or post-filtering
  • G10L19/265—Pre-filtering, e.g. high frequency emphasis prior to encoding
• • 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
  • G10L19/04—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

Definitions

• the present invention relates to the field of coding digital signals.
• the coding according to the invention is particularly suitable for the transmission and / or storage of digital signals such as audio-frequency signals (speech, music or other).
• the present invention relates more particularly to the coding of the ADPCM type of waveform coding (for "Pulse Modulation and Adaptive Differential Coding"), or "ADPCM” (for "Adaptive Differential Pulse Code Modulation”) in English, and in particular to the coding of a nested code ADPCM type for delivering scalable bit stream quantization indices.
• ITU-T Recommendation G.722 or ITU-T G.121 The general principle of nested code ADPCM coding / decoding specified by ITU-T Recommendation G.722 or ITU-T G.121 is as described with reference to Figures 1 and 2.
• FIG. 1 thus represents an encoder with nested codes of the ADPCM type.
• a prediction module 110 making it possible to give the prediction of the signal from the previous samples of the quantized error signal Q v is the scaling factor, and reconstituted signal where n is the current moment.
• a subtraction module 120 which subtracts from the input signal x (n) its prediction to obtain a prediction error signal denoted e (ri).
• a quantization module 130 of the error signal which receives as input the error signal e (n) to give quantization indices constituted of B + K bits.
• the quantification module is nested codes that is to say it has a core quantizer B bits and quantizers bit that are embedded on the quantizer heart.
• the bit quantization index at the output of the module of quantification is transmitted via the transmission channel 140 to the decoder as described with reference to FIG. 2.
• the encoder also includes:
• an inverse quantization module 120 for outputting a quantized error signal on B bits
• an adaptation module 170 p quantizers and inverse quantizers to give a level control parameter v (n) still called scale factor, for the next moment;
• the dashed portion referenced 155 represents the low-rate local decoder which contains the predictors 165 and 175 and the inverse quantizer 120.
• This local decoder thus makes it possible to adapt the inverse quantizer to 170 from the low bit rate index I B (n) and adapt the predictors 165 and 175 from the reconstructed low bit rate data.
• the nested code ADPCM decoder of FIG. 2 receives as input the indices from transmission channel 140, version of possibly disturbed by bit errors, and performs inverse quantization by the inverse quantization module 210 (Q B ) of bit rate B bits per sample to obtain the signal
• Q B inverse quantization module 210
• the symbol "'" indicates a value received at the decoder, possibly different from that transmitted by the encoder due to transmission errors.
• the output signal r ' B (n) for B bits will be equal to the sum of the signal prediction and the output of the B-bit inverse quantizer.
• This part 255 of the decoder is identical to the low speed local decoder 155 of FIG.
• the decoder can improve the restored signal.
• the output will be equal to the sum of the prediction and the output of the inverse quantizer 230 to B + 1.
• G.722 nested code ADPCM (hereinafter referred to as G.722) coding broadband signals which are defined with a minimum bandwidth of [50-7000 Hz] and sampled at 16 kHz.
• G.722 coding is an ADPCM coding of each of the two signal subbands [50-4000 Hz] and [4000-7000 Hz] obtained by decomposition of the signal by quadrature mirror filters.
• the low band is coded by a 6, 5 and 4 bit nested code ADPCM coding while the high band is coded by a 2 bit ADPCM coder per sample.
• the total bit rate will be 64, 56 or 48 bit / s depending on the number of bits used for decoding the low band.
• This coding was first used in ISDN (Digital Integrated Services Network) and then in IP network audio coding applications.
• the 8 bits are distributed in the following manner as represented in FIG.
• bits I LS and I 16 may be "stolen" or replaced by data and constitute the improvement bits of the low band. Bits constitute the core bits of the low band.
• a quantized signal frame according to the G.722 standard consists of 8, 7 or 6 bit coded quantization indices.
• the transmission frequency of the index being 8 kHz
• the bit rate will be 64, 56 or 48 kbit / s.
• the quantization noise spectrum will be relatively flat as shown in FIG. 4.
• the spectrum of the signal is also represented in FIG. 4 (here a voiced signal block). This spectrum has a great dynamic ( ⁇ 40dB). It can be seen that in areas of low energy, the noise is very close to the signal and is therefore not necessarily masked. It can then become audible in these regions, essentially in the frequency range [2000-2500 Hz] in Figure 4.
• Coding noise formatting is therefore necessary. Coding noise formatting suitable for nested code coding would also be desirable.
• G.711.1 Wideband embedded extension for G.711
• modulation code "or” G.711.1 A wideband extension to ITU-T G.711 ". Y. Hiwasaki, S. Sasaki, H. Ohmuro, T. Mon, J. Seong, MS Lee, B. Kovesi, S. Ragot, J.-L. Garcia, C. Marro, LM, J. Xu, V. Malenovsky, J. Lapierre, R. Lefebvre. EUSIPCO, Lausanne, 2008.
• This recommendation thus describes coding with coding noise formatting for heart rate coding.
• a perceptual filter for encoding noise shaping is calculated based on past decoded signals from a reverse core quantizer.
• a local heart rate decoder thus makes it possible to calculate the noise shaping filter.
