CN1458646A - Filter parameter vector quantization and audio coding method via predicting combined quantization model - Google Patents
Filter parameter vector quantization and audio coding method via predicting combined quantization model Download PDFInfo
- Publication number
- CN1458646A CN1458646A CN03122000A CN03122000A CN1458646A CN 1458646 A CN1458646 A CN 1458646A CN 03122000 A CN03122000 A CN 03122000A CN 03122000 A CN03122000 A CN 03122000A CN 1458646 A CN1458646 A CN 1458646A
- Authority
- CN
- China
- Prior art keywords
- coefficient
- parameter
- vector quantization
- coding
- prediction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
The present invention relates to a kind of audio signal encoding method, and is audio signal encoding method of filtering parameter vector quantization and combining quantization model for predicting. The present invention is used to solve the pre-echo problem in coding audio signal, and can lower code rate and control noise effectively. In the present invention, audio signal is filtered to obtain filtering parameters, the filtering parameters are predicted and analyzed to find out predicted model parameters, which are parameter treated in vector quantization method. The said process is similar to increase one new module between the traditional sensing encoder and filter module and quantizing module.
Description
Technical field
The present invention relates to a kind of method, be meant that specifically a kind of filtering parameter vector quantization and combination quantize the audio coding method of model prediction coding audio signal.Mainly be the audio-frequency signal coding method that filtering parameter vector quantization and combination quantize model prediction, it can effectively reduce code check and control noise simultaneously again with solving Pre echoes problem in the audio-frequency signal coding.
Background technology
The design of traditional perceptual audio coder mainly utilizes in the occlusion masking effect simultaneously, it before the effect of time domain can produce noise, shelter and after shelter.This has just caused the extra demand to the perceptual audio coder time domain specification: promptly will accomplish transparent coding quality, quantizing noise also must satisfy the relevant masking threshold of a time domain.But in reality realized, this required concerning perceptual audio coder and is not easy to realize, by the uncertainty principle of time-frequency as can be known, decomposed to do with spectrum signal and quantized and quantization error that coding causes, after composite filter reconstruct, can spread in time domain.To Design of Filter commonly used, MDCT as 1024, this can cause the about 40ms of quantization error propagation, if the energy that signal is stronger in analysis window mainly only concentrates on a very little part, so, before quantizing noise will be diffused into the signal appearance, under the extreme case, in some time period, it in addition can be higher than the energy level of original signal, phenomenon that Here it is so-called " Pre echoes ".According to the characteristic of people's ear, if coding noise words not obviously for a long time before signal begins, Pre echoes can be masked falls.Otherwise coding noise can be perceived by people's ear, sounds that just as before beginning at signal a bit of noise is arranged.For fear of this human factor, the design scrambler will be considered the time domain specification of quantizing noise, guarantee to satisfy the time domain masking condition.The Pre echoes phenomenon is to become the main difficulty that class signal such as castanets signal etc. can't be accomplished low code check soon always.
In the encoding and decoding sound signal, for solving the Pre echoes phenomenon, current more existing technology roughly comprise following several:
Pre echoes control with than position pool technology: bank of filters is covered the fast spectral coefficient that becomes in the section window, increases encoding precision.This can increase greatly fast frame needed bit number of encoding that becomes, and this method can not be used for fixing the code check scrambler.In Moving Picture Experts Group-1, adopt bit pond method, the bit that the frame of use front stays when bit needs peak value, thus keep average constant code rate.If run into the very fast signal of variation, what the size in required bit pond can be big during coding is unreasonable yet in fact.
The adaptive windows handoff technique: what use in many perceptual audio coders is the adaptive windows handoff technique.This method can be according to the characteristic of input signal, the size of adaptive adjustment bank of filters window; Window when stable state part or gradual part adopt length, fast changed signal partly adopts short window to encode.Can reduce the peak value bit demand greatly like this, because need the signal section of high encoding precision to obtain restriction in time.The major defect of this method is that it has increased the scrambler calculated amount, and makes coder structure complicated.Because different window length needs different explanation of psychoacoustic model and normalization, and different frequency bands and noiseless coding structure, window switches the complexity that has increased coder structure significantly.In addition, when adopting overlap-add Structure Filter group, the window switching judging needs extra buffering of scrambler and delay, can cause bigger end-to-end delay.At last, though long window and short window have time-frequency local characteristics preferably, begin window and end window and but can introduce bigger poor efficiency coding.
