CN1256001A - Method and device for coding lag parameter and code book preparing method - Google Patents
Method and device for coding lag parameter and code book preparing method Download PDFInfo
- Publication number
- CN1256001A CN1256001A CN99800072A CN99800072A CN1256001A CN 1256001 A CN1256001 A CN 1256001A CN 99800072 A CN99800072 A CN 99800072A CN 99800072 A CN99800072 A CN 99800072A CN 1256001 A CN1256001 A CN 1256001A
- Authority
- CN
- China
- Prior art keywords
- coding
- lag parameter
- code book
- code
- parameter
- 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
- 238000000034 method Methods 0.000 title claims description 52
- 238000013459 approach Methods 0.000 claims description 17
- 238000004891 communication Methods 0.000 claims description 11
- 238000004519 manufacturing process Methods 0.000 claims description 11
- 238000005259 measurement Methods 0.000 claims description 4
- 230000015556 catabolic process Effects 0.000 abstract 1
- 238000006731 degradation reaction Methods 0.000 abstract 1
- 230000035807 sensation Effects 0.000 description 10
- 230000006866 deterioration Effects 0.000 description 9
- 239000013598 vector Substances 0.000 description 9
- 230000003044 adaptive effect Effects 0.000 description 8
- 230000005284 excitation Effects 0.000 description 8
- 239000002131 composite material Substances 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000005086 pumping Methods 0.000 description 3
- 230000008676 import Effects 0.000 description 2
- 238000011002 quantification Methods 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 206010021403 Illusion Diseases 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
Images
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
- G10L13/00—Speech synthesis; Text to speech systems
-
- 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
- G10L19/08—Determination or coding of the excitation function; Determination or coding of the long-term prediction parameters
-
- 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
- G10L2019/0001—Codebooks
- G10L2019/0007—Codebook element generation
-
- 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
- G10L2019/0001—Codebooks
- G10L2019/0011—Long term prediction filters, i.e. pitch estimation
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
A lag parameter coding means (215b) generates a code corresponding to a lag parameter value by using a lag parameter code book (215a). On the decoding side, a lag parameter value corresponding to the lag parameter code generated on the coding side is decoded by using the same lag parameter code book (215a) and outputted. In the lag parameter code book (215a), the relation between the lag parameter value and the corresponding code (P) is shown. The relation is so determined as to increase the rate at which the decoded lag parameter value of when a bit error occurs in the code deviates to approximately an integral multiple (including one time) or an integral submultiple of the decoded lag parameter value of when no bit error occurs. As a result, the auditory degradation of quality of decoded sound is suppressed even when the code has a bit error.
Description
Technical field
The present invention relates to a kind of voice processing apparatus that can be applicable to digital cellular telephone and personal computer etc., be particularly related to a kind of method and apparatus of coding lag parameter and the method for making the code book that is used for this, this lag parameter is represented pitch period or related parameter is arranged, and is one of multiple parameter of expression phonic signal character.
Background technology
One of multiple important parameter of expression phonic signal character is pitch period and lag parameter.These parameters are used as voice coding and are used for effectively the coding parameter that voice signal is encoded and the synthetic parameters of phonetic synthesis in handling.When sending or store lag parameter, need parameter value be encoded to the code corresponding with this value according to ad hoc rules.
The coding lag parameter method that is used for voice coding is described in the ITU-T of international organization and recommends G.729 (8kbps CS-ACELP voice coding method).
Lag parameter according to this recommendation coding is sent with other coding parameters.Lag parameter in this conventional example be indication when make when being used for synthesize the pumping signal of decoded speech by the CS-ACELP algorithm, the value (lagged value) of which section of signal that use is included in the code book that is called " adaptive codebook ", the CS-ACELP algorithm is a kind of speech coding algorithm of this conventional example.This lagged value T is made up of integral part T1 (T1=19 to 143) and fraction part frac/3 (frac=-1,0,1).
This lagged value T uses above-mentioned T1 and frac to be encoded to code P (P=0 to 255) in the formula (1) by scrambler.
On the other hand, decoding lagged value T1 and frac are decoded according to code P according to the rule opposite with formula (1) by demoder.
Lag parameter is the time t1 from voice signal, the retardation of the t0 before the T1, and the waveform at t0 place is similar to the waveform at t1 place.That is, lag parameter is indicated the parameter of the pitch period of periodic waveform typically, and is the pitch period of voice itself.Yet lag parameter is the parameter with broad sense in a sense, it be included in the aperiodicity signal, for example in the initial voice to the retardation of the similar position of waveform.
Yet, in the lag parameter code that obtains by above-mentioned traditional coding lag parameter method, if in the process that sends or store bit error takes place, the lagged value of then decoding is same as the correct lagged value of zero defect far from, and this can make decoded speech deteriorate significantly.
Usually, a kind of method that the quality that bit error in the inhibition code causes worsens provides certain correlationship between the distortion, as the Euclidean distance between the parameter value of coding parameter with indicate distance (Hamming distance) between the code of these parameter values, and reduce the influence of bit error.