• this noise shaping filter is possible to calculate from decoded heart rate signals.
• a quantizer delivering improvement bits is used at the encoder.
• the decoder receiving the core bit stream and the improvement bits, calculates the coding noise shaping filter in the same way as the coder from the decoded heart rate signal and applies this filter to the output signal of the decoder.
• inverse quantizer of the enhancement bits the shaped high-speed signal being obtained by adding the filtered signal to the decoded heart signal.
• the shaping of the noise thus improves the perceptual quality of the heart rate signal. It offers a limited improvement in quality for improvement bits. Indeed, the formatting of the coding noise is not carried out for the coding of the improvement bits, the input of the quantizer being the same for the quantization of the core as for the improved quantization.
• the decoder must then remove a resulting parasitic component by a matched filtering, when the improvement bits are decoded in addition to the core bits.
• This technique is not used in existing standard scalable decoders of the G.722 or G.727 decoder type. There is therefore a need to improve the quality of the signals regardless of the bit rate while remaining compatible with standard scalable existing decoders.
• the present invention improves the situation.
• the enhancement coding comprises a step of obtaining a coding noise shaping filter used to determine a target signal and that the scalar quantization indices of said enhancement signal, are determined by minimizing the error between a set of possible scalar quantization values and said target signal.
• the signal received at the decoder can therefore be decoded by a standard decoder capable of decoding the heart rate and nested rate signal that does not require noise shaping calculation or correction term.
• the quality of the decoded signal is therefore improved whatever the bit rate available to the decoder.
• an embodiment of the determination of the target signal is such that for a current improvement coding stage, the method comprises the following steps for a current sample:
• the set of possible scalar quantization values and the quantization value of the error signal for the current sample are values designating quantization reconstruction levels, scaled by a parameter. level control calculated with respect to the heart rate quantization indices.
• the values are adapted to the output level of the core coding.
• the values designating quantization reconstruction levels for an improvement stage k are defined by the difference between the values designating the reconstruction levels of the quantization of a nested quantizer with B + k bits, B denotes the number of bits of the core coding and the values designating the quantization reconstruction levels of a nested quantizer at B + k-1 bits, the reconstruction levels of the nested quantizer at B + k bits being defined by doubling the levels. of reconstruction of the nested quantizer at B + k-1 bits.
• the values designating quantization reconstruction levels for the improvement stage k are stored in a memory space and indexed according to the core rate and enhancement quantization indices.
• Output values of the enhancement quantizer are not recalculated for each sampling time by subtracting the output values of the quantizer at B + k bit from those of the quantizer at B + k-1 bits. They are more, for example, arranged 2 by 2 in a table easily indexable by the index of the previous stage.
• the number of possible scalar quantization values varies for each sample.
• the number of coded samples of said improvement signal giving the scalar quantization indices is smaller than the number of samples of the input signal.
• a possible embodiment of the core coding is for example an ADPCM coding using a scalar quantization and a prediction filter.
• Another possible embodiment of the core coding is, for example, a PCM coding.
• the core coding may also comprise a formatting of the coding noise, for example with the following steps for a current sample:
• a shaping of the coding noise of less complexity is thus carried out for the core coding.
• the noise shaping filter is defined by an ARMA filter or a succession of ARMA filters.
• this type of weighting function with a numerator value and a denominator value has the advantage that the denominator value takes signal peaks into account and the numerator value to mitigate these values. peaks thus providing an optimal shaping of the quantization noise.
• the succession of cascade ARMA filters makes it possible to better model the masking filter by modeling components of the envelope of the signal spectrum and of periodicity or quasi-periodicity components.
• the noise shaping filter is decomposed into two ARMA filtering cells in a cascade of decoupled spectral slope and formational pitch.
• each filter is adapted according to the spectral characteristics of the input signal and is therefore suitable for signals having various types of spectral slopes.
• the noise shaping filter (W (z)) used by the enhancement coding is also used by the core coding, thus reducing the implementation complexity.
• the noise shaping filter is calculated as a function of said input signal so as best to adapt to different input signals.
• the noise shaping filter is calculated from a signal locally decoded by the core coding.
• the present invention also relates to a hierarchical coder of a digital audio signal for a current frame of the input signal comprising:
• a core coding stage delivering a scalar quantization index for each sample of the current frame
• At least one improvement coding stage delivering scalar quantization indices for each coded sample of an improvement signal.