The gain correction technique: the third suppresses to quantize the technology of noise diffusion, is before the spectral factorization of signal calculated, and signal application dynamic gain modification method is carried out gain control.The dynamic range of input signal is reduced by a product pretreater, enters scrambler then, and like this, " peak value " of signal obtained slackening before coding.The gain corrected parameter also is encoded in the bit stream, in decoding end, carries out an opposite process with this information.Signal spectrum is decomposed the dynamic product correction of previous crops, be equivalent to the analysis window of dynamic correction wave filter group, depend on the shape difference of gain correction function, the frequency response of analysis filter changes according to synthetic window function.In order can both effectively to handle to most of signals, it is highly important that and to allow the independently different piece of applied audio signal spectrum of gain makeover process, this be because, the fast changed signal time generally only accounts for leading on partial spectrum, and, usually do not wish to widen the frequency response of the low-pass filtering passage of bank of filters, because can increase not matching of critical bandwidth like this.
In the mpeg audio coding techniques, the defined scrambler series of MPEG-1 and MPEG-2, layer 1 and layer 2 use the bank of filters of low frequency resolution, do not compose tolerance and handle the Pre echoes phenomenon.Layer 3 has used high frequency resolution bank of filters, carries out Pre echoes in conjunction with Pre echoes control, bit pond and adaptive windows handoff technique and handles.In the coding tools of MPEG-4, increased a kind of " time domain is noise shaped " technology, in structure based on perceptual audio encoders, after signal is transformed into frequency coefficient through bank of filters, according to judgement, if become the class signal soon to signal type, directly frequency coefficient is not quantized, but by increasing a TNS wave filter, be transformed into again in the time domain, in time domain, quantize and encode.The advantage of this technology is to have utilized the characteristics of fast changed signal at frequency domain, can encode to conversion coefficient by forecast method, make quantizing noise similar at the shape and the original signal envelope of time domain, effectively controlled the diffusion problem of quantizing noise, the side information that weak point increases is more, has influenced binary encoding efficient.
Summary of the invention
The present invention proposes a kind of method of filtering parameter vector quantization, and predicts in conjunction with quantitative model, handles the Pre echoes phenomenon.At the deficiency of the noise shaped technology of time domain, the TNS wave filter is improved.Utilize frequency domain and the time domain square Hilbert envelope character of antithesis each other, the filtering parameter that fast changed signal is obtained behind wave filter, carry out residual filtering with linear forecasting technology, residual signals is carried out quantization encoding, simultaneously in order to improve code efficiency, filter parameter is carried out vector quantization, the code word sequence number in the code stream behind the main just vector quantization that increases.The Pre echoes phenomenon of fast changed signal can be effectively solved, the code efficiency of original system can be improved again.
A kind of filtering parameter vector quantization of the present invention and in conjunction with the audio coding method that quantizes model prediction may further comprise the steps the method for audio-frequency signal coding:
(1) input audio signal with the analysis filter analysis, be transformed into the coefficient of frequency domain;
(2) frequency coefficient is carried out linear prediction analysis;
(3) predictive coefficient is carried out vector quantization;
(4) spectral coefficient carries out predictive filtering and obtains the residual error coefficient sequence;
(5) spectral coefficient residual sum predictive coefficient vector quantization result together quantizes;
(6) quantized result and side information are encoded, and obtain output bit flow.
Described quantification to frequency coefficient in the method for audio-frequency signal coding is made up of following steps:
(1), determines the size of predicted frequency scope and window to the output of analysis filter;
(2) selected coefficient is carried out the linear prediction analysis process;
(3) determine whether to use Prediction Parameters vector quantization and spectral coefficient filtering according to prediction gain;
(4) spectral coefficient is done error filtering with the filter coefficient that quantizes reconstruct;
(5) residual sequence is handled with traditional quantification and coding method.
Described in the method for audio-frequency signal coding the predictor parameter vector quantization being made up of following steps:
(1) with the linear prediction analysis model spectral coefficient is predicted;
(2) parameter of predictive analyzer is made up of predictor coefficient, represents with the form of LSF parameter;
(3) the LSF parameter adopts vector quantization method to send;
(4) structure of vector quantizer adopts hierarchical form, minimum two-stage, and the first order quantizes the LSF coefficient, and the second level quantizes the error vector of the first order, by that analogy;
(5) distortion metrics of vector quantization LSF parameter adopts best distortion metrics criterion to determine;
What (6) result of vector quantization obtained is the sequence number of code word in the code book, sends to decoding end as the parameter that quantizes.
A kind of filtering parameter vector quantization of the present invention and in conjunction with the audio coding method that quantizes model prediction may further comprise the steps the method for audio signal decoding:
(1) from bit stream, decodes label information, the spectrum data that obtain corresponding predictor parameter coding and quantize;
(2) the inverse quantization process of predictor parameter obtains predictor coefficient;
(3) the prediction spectral coefficient is carried out the liftering process;
(4) obtain the sound signal of reconstruct with composite filter.
Described inverse quantization to Prediction Parameters in the method for audio signal decoding is made up of following steps:
(1) from bit stream, obtains the code word sequence number of vector quantization;
(2) according to the code word sequence number, hierarchical reconfiguration, interpolation are obtained the LSF parameter that quantizes after the reconstruct;
(3) conversion LSF parameter is to the predictor parameter form;
(4) with the predictor parameter of reconstruct, the spectrum prediction residual is carried out liftering handle.
Filtering parameter vector quantization of the present invention and combination quantize the audio coding method of model prediction, solve Pre echoes problem in the audio-frequency signal coding, effectively reduce code check and control noise simultaneously.
Description of drawings
Fig. 1 is the structured flowchart of audio coding method encoder-side of the present invention.
Fig. 2 is the structured flowchart of audio coding method decoder end of the present invention.
Fig. 3 is an algorithm flow block diagram among the scrambler embodiment of the present invention.
Embodiment
The present invention is described in further detail below in conjunction with drawings and Examples.
To fast change class signal, its optimum coding method can obtain from the following fact: suppose to have real signal x (t), its square Hilbert envelope can be expressed as the contrary Fourier conversion of the autocorrelation function of its frequency spectrum; We know that the power spectral density function of signal can be expressed as the Fourier conversion of the autocorrelation function of its time domain waveform simultaneously.Therefore, we can say that signal is duality relation each other at square Hilbert envelope of time domain and signal at the power spectral density function of frequency domain.
Can obtain from top duality relation, the part bandpass signal in each certain frequency scope, constant if its Hilbert envelope keeps, the auto-correlation of adjacent spectral values also will keep constant so.This just means that the spectral coefficient sequence is the stable state sequence with regard to frequency, thereby can come the spectrum value is handled with predictive coding, effectively represents signal with one group of public predictive coefficient.For fast changed signal such as castanets signal, the measurable coding character of this spectral coefficient can also be understood more intuitively.To a pure sinusoidal signal, its temporal envelope is smooth, and corresponding frequency domain representation is the spectrum of a uneven monopulse shape of extreme.Obviously, to its optimum code mode,, be exactly transition coding technology to the spectrum direct coding if at frequency domain, if in time domain, then be with predictive coding such as LPC linear forecasting technology; The input signal of supposing time domain is a pulse signal, so it frequency domain just have a smooth power spectrum, this signal is handled frequency coefficient no matter be with transition coding for traditional audio coding technology, still handling time domain waveform with forecasting techniques, all is the worst a kind of situation.Yet if adopt opposite technology this moment, promptly time domain waveform can be more effective to signal encoding with transition coding technology or frequency coefficient Forecasting Methodology.
After having set up the duality relation of time domain and frequency domain,, now frequency domain is also set up all about the prediction character of time domain so frequency domain filtering coefficient applied forcasting Methods for Coding.Because antithesis each other between power spectrum density and square Hilbert envelope, the residual energy that the gain of spectral coefficient prediction is promptly encoded depends on the flatness tolerance of squared envelope, therefore to having the signal of fast change characteristic, its squared envelope unevenness degree has determined to have the character of higher coding gain.To the fast changed signal section, predicting strategy can be selected open-loop prediction and two kinds of schemes of closed loop prediction, can obtain coding gain, but on the overall performance of system, the two respectively has different advantages.The error energy of closed loop prediction increases along with prediction gain in decoded signal and reduces, because the error of introducing in the spectral coefficient data has smooth power density spectrum, shape in decoding end quantizing noise time domain is still keeping smooth, be that error energy also is equally distributed in time domain, this means that it can not eliminate the non-occlusion of the time domain of quantizing noise in the quiet section of coding window.If what adopt is the open-loop prediction method, do not reduce at the total error energy of decoding end so, with regard to the global error energy, there is no gain.Yet this moment, error energy was consistent with signal time domain shape in the shape of time domain, can be controlled at quantizing noise under the actual signal, concerning fast changed signal, had just solved the problem of time domain masking.This predictive coding method of spectral coefficient is exactly the noise shaped algorithm of time domain.