• the encoder is such that the enhancement coding stage comprises a module for obtaining a coding noise shaping filter used to determine a target signal and a quantization module delivering the scalar quantization indices of said signal. improvement method by minimizing the error between a set of possible scalar quantization values and said target signal.
• ElIe also relates to a computer program comprising code instructions for implementing the steps of the encoding method according to the invention, when these instructions are executed by a processor.
• the invention relates to a storage means readable by a processor storing a computer program as described.
• FIG. 1 illustrates a coder of the ADPCM type with nested codes according to the state of the art and as previously described;
• FIG. 2 illustrates a decoder of the ADPCM type with nested codes according to the state of the art and as described above;
• FIG. 3 illustrates an example of a quantization index frame of a coder of the ADPCM type with nested codes according to the state of the art and as previously described;
• FIG. 4 represents a spectrum of a signal block with respect to the spectrum of a quantization noise present in an encoder not implementing the present invention
• FIG. 5 represents a block diagram of an embedded coder and a coding method according to a general embodiment of the invention
• FIGS. 6a and 6b show a block diagram of an improvement coding stage and an improvement coding method according to the invention
• FIG. 7 illustrates different configurations of decoders adapted to the decoding of a signal derived from the coding according to the invention
• FIG. 8 represents a block diagram of a first detailed embodiment of an encoder according to the invention and of a coding method according to the invention
• FIG. 9 illustrates an example of calculation of a coding noise for the core coding stage of an encoder according to the invention
• FIG. 10 illustrates a detailed function for calculating a coding noise of FIG. 9
• FIG. 11 illustrates an example of obtaining a set of quantization reconstruction levels according to the coding method of the invention
• FIG. 12 illustrates a representation of the improvement signal according to the coding method of the invention
• FIG. 13 illustrates a flowchart representing the steps of a first embodiment of the calculation of the masking filter for coding according to the invention
• FIG. 14 illustrates a flowchart representing the steps of a second embodiment of the calculation of the masking filter for coding according to the invention
• FIG. 15 represents a block diagram of a second detailed embodiment of an encoder according to the invention and of a coding method according to the invention
• FIG. 16 represents a block diagram of a third detailed embodiment of an encoder according to the invention and of a coding method according to the invention.
• FIG. 17 represents a possible embodiment of an encoder according to the invention.
• This encoder comprises a heart rate encoding stage 500 with B-bit quantization, of the type for example ADPCM encoding such as the G.722 or G.727 or PCM standardized coder for "Coded pulse modulation” or PCM in English (for "Pulse code modulation") such as the standardized encoder G.71 1 and modified according to the outputs of block 520.
• the block referenced 510 represents this core coding stage with shaping of the coding noise, that is to say masking of the noise of the core coding, described in more detail later with reference to FIGS. 8, 15 or 16.
• the invention as presented also relates to the case where no masking of the coding noise in the core part is performed.
• core coder is used in a broad sense in this document.
• core encoder an existing multi-rate encoder such as ITU-T G.722 56 or 64 kbit / s.
• core encoder at 0 kbit / s, that is to say, apply the enhancement coding technique that is the subject of the present invention from the first step of the coding. In the latter case the improvement coding becomes core coding.
• the core coding stage described here with reference to FIG. 5, with noise shaping comprises a filtering module 520 performing the prediction P r (z) from the quantization noise q B (n) and the noise of the noise. filtered quantization q B (n) to provide a prediction signal p
• the filtered quantization noise q B (n) is obtained for example by adding K M partial predictions of the filtered noise to the quantization noise as described later with reference to FIG. 9.
• the heart coding stage receives the signal x (n) as input and outputs the quantization index I B (n), the reconstructed signal r B (n) from I B (n) and the factor d quantizer scale v (n) in the case for example of an ADPCM coding as described with reference to FIG.
• the encoder as represented in FIG. 5 also comprises several improvement coding stages.
• the EAl stage (530), the EAk stage (540) and the EAk2 stage (550) are represented here.
• each enhancement coding stage k inputs the signal x (n), the optimal index. , concatenation of the coding index heart and indexes of the previous improvement stages J or of equivalently all these indices, the signal reconstructed in the previous step the parameters of the masking filter and, if appropriate, the scale factor v (n) in the case of adaptive coding.
• This improvement stage outputs the quantization index of the improvement bits of this coding stage which will be concatenated with the index in the concatenation module 560.
• the improvement stage k also outputs the reconstructed signal .
• the index J k (n) represents here a bit for each sample of index n; however, in the general case J k (n) can represent several bits per sample if the number of possible quantization values is greater than 2.
• stages correspond to bits to be transmitted which will be concatenated to the index so that the resulting index can be decoded by a standard decoder as shown and subsequently described in Figure 7. It is therefore not necessary to change the remote decoder; in addition, no additional information is needed to "inform" the remote decoder of the processing performed at the encoder.
• bits correspond to improvement bits by increasing the bit rate and masking and require an additional decoding module described with reference to FIG. 7.
• the encoder of FIG. 5 also comprises a module 580 for calculating the noise shaping filter or masking filter, from the input signal or the coefficients of the synthesis filters of the encoder as described later with reference to FIGS. and 14. It should be noted that the 580 module could have as input the decoded signal locally rather than the original signal.
• the enhancement coding stages as shown herein provide enhancement bits providing increased signal quality to the decoder, regardless of the rate of the decoded signal and without modifying the decoder and therefore without any additional complexity to the decoder.
• a module Eak of FIG. 5 representing an improvement coding stage k according to one embodiment of the invention is now described with reference to FIG. 6a.
• the improvement coding performed by this coding stage includes a quantization step which outputs an index and a value of quantization minimizing the error between a set of possible quantization values and a target signal determined by using the encoding noise shaping filter.
• the stage k makes it possible to obtain the improvement bit or a group of bits
• It comprises an EAk-I module for subtracting the input signal x (n) from the signal synthesized at the stage for each preceding sample a current frame and the signa of the previous stage for sample n, to give a coding error signal
• a weighted quadratic error criterion will be minimized in the quantization step, so that noise Spectrally shaped is less audible.
• the stage k thus comprises a filter module EAk-2 of the error signal by the weighting function W (z).
• This weighting function can also be used for noise shaping in the heart coding stage.
• the noise shaping filter is here equal to the inverse of the spectral weighting, that is to say:
• This formatting filter is ARMA type ("AutoRegressive Moving Average"). Its transfer function includes a numerator of order N N and a denominator of order Np.
• the EAk-I block serves essentially to define the memories of the non-recursive part of the filter W (z), which correspond to the denominator of H M (z).
• the definition of the memories of the recursive part of W (z) is not shown for the sake of brevity, but it is deduced from and of
• This filtering module outputs a corresponding filtered signal to the target signal.
• spectral weighting The role of spectral weighting is to shape the spectrum of the coding error, which is achieved by minimizing the energy of the weighted error.
• An EAk-3 quantization module performs the quantization step which, based on possible quantization output values, seeks to minimize the weighted error criterion according to the following equation:
• This module EAk-3 makes a quantification of improvement Q enh having as first output the value of the optimal bit J k to concatenate with the index of the previous stage and for the second output, the signal output of the quantizer for the optimal index J k where v (n) represents a scale factor defined by the core coding to adapt the output level of the quantizers.
• the improvement coding stage finally comprises an EAk-4 module for adding the quantized error signal. to the synthesized signal at
• the number of samples to be kept in memory is therefore equal to the number of coefficients of the denominator of the noise shaping filter.
• the index n is incremented by one unit.
• the invention shown in Figure 6a can be achieved by equivalent variants.
• the reconstructed signal can be broken down into a part determined only by the samples already available (past samples samples present from previous stages, memories of the filters) and another part to be determined s opt (n) depending solely on the sample present to be optimized.
• the calculation of the error to be minimized which is the error weighted between the input signal and the reconstructed signal r can also be broken down into two parts.
• W (z) the weighted difference by W (z) between the input sample x (n) and (EAK-I and EAK-2 modules of Figure 6a).
• the value thus obtained is the target signal at the instant n which is reduced to a single target value, it is to be calculated once for each possible value of quantification. Then in the loop For optimization, we must simply find out of all possible scalar quantization values the one that is closest in the sense of the Euclidean distance of this target value.
• Another variant of calculating the target value is to carry out two weighting filterings W (z). The first weights the difference between the input signal and the reconstructed signal of the previous stage The second filter has zero input but these memories are updated using enh (n) v (n). The difference between the outputs of these two filterings gives the same target signal.
• Block 601 gives the coding error of the previous stage
• Block 602 derives one by one all the possible scalar quantization values en h, which are subtracted from block 603 to obtain the coding error of the current stage.
• This error is weighted by the noise shaping filter W (z) (block 604) and minimized (block 605) for controlling the block 602.
• W (z) the noise shaping filter