In traditional audio coder, can followingly handle in conjunction with the process of improved elimination Pre echoes technology based on perception principle:
Input signal is broken down into one group of spectral coefficient by a high-resolution bank of filters or conversion.Based on the spectral coefficient that calculates,, carry out the coefficient that the Levinson-Durbin algorithm is asked fallout predictor as calculating autocorrelation matrix, recursion to the lpc analysis process of corresponding spectral coefficient operative norm in range of target frequencies.Linear prediction analysis is carried out according to the exponent number of the noise shaped wave filter of high permission, when the prediction gain that calculates surpasses pre-set threshold, just can activate noise shaped predictive coding module.
To the spectral coefficient of range of target frequencies, carry out error filtering with the quantitative prediction device coefficient that calculates, calculate the residual sequence of prediction spectrum.The output of predictive filter has replaced original spectral coefficient, is sent to quantize in the quantizer of standard and encode.By the linear prediction analysis coding principle as can be known, the dynamic range of the residual sequence of spectral coefficient is less than the dynamic range of input spectrum coefficient, therefore in quantifying unit, less bit number can be distributed, perhaps under the condition of same number of bits, improved coding gain can be obtained.
The side information that sends to demoder is expanded by a marker bit, just whether has used improved filtering parameter vector quantization and filter factor predictive coding method.If be labeled as very, range of target frequencies and remaining necessary information arranged in the bit stream so.The parameter of predictor coefficient is the LPC transformation of coefficient form through vector quantization.
The vector quantization of predictor parameter LPC predictive filter is converted into line spectrum pair coefficient of frequency LSF to carry out, and vector quantization adopts the scalar quantization mode to carry out.Current predictor coefficient is quantified as the corresponding code word in the code books at different levels, to line spectrum pair coefficient of frequency LSF vector parameters, select best distortion metrics criterion (as the arest neighbors criterion), searching and computing goes out the codewords indexes of code books at different levels, sends as fallout predictor vector quantization parameter.
The calculating of the quantification of scrambler and coding module and psychoacoustic model can adopt traditional technology based in the audio coder of perception to handle.
In corresponding decoder, need do following processing to each sound channel:
Side information is decoded, setting according to marker bit, vector quantization codewords indexes to predictor coefficient, carry out opposite re-quantization process, obtain the line spectrum pair coefficient of frequency LSF of multi-stage quantization, according to the LSF coefficient, the filtering parameter of the linear prediction error wave filter of quantification is obtained in decoding then.
Residual sequence to the spectral coefficient that quantizes carries out the re-quantization process, obtains residual sequence.Allow residual sequence by the linear prediction inverse filter then, obtain the spectral coefficient sequence of frequency domain.
Follow-up processing procedure adopts traditional treatment method to decode, and finally obtains the sound signal of reconstruct.
The relevant block diagram of a embodiment that below will be by said method illustrates the present invention, but to have more than be to be confined to the implementation procedure shown in these block diagrams in the present invention, and they only are exemplary.According to the principle in the claim proposed by the invention to said process any in form and the change of non-intrinsically safe also belongs within the scope of the invention.
Fig. 1 has provided an embodiment block diagram of encoding scheme of the present invention.Original audio signal is sent to analysis filterbank 101 modules and carries out analysis filtered, has obtained the filter factor of frequency domain, and synchronous signal is also through fast changed signal detection 102 and psychoacoustic analysis 106.Detect in 102 at fast changed signal signal is done the whether detection of fast changed signal, as the judgement of selecting the long or utilization algorithm of the present invention of window.Psychoacoustic analysis 106 is done the psychoacoustic model analysis to signal simultaneously.In spectrum prediction and vector quantization 103 modules, filter factor is used the method for linear prediction analysis at frequency domain then, spectral coefficient is predicted, simultaneously the parameter of fallout predictor is carried out vector quantization.The result of predictor parameter vector quantization directly delivers to synthetic bit stream 105, puts into final bit stream as side information and sends to decoding end.Module quantizes and encodes 104 identical with the effect in the classic method with structure with the function of synthetic bit stream 105.