• block 605 the value locally decoded by the enhancement coding stage is) (block 606).
• the decoding device implemented depends on the signal transmission rate and for example the origin of the signal depending on whether it comes from an ISDN network 710 for example or an IP network 720.
• the restored signal resulting from this decoding will benefit from improved quality through the enhancement coding stages implemented in the encoder.
• an additional decoder 730 then performs inverse quantization in addition to inverse quantifications at B + 1 and B + 2 bits described with reference to Figure 2 to provide the quantized error that added to the prediction signal will give the enhanced broadband signal
• the heart rate coding stage 800 performs an ADPCM type coding with a formatting of coding noise.
• the heart coding stage comprises a module 810 for calculating the prediction of the signal x B (n) produced from the preceding samples of the quantized error signal via the low flow rate index I B (n) of the core layer and the reconstructed signal r as that described with reference to FIG.
• a subtraction module 801 of the prediction to the input signal x (n) is provided to obtain a prediction error signal B
• the core coder also comprises a noise prediction module P r (z) 802 made from previous samples of the quantization noise and filtered noise
• An addition module 803 of the noise prediction / to the prediction error signal d is also provided to obtain an error signal noted e B (n).
• a core quantization module Q B receives the error signal as input to give quantization indices
• the optimal quantification index I B (ri) and the quantified value minimize the error criterion where the values y B (ri) are the reconstructed levels and v (n) the scaling factor from the quantizer adaptation module 804.
• the reconstruction levels of the core quantizer Q B are defined by the table VI of the article of X. Master. "7 kHz audio coding within 64 kbit / s", IEEE Journal on Selected Areas in Communication, Vol.6-2, February 1988 ".
• the quantization index I B (n) of B bits at the output of the quantization module Q B will be multiplexed in the multiplexing module 830 with the bits improvement J t , ..., J ⁇ before being transmitted via the transmission channel 840 to the decoder as described with reference to Figure 7.
• the core coding stage also comprises a module 805 for calculating the quantization noise, the difference between the input of the quantizer and its output a module 806 for calculating the quantization noise filtered by addition of the quantization noise to the quantization noise prediction and a module 807 for calculating the reconstructed signal in adding the signal prediction to the quantized error r
• the adaptation module 804 of the quantizer Q B gives a parameter control level v (n) still called scale factor for the next moment
• the prediction module 810 comprises an adaptation module 811 from the samples of the reconstructed quantized error signal and possibly reconstructed quantized error signal Q filtered by
• the module 850 Cale Mask detailed later is intended to provide the coding noise shaping filter that can be used by both the core coding stage and the improvement coding stages, either from the signal of the coding noise. input, either from the signal locally decoded by the core coding (at the heart rate), or from the prediction filter coefficients calculated in the ADPCM coding by a simplified gradient algorithm.
• the noise shaping filter can be obtained from the coefficients of a prediction filter used for heart rate coding, by adding damping constants and adding a de-rating filter. emphasis.
• the masking module is also possible to use the masking module only in the improvement stages; this alternative is advantageous in the case where the core coding uses few bits per sample, in which case the coding error is not white noise and the signal-to-noise ratio is very low - this situation is reflected in ADPCM coding at 2 bits per sample of the high band (4000-8000 Hz) in G.722, in which case feedback noise shaping is not effective.
• the noise shaping of the core coding corresponding to blocks 802, 803, 805, 806 in Figure 8, is optional.
• the invention as shown in FIG. 16 applies even for a reduced ADPCM core coding to blocks 801, 804, 807, 810, 811, 820.
• FIG. 9 describes in more detail the module 802 performing the calculation of the quantization noise prediction by an ARMA filter (for "AutoRegressive to Adjusted Average” of general expression:
• the filter is represented by cascading ARMA filtering cells 900, 901, 902:
• Quantization noise filtered of Figure 9 at the end of this filter cascade, will be given as a function of the quantization noise by :
• FIG. 10 shows in more detail a module F k (z) 901.
• the quantization noise at the output of this cell k is given by:
• perceived coding is shaped by the filter so less audible.
• an ARMA filter cell can be deduced
• This type of weighting function with a numerator value and a denominator value has the advantage that the denominator value takes into account the signal peaks and the numerator value of attenuating these peaks. thus an optimal shaping of the quantization noise.
• the values of g] and g 2 are such that: 0
• a slight shaping from the fine structure of the signal revealing the periodicities of the signal reduces the perceived quantization noise between the harmonics of the signal.
• the improvement is particularly significant in the case of signals with a fundamental frequency or a relatively high pitch, for example greater than 200 Hz.
• An ARMA cell for long-term noise shaping is given by:
• the coder also comprises several improvement coding stages. Two stages EA1 and EAk are represented here.
• This coding stage comprises an EAk-I module for subtracting from the input signal x ⁇ ) the signal r B + k (n) formed of the signal synthesized at the stage kr B + k (n) for the instants of sampling n - l, ..., n - N D and signal synthesized at the stage k-1 for the moment n, to give a coding error signal e B + k (n).
• An EAk-2 filtering module of e B + k (n) by the weighting function W (z) is also included in the coding stage k.
• This weighting function is equal to the inverse of the masking filter H M (z) given by the core coding such as previously described.
• a filtered signal e w + (rc) is obtained.
• the stage k also comprises an addition module EAk-4 of the quantized error signal to the signal synthesized on the previous stage to give the signal synthesized upstairs
• the filtered error signal is then given in z-transformed notation, by:
• a partial reconstituted signal r B + k (n) is calculated from the signal reconstructed on the previous stage r B + k ⁇ ] (n) and past samples of the signal
• This signal is subtracted from the signal x (n) to give the error signal
• the error signal is filtered by the filter having an ARMA filtering cell W x to give:
• the weighted error criterion is to minimize the squared error for the two values (or N G values if several bits) of possible outputs of the quantizer:
• cascade filtering is carried out.
• the output of the first filter cell will be equal to:
• the improvement bits are obtained bit by bit or group of bits per group of bits in cascade improvement stages.
• the improvement bits according to the invention are calculated in such a way that the improvement signal at the output of the standard decoder is reconstructed with shaping of the quantization noise.
• the values designating levels of quantization reconstructions for an improvement stage k are defined by the difference between the values designating the quantization reconstruction levels of a nested quantizer at B +. k bits, where B denotes the number of bits of the core coding and the values designating the quantization reconstruction levels of a nested quantizer at B + k-1 bits, the reconstruction levels of the nested quantizer at B + k bits being defined by duplication of reconstruction levels of the nested quantizer at B + k-1 bits.
• v (n) representing the scale factor defined by the core coding to adapt the output level of the fixed quantizers.
• the quantization for the quantizers at B, B + 1, ..., B + K bits was performed at one time by identifying in which decision range the quantizer at B + K bits is the value e (n) to quantify.
• the present invention provides a different method. Knowing the quantized value from the quantizer at B + k-1 bits, the quantization of the signal e ⁇ + li (n) at the input of the quantizer is done by minimizing the quantization error and without using the decision thresholds, which advantageously makes it possible to reduce the calculation noise for a fixed-point implantation of the product enh tel than:
• a weighted quadratic error criterion will be minimized so that the spectrally shaped noise is less audible.
• the spectral weighting function used is W (z), which can also be used for noise shaping in the core coding stage.
• the restored heart signal is equal to the sum of the prediction and the output of the inverse quantizer, ie:
• the two reconstructed signals possible at the k-stage are given as a function of the signal actually reconstructed at the k-1 stage by the following equation:
• the signal is defined as being equal to the sum of the two signals: representing the concatenation of all the values
• the filtered error signal () will be equal to:
• the optimal index J k is the one that minimizes the criterion E for realizing scalar quantization from both levels of improvement calculated from the scalar quantizer reconstruction levels to bits and knowing the optimal core index and clues or equivalent
• the quantizer output value for the optimal index is: and the value of the reconstituted signal at time n will be given by:
• FIGS. 13 and 14 illustrate two masking filter calculation embodiments implemented by the module 850 for calculating the masking filter.
• a current signal block that corresponds to the block of the current frame completed by a segment sample of the previous frame is taken into account.
• the signal is pre-processed (pre-emphasis processing) before the calculation in E60 of the correlation coefficients by a filter A 1 (z) whose coefficient or coefficients are either fixed are adapted by linear prediction as described in the patent FR2742568.
• the signal to be analyzed S p (n) is calculated by inverse filtering:
• the signal block is then weighted at E 61 by a Hanning window or a window formed by the concatenation of sub-windows, as known from the state of the art.
• Constants allow you to adjust the spectrum of Masking filter including the first two that regulate the slope of the filter spectrum.
• FIG. 1 A second example of implementation of the masking filter, of low complexity, is illustrated with reference to FIG.
• the principle here is to directly use the synthesis filter of the ARMA filter reconstruction of the decoded signal with a de-emphasis applied by a compensation filter according to the slope of the input signal.
• the ARMA ADPCA predictor has 2 denominator coefficients.
• the compensation filter calculated in E71 will be of the form:
• masking filter Another very simple form of masking filter is that obtained by taking only the denominator of the ARMA predictor with a slight damping: with for example
• This AR partial reconstruction filter of the signal leads to a reduced complexity.