Fig. 2 has provided an embodiment block diagram of decoding scheme of the present invention.Step bit stream decoding 201 is implemented bit stream decoding process, the various bit stream information when obtaining encoding.The value of the marker bit that Prediction Parameters inverse quantization 202 obtains according to decoding is at first carried out inverse quantization to the predictor coefficient through vector quantization, and is constructed the inverse prediction error-filter of frequency coefficient according to this.Spectral coefficient inverse quantization 203 carries out filtering according to the inverse predictor parameter of reconstruct, carries out the inverse quantization process of spectral coefficient, obtains the sound signal of reconstruct at last by composite filter 204 steps.
Fig. 3 has provided an algorithm flow block diagram of the inventive method among the scrambler embodiment.Obtain spectral coefficient after bank of filters 101 filtering by analysis.The definite size that will carry out the spectral limit of forecast analysis and select employed window in determining long 303 steps of frequency range 302 and definite block type and window.Linear prediction analysis 304 steps are carried out the linear prediction analysis process, obtain the spectral coefficient autocorrelation matrix in the frequency range, obtain predictor parameter with the Levison-Durbin algorithm then, can obtain prediction gain simultaneously.Whether gain is satisfied greater than the condition of given threshold value as the selection predictive filtering according to prediction gain greater than threshold values 305, if gain is not more than threshold value, then skips this module 311 steps and finishes forecasting process, continues to handle with traditional method.If satisfy condition, predictor parameter LP is converted into line spectrum pair coefficient of frequency LSF, because line spectrum pair coefficient of frequency LSF had both had quantized character preferably, a kind of tolerance of direct metric vector quantizing distortion is provided again, the vector quantization of predictor parameter LP has been based on line spectrum pair coefficient of frequency LSF code book.The LSF parameter vector quantizes the multi-stage vector quantization that 307 steps are carried out the LSF parameter, obtains the code word sequence number of Codebook of Vector Quantization.In spectral coefficient predictive filtering 308,, spectral coefficient is carried out the error Filtering Processing with the reconstruct linear prediction analysis parameter after quantizing.Side information quantification 309 and the result who transfers from one department to another to obtain after 310 pairs of above-mentioned steps of quantification module quantize, and finally finish the single treatment process of algorithm.
The abbreviation of literary composition Chinese and English is as shown in the table:
| English abbreviation | English full name | Explanation in Chinese |
| MDCT | ?Modified?Discrete?Cosine ?Trans?form | Revise discrete cosine transform |
| MPEG | ?Moving?Picture?Expert?Group | Mobile picture experts group |
| TNS | ?Time?Noise?Shaping | Time noise typing |
| Hilbert | ?Hilbert?Transform | Hilbert transform |
| Fourier | ?Fourier?Transform | Fourier transform |
| Levinson-Durbin | ?Levinson-Durbin?Algorithm | Paul levinson-moral guest's algorithm |
| LSF | ?Line?Spetrum?Frequency | Line spectral frequencies |
| LPC | ?Linear?Predicting?Coding | Linear predictive coding |
Claims (5)
1, a kind of filtering parameter vector quantization and combination quantize the audio coding method of model prediction, it is characterized in that the method for audio-frequency signal coding be may further comprise the steps:
(1) input audio signal analyzes, is transformed into the coefficient of frequency domain with analysis filter (101);
(2) frequency coefficient is carried out linear prediction analysis (304);
(3) predictive coefficient is carried out vector quantization (307);
(4) spectral coefficient carries out predictive filtering and obtains the residual error coefficient sequence;
(5) spectral coefficient residual sum predictive coefficient vector quantization result together quantizes;
(6) quantized result and side information are encoded, and obtain output bit flow.
2, the method to audio-frequency signal coding according to claim 1 is characterized in that the quantification of described frequency coefficient is made up of following steps:
(1), determines the size of predicted frequency scope (302) and window to the output of analysis filter (101);
(2) selected coefficient is carried out linear prediction analysis (304) process;
(3) determine whether to use Prediction Parameters vector quantization and spectral coefficient filtering according to prediction gain (305);
(4) spectral coefficient is done error filtering with the filter coefficient that quantizes reconstruct;
(5) residual sequence is handled with traditional quantification and coding method.
3, the method to audio-frequency signal coding according to claim 1 is characterized in that described predictor parameter vector quantization is made up of following steps:
(1) with linear prediction analysis (304) model spectral coefficient is predicted;
(2) parameter of predictive analyzer is made up of predictor coefficient, represents with the form of LSF parameter;
(3) the LSF parameter adopts vector quantization method to send;
(4) structure of vector quantizer adopts hierarchical form, minimum two-stage, and the first order quantizes the LSF coefficient, and the second level quantizes the error vector of the first order, by that analogy;
(5) distortion metrics of vector quantization LSF parameter adopts best distortion metrics criterion to determine;
What (6) result of vector quantization obtained is the sequence number of code word in the code book, sends to decoding end as the parameter that quantizes.
4, a kind of filtering parameter vector quantization and combination quantize the audio coding method of model prediction, it is characterized in that the method for audio signal decoding be may further comprise the steps:
(1) from bit stream (201), decodes label information, the spectrum data that obtain corresponding predictor parameter coding and quantize;
(2) the inverse quantization process of predictor parameter (202) obtains predictor coefficient;
(3) the prediction spectral coefficient is carried out the liftering process;
(4) obtain the sound signal of reconstruct with composite filter.
5, the method to audio signal decoding according to claim 4 is characterized in that the inverse quantization of described Prediction Parameters (202) is made up of following steps:
(1) from bit stream, obtains the code word sequence number of vector quantization;
(2) according to the code word sequence number, hierarchical reconfiguration, interpolation are obtained the LSF parameter that quantizes after the reconstruct;
(3) conversion LSF parameter is to the predictor parameter form;
(4) with the predictor parameter of reconstruct, the spectrum prediction residual is carried out liftering handle.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN03122000A CN1458646A (en) | 2003-04-21 | 2003-04-21 | Filter parameter vector quantization and audio coding method via predicting combined quantization model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN03122000A CN1458646A (en) | 2003-04-21 | 2003-04-21 | Filter parameter vector quantization and audio coding method via predicting combined quantization model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1458646A true CN1458646A (en) | 2003-11-26 |
Family
ID=29430282
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN03122000A Pending CN1458646A (en) | 2003-04-21 | 2003-04-21 | Filter parameter vector quantization and audio coding method via predicting combined quantization model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1458646A (en) |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101222793B (en) * | 2008-01-25 | 2010-06-16 | 中兴通讯股份有限公司 | Method and system for regulating bluetooth stereo acoustic quality |
| CN1848960B (en) * | 2005-01-06 | 2011-02-09 | 高通股份有限公司 | Residual coding in compliance with a video standard using non-standardized vector quantization coder |
| CN101436406B (en) * | 2008-12-22 | 2011-08-24 | 西安电子科技大学 | Audio encoder and decoder |
| CN102243876A (en) * | 2010-05-12 | 2011-11-16 | 华为技术有限公司 | Quantization coding method and quantization coding device of prediction residual signal |
| CN101388213B (en) * | 2008-07-03 | 2012-02-22 | 天津大学 | Preecho control method |
| CN101527139B (en) * | 2009-02-16 | 2012-03-28 | 成都九洲电子信息系统股份有限公司 | Audio encoding and decoding method and device thereof |
| CN101421780B (en) * | 2006-04-10 | 2012-07-18 | 高通股份有限公司 | Method and device for encoding and decoding time-varying signal |
| CN102160114B (en) * | 2008-09-17 | 2012-08-29 | 法国电信公司 | Method and device of pre-echo attenuation in a digital audio signal |
| CN101911185B (en) * | 2008-01-16 | 2013-04-03 | 松下电器产业株式会社 | Vector quantizer, vector inverse quantizer, and methods thereof |