• the coefficients of the filter to be buffered over a signal frame can be fixed or several times per frame to maintain a smoothing effect.
• One way to perform the smoothing is to detect sudden changes in dynamics on the signal at the input of the quantizer or equivalent but minimum complexity directly on the output indices of the quantizer. Between two abrupt variations of indices we obtain an area where the spectral characteristics are less fluctuating, and therefore with ADPCM coefficients more suitable for the purpose of masking.
• the pitch period is calculated, for example, by minimizing the quadratic long-term prediction error at the input e B (n) of the quantizer Q B of FIG. 8, by maximizing the correlation coefficient:
• Pitch is such that:
• the pitch prediction gain Cor f (i) used to generate the masking filters is given by:
• FIG. 15 proposes a second embodiment of an encoder according to the invention.
• This embodiment uses prediction modules in place of the filter modules described with reference to FIG. 8, for both the core coding stage and the improvement coding stages.
• the ADPCM type encoder with heart quantization noise formatting includes a noise prediction module 1505. of reconstruction difference between the input signal x (n) and the low speed synthesized signal r B (n) and an addition module 1510 of the prediction to the input signal x (n),
• the core encoder also comprises a noise prediction calculation module P N , (Z) 1530 made from the preceding quantization noise samples and a subtraction module 1540 of the prediction thus obtained to the prediction error signal to obtain an error signal noted e B (n).
• a core quantization module Q B in 1550 minimizes the quadratic error criterion where the values are the levels reconstructed and v (n) the scaling factor from the adaptation module 1560 of the quantizer.
• the quantization module receives as input the error signal to output quantization indices I B (n) and the quantized signal
• the Q B core quantizer reconstruction levels are defined by Table VI of the X. Master article. "7 kHz audio coding within 64 kbit / s". IEEE Journal on Selected Areas in Communication, Vol.6-2, February 1988 ".
• the quantification index I B (n) of B bits at the output of the quantization module Q B will be multiplexed at 830 with the improvement bits J 1 ,..., J k before being transmitted via the transmission channel. 840 to the decoder as described with reference to Figure 7.
• a quantization noise calculation module 1570 makes the difference between quantizer input and quantizer output
• a module 1580 calculates the reconstructed signal by adding the prediction of the signal to the quantized error
• Q Adapt adaptation module 1560 quantifier gives a level of control parameter v (n) also called scale factor for the next moment.
• An adaptation module P Ada t 811 of the prediction module performs an adaptation from the passed samples of the reconstructed signal and the signal of quantized error reconstructed
• the improvement stage EAk comprises an EAk-10 module for subtracting the signal reconstructed at the stage preceding the input signal x (n) for give the signal
• the signal filtering is carried out by the filter module EAk-11 by the filter to give the filtered signal
• An EAk-12 module for calculating a prediction signal also calculated from the previous quantized samples of the quantized error signal and samples thereof. signal filtered by.
• the EA-k enhancement stage also includes a module EA-kl3 subtraction of the prediction ) to the signal to give a target signal
• the improvement quantification module performs a step of minimizing the quadratic error criterion:
• This module receives as input the signal and provides for first output the quantized signal and for second output the index J k .
• the reconstructed levels of the N + k-bit nested quantizer are computed by splitting the nested output levels of the quantizer to B + k-1 bits. Difference values between these reconstructed levels of the nested quantizer at B + k bits and those of the quantizer at B + k-1 bits are calculated. The difference values are then stored once and for all in processor memory and are indexed by the combination of the core quantization index and quantizer enhancement indices of the previous stages.
• An addition module EAk-15 of the output signal of the quantizer e to the prediction P is also integrated into the improvement stage k as well as a module EAk-16 for adding the signal preceding the signal reconstituted on the previous stage to give the signal restored to the floor
• the previously detailed module Cale Mask 850 provides the masking filter either from the input signal (FIG. 13) or from the coefficients of the ADPCM synthesis filters as explained with reference. in Figure 14.
• the improvement stage k implements the following steps for a current sample: - obtaining a difference signal) by calculating the difference between the input signal x (n) of the hierarchical coding and a reconstructed signal derived from an improvement coding of a preceding improvement coding stage;
• Fig. 15 is given for a masking filter consisting of a single ARMA cell for simple explanation purposes. It is understood that cascading multi-cell ARMA generalization will be performed according to the method described by equations 7 to 17 and FIGS. 9 and 10.
• FIG. 16 represents a third embodiment of the invention with this time a PCM core coding stage.
• the core coding stage 1600 comprises a coding noise shaping via a prediction module P, (z) 1610 calculating the noise prediction. from previous samples of standardized G.71 1 MIC quantization noise and filtered noise
• noise shaping of the core coding corresponding to the blocks 1610, 1620, 1640 and 1650 in Figure 16, is optional.
• the invention as shown in FIG. 16 applies even for a MIC core coding reduced to block 1630.
• a module 1620 performs the addition of the prediction) to the signal input x (n) to obtain an error signal noted e (n).
• a heart quantization module 1630 receives as input the signal e (") to give quantization indices I B (n). The optimal quantification index I B (n) and the quantized value minimize
• PCM Pulse Code Modulation
• the quantization index of B bits at the output of the module of quantification will be concatenated in 830 with the improvement bits J t , ..., J ⁇ before being transmitted via the transmission channel 840 to the standard G.71 1 type decoder.
• a quantization noise calculation module 1640 makes the difference between the input of the quantizer MIC and the quantized output
• a filtered quantization noise calculation module 1650 performs the addition of the quantization noise to the quantization noise prediction Improvement coding consists of improving the quality of the decoded signal by successive addition of quantization bits while keeping optimal formatting of the reconstruction noise for the intermediate rates.
• stage k making it possible to obtain the improvement MIC bit J k or a group of bits is described by the EAk block.
• This improvement coding stage is similar to that described with reference to FIG. 8.
• It comprises an EAk-I subtraction module of the input signal x (n) of the signal r B + k (n) formed of the signal synthesized at the stage k for the samples and the signal synthesized at the k-1 stage for the moment n for give a coding error signal e .
• An EAk-4 addition module of the quantized error signal at signal synthesized in the previous step gives the signal synthesized in step k.
• the signal and the memories of the filter are adapted as described previously for Figures 6 and 8.
• the module 850 calculates the masking filter used for both the core coding and the enhancement coding. It is possible to envisage other versions of the hierarchical coder, represented in FIGS. 8, 15 or 16.
• the number of possible quantization values in the improvement coding varies for each coded sample.
• Improvement coding uses a variable number of bits depending on the samples to be coded.
• the number of enhancement bits allocated can be adapted according to an allocation rule, fixed or variable.
• An example of a variable allocation is given, for example, by the low band enhancement MIC coding in ITU-T G.711.1.
• the allocation algorithm if it is variable must use information available to the remote decoder, so that no additional information needs to be transmitted, which is the case for example in the ITU standard -T G.711.1.
• the number of coded samples of the improvement signal giving the scalar quantization indices (J k (n)) in the enhancement coding may be less than the number of samples of the signal of enhancement. 'Entrance. This variant is deduced from the previous variant when the number of improvement bits allocated is set to zero for some samples.
• an encoder as described according to the first, the second or the third embodiment in the sense of the invention typically comprises a ⁇ P processor cooperating with a memory block BM including a storage and / or working memory, as well as a aforementioned MEM buffer memory as a means for storing, for example, quantization values of the preceding coding stages or a dictionary of quantization reconstruction levels or any other data necessary for the implementation of the coding method as described with reference to FIGS. 6, 8, 15 and 16.
• This encoder receives as input successive frames of the digital signal x (n) and delivers concatenated quantization indices I B + K.
• the memory block BM may comprise a computer program comprising the code instructions for implementing the steps of the method according to the invention when these instructions are executed by a ⁇ P processor of the encoder and in particular a coding at a predetermined rate called core rate, delivering a scalar quantization index for each sample of the current frame and at least one enhancement coding delivering scalar quantization indices for each coded sample of an enhancement signal.
• This enhancement coding comprises a step of obtaining a coding noise shaping filter used to determine a target signal. The scalar quantization indices of said enhancement signal are determined by minimizing the error between a set of possible scalar quantization values and said target signal.
• a means of storage readable by a computer or a processor, integrated or not integrated with the encoder, possibly removable, stores a computer program implementing a coding method according to the invention.
• Figures 8, 15 or 16 may for example illustrate the algorithm of such a computer program.

Landscapes

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

Priority Applications (5)

Applications Claiming Priority (2)

Publications (1)

Family

ID=40661226

Family Applications (1)

Country Status (7)

Cited By (1)

Families Citing this family (24)

Citations (3)

Family Cites Families (72)

• 2008
  • 2008-11-18 FR FR0857839A patent/FR2938688A1/fr not_active Withdrawn
• 2009
  • 2009-11-17 EP EP09768200.9A patent/EP2366177B1/fr active Active
  • 2009-11-17 KR KR1020117014240A patent/KR101339857B1/ko active IP Right Grant
  • 2009-11-17 CN CN2009801546871A patent/CN102282611B/zh active Active
  • 2009-11-17 JP JP2011543801A patent/JP5474088B2/ja active Active
  • 2009-11-17 WO PCT/FR2009/052194 patent/WO2010058117A1/fr active Application Filing
  • 2009-11-17 US US13/129,483 patent/US8965773B2/en active Active

Patent Citations (3)

Non-Patent Citations (5)

Also Published As

Similar Documents

Legal Events