| CN101925950B (en) * | 2008-01-04 | 2013-10-02 | 杜比国际公司 | Audio encoder and decoder |
| CN105849801A (en) * | 2013-12-27 | 2016-08-10 | 索尼公司 | Decoding device, method, and program |
| CN106205626A (en) * | 2015-05-06 | 2016-12-07 | 南京青衿信息科技有限公司 | A kind of compensation coding and decoding device for the subspace component being rejected and method |
| CN108107411A (en) * | 2016-11-24 | 2018-06-01 | 北京遥感设备研究所 | Unevenness determines method at the top of a kind of wide-band Chirp pulse signal |
| CN108352166A (en) * | 2015-09-25 | 2018-07-31 | 弗劳恩霍夫应用研究促进协会 | The encoder and method that audio signal is encoded in a manner of so that ambient noise is reduced using linear predictive coding |
| CN110709925A (en) * | 2017-04-10 | 2020-01-17 | 诺基亚技术有限公司 | Audio coding |
| US10546594B2 (en) | 2010-04-13 | 2020-01-28 | Sony Corporation | Signal processing apparatus and signal processing method, encoder and encoding method, decoder and decoding method, and program |
| CN111179952A (en) * | 2014-03-07 | 2020-05-19 | 弗劳恩霍夫应用研究促进协会 | Concept for information coding |
| CN112119457A (en) * | 2018-04-05 | 2020-12-22 | 瑞典爱立信有限公司 | Truncatable predictive coding |
| WO2021136343A1 (en) * | 2019-12-31 | 2021-07-08 | 华为技术有限公司 | Audio signal encoding and decoding method, and encoding and decoding apparatus |
-
2003
- 2003-04-21 CN CN03122000A patent/CN1458646A/en active Pending
Cited By (29)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1848960B (en) * | 2005-01-06 | 2011-02-09 | 高通股份有限公司 | Residual coding in compliance with a video standard using non-standardized vector quantization coder |
| CN101421780B (en) * | 2006-04-10 | 2012-07-18 | 高通股份有限公司 | Method and device for encoding and decoding time-varying signal |
| CN101925950B (en) * | 2008-01-04 | 2013-10-02 | 杜比国际公司 | Audio encoder and decoder |
| CN101911185B (en) * | 2008-01-16 | 2013-04-03 | 松下电器产业株式会社 | Vector quantizer, vector inverse quantizer, and methods thereof |
| CN101222793B (en) * | 2008-01-25 | 2010-06-16 | 中兴通讯股份有限公司 | Method and system for regulating bluetooth stereo acoustic quality |
| CN101388213B (en) * | 2008-07-03 | 2012-02-22 | 天津大学 | Preecho control method |
| CN102160114B (en) * | 2008-09-17 | 2012-08-29 | 法国电信公司 | Method and device of pre-echo attenuation in a digital audio signal |
| CN101436406B (en) * | 2008-12-22 | 2011-08-24 | 西安电子科技大学 | Audio encoder and decoder |
| CN101527139B (en) * | 2009-02-16 | 2012-03-28 | 成都九洲电子信息系统股份有限公司 | Audio encoding and decoding method and device thereof |
| US10546594B2 (en) | 2010-04-13 | 2020-01-28 | Sony Corporation | Signal processing apparatus and signal processing method, encoder and encoding method, decoder and decoding method, and program |
| CN102243876B (en) * | 2010-05-12 | 2013-08-07 | 华为技术有限公司 | Quantization coding method and quantization coding device of prediction residual signal |
| CN102243876A (en) * | 2010-05-12 | 2011-11-16 | 华为技术有限公司 | Quantization coding method and quantization coding device of prediction residual signal |
| CN105849801A (en) * | 2013-12-27 | 2016-08-10 | 索尼公司 | Decoding device, method, and program |
| US11705140B2 (en) | 2013-12-27 | 2023-07-18 | Sony Corporation | Decoding apparatus and method, and program |
| CN105849801B (en) * | 2013-12-27 | 2020-02-14 | 索尼公司 | Decoding device and method, and program |
| US10692511B2 (en) | 2013-12-27 | 2020-06-23 | Sony Corporation | Decoding apparatus and method, and program |
| US11640827B2 (en) | 2014-03-07 | 2023-05-02 | Fraunhofer-Gesellschaft Zur Foerderung Der Angewandten Forschung E.V. | Concept for encoding of information |
| CN111179952B (en) * | 2014-03-07 | 2023-07-18 | 弗劳恩霍夫应用研究促进协会 | Concept for information encoding |
| CN111179952A (en) * | 2014-03-07 | 2020-05-19 | 弗劳恩霍夫应用研究促进协会 | Concept for information coding |
| CN106205626A (en) * | 2015-05-06 | 2016-12-07 | 南京青衿信息科技有限公司 | A kind of compensation coding and decoding device for the subspace component being rejected and method |
| CN106205626B (en) * | 2015-05-06 | 2019-09-24 | 南京青衿信息科技有限公司 | A kind of compensation coding and decoding device and method for the subspace component being rejected |
| CN108352166A (en) * | 2015-09-25 | 2018-07-31 | 弗劳恩霍夫应用研究促进协会 | The encoder and method that audio signal is encoded in a manner of so that ambient noise is reduced using linear predictive coding |
| CN108352166B (en) * | 2015-09-25 | 2022-10-28 | 弗劳恩霍夫应用研究促进协会 | Encoder and method for encoding an audio signal using linear predictive coding |
| CN108107411B (en) * | 2016-11-24 | 2021-08-13 | 北京遥感设备研究所 | Broadband linear frequency modulation pulse signal top unevenness determining method |
| CN108107411A (en) * | 2016-11-24 | 2018-06-01 | 北京遥感设备研究所 | Unevenness determines method at the top of a kind of wide-band Chirp pulse signal |
| CN110709925A (en) * | 2017-04-10 | 2020-01-17 | 诺基亚技术有限公司 | Audio coding |
| CN110709925B (en) * | 2017-04-10 | 2023-09-29 | 诺基亚技术有限公司 | Method and apparatus for audio encoding or decoding |
| CN112119457A (en) * | 2018-04-05 | 2020-12-22 | 瑞典爱立信有限公司 | Truncatable predictive coding |
| WO2021136343A1 (en) * | 2019-12-31 | 2021-07-08 | 华为技术有限公司 | Audio signal encoding and decoding method, and encoding and decoding apparatus |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7693709B2 (en) | Reordering coefficients for waveform coding or decoding | |
| KR101278805B1 (en) | Selectively using multiple entropy models in adaptive coding and decoding | |
| US7684981B2 (en) | Prediction of spectral coefficients in waveform coding and decoding | |
| CN1458646A (en) | Filter parameter vector quantization and audio coding method via predicting combined quantization model | |
| US8332216B2 (en) | System and method for low power stereo perceptual audio coding using adaptive masking threshold | |
| EP2267698B1 (en) | Entropy coding by adapting coding between level and run-length/level modes. | |
| US7433824B2 (en) | Entropy coding by adapting coding between level and run-length/level modes | |
| CN101601087B (en) | Device for encoding and decoding | |
| JP6970789B2 (en) | An audio encoder that encodes an audio signal taking into account the detected peak spectral region in the high frequency band, a method of encoding the audio signal, and a computer program. | |
| US20070106502A1 (en) | Adaptive time/frequency-based audio encoding and decoding apparatuses and methods | |
| KR20080025403A (en) | Frequency segmentation to obtain bands for efficient coding of digital media | |
| KR20080025404A (en) | Modification of codewords in dictionary used for efficient coding of digital media spectral data | |
| US20060122825A1 (en) | Method and apparatus for transforming audio signal, method and apparatus for adaptively encoding audio signal, method and apparatus for inversely transforming audio signal, and method and apparatus for adaptively decoding audio signal | |
| KR20010021226A (en) | A digital acoustic signal coding apparatus, a method of coding a digital acoustic signal, and a recording medium for recording a program of coding the digital acoustic signal | |
| CN1735925A (en) | Reducing scale factor transmission cost for MPEG-2 AAC using a lattice | |
| TWI306336B (en) | Sacle factor based bit shifting in fine granularity scalability audio coding | |
| CN1240050C (en) | Invariant codebook fast search algorithm for speech coding | |
| CN101308657B (en) | Code stream synthesizing method based on advanced audio coder | |
| JPH0761044B2 (en) | Speech coding method | |
| Ravelli et al. | Extending fine-grain scalable audio coding to very low bitrates using overcomplete dictionaries | |
| CA3202969A1 (en) | Method and device for unified time-domain / frequency domain coding of a sound signal | |
| CN101430879B (en) | Multi-speed audio encoding method | |
| Imm et al. | Lossless coding of audio spectral coefficients using selective bitplane coding | |
| CN102226945A (en) | Multi-rate speech audio encoding method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| PP01 | Preservation of patent right | ||
| PP01 | Preservation of patent right |
Effective date of registration: 20051209 Pledge (preservation): Preservation |
|
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20031126 |