If as the measuring of distortion between the parameter value of lag parameter, then as long as these values are very little, they are exactly effective with Euclidean distance between these lagged values and difference etc.Yet, if should value surpass certain value, no longer may keep the corresponding relation with perceptual distortion, and to use above-mentioned universal method to handle for the coding/decoding of lag parameter be not very effective.
In order to tackle these bit errors, a kind of detection of bit errors and the method that prevents to use the lagged value that comprises mistake are arranged, but the error detection occurs of this method itself is very complicated, in addition, it is inappropriate redundant bit, for example check bit being added in low bitrate communication method, for example voice communication.
The present invention proposes in light of this situation, its objective is at the lag parameter code to have bit error and cause thus under the situation that the sensation speech quality worsens, the method that a kind of excellent process that lag parameter is encoded and device is provided and makes the code book that can compress.
Summary of the invention
In order to address the above problem, the present invention uses the coding that carries out lag parameter as the code book of giving a definition.Code book is following generation, that is, make have the decoding hysteresis parameter values of bit error approach not have the bit mistake decode value n doubly (comprising 1 times) or 1/n doubly the ratio of the lag parameter code of (n is an integer) improve.
Following generation code book, promptly, make in the code book summation of Hamming distance distortion of decode value between the specified bit number is with interior code be minimized or near minimizing, and use following distortion measure, that is, the n that makes decoding hysteresis parameter values and this value doubly or between another value of 1/n times the distortion of measurement less.
Consequently, following generation code book, promptly, make have the decoding hysteresis parameter values of bit error approach not have the bit mistake decode value n doubly (comprising 1 times) or 1/n doubly the ratio of the lag parameter code of (n is an integer) improve, this makes it possible to worsen with littler sensation speech quality voice signal is carried out coding/decoding.
Brief description of the drawings
Fig. 1 is the block scheme of wireless transmitter of using the coding lag parameter method and apparatus of the embodiment of the invention 1;
Fig. 2 is the block scheme of voice coding portion of the radio communication device of embodiment 1;
Fig. 3 is the block scheme of wanting portion of voice coding portion of the radio communication device of embodiment 1;
Fig. 4 is the block scheme of wanting portion of tone decoding portion of the radio communication device of embodiment 1; And
Fig. 5 is a process flow diagram of making the code book of the radio communication device that can be applicable to embodiment 1.
Implement best way of the present invention (embodiment 1)
Below with reference to Fig. 1 to Fig. 5 embodiments of the invention 1 are described.
Fig. 1 is a block scheme of using wireless transmitter of the present invention.
Voice send handles following carrying out: is numeral by A/D converter 102 from analog-converted from the voice signal of microphone 101 inputs, outputs to voice coding portion 103, and encodes according to for example CELP algorithm.Coding output is by modulator/demodulator 104 modulation of cdma system etc., through the wireless transmission part 105 and antenna 106 send.
Following carrying out receive to be handled in voice: the modulated signal that receives through antenna 107 and wireless receiving portion 108 is by modulator/demodulator 104 demodulation, then by 109 decodings of tone decoding portion, is simulation by D/A converter from digital conversion, exports as voice from loudspeaker 111.
The present invention is applied to the part that adaptive codebook search used in the voice coding portion 103 of above-mentioned radio communication device and the tone decoding portion 109 is handled.
Fig. 2 is the block scheme of the voice coding portion 103 of radio communication device, and the general structure of CELP type speech coders/decoders is shown.Import the voice signal that A/D changed from terminal 201, and output to lpc analysis portion 202.Lpc analysis portion 202 carries out linear prediction analysis according to input speech signal, and the output linear predictor coefficient.203 pairs of linear predictor coefficients of LPC parameter quantification portion (L) quantize, and quantized result is outputed to composite filter 204 and multiplexer 205.
Composite filter 204 constitutes the wave filter with the given characteristic of above-mentioned linear predictor coefficient, the pumping signal of importing from totalizer 206 is carried out filtering, and the result is outputed to totalizer 207.This totalizer 207 is calculated from the input speech signal of terminal 201 with from the error between the output of composite filter 204, and error signal is outputed to perceptual weighting portion 208.Perceptual weighting portion 208 carries out the weighted corresponding with the sensation of error signal, and the result is outputed to error minimize portion 209.
The code vector of adaptive codebook 210 and constant excitation code book 211 is selected to be used for by error minimize portion 209, makes can be minimized from the error signal of perceptual weighting portion 208 outputs, and the gain of the code book 212 of selecting to be used to gain.
Adaptive codebook 210 is pumping signal tables, its storage excitation vector in the past, and optionally output error minimizes the particular code vector that portion 209 is selected.Multiplier 213 multiply by the gain that gain code book 212 is selected with output, and the result is outputed to totalizer 206.
Point out in passing, this adaptive codebook 210 comprises impact damper, storage finally is defined as the history of certain period of excitation vector of the output of totalizer 206, and according to the code vector that error minimize portion 209 is selected, the lagged value that indication should be extracted which section of the burst that is stored in the described impact damper outputs to coding lag parameter portion 215.This coding lag parameter portion 215 comprises in advance lag parameter code book 215a and the 215b of coding lag parameter portion that rule according to the rules produces, and according to certain rule the lagged value of adaptive codebook 210 is encoded, and it is outputed to multiplexer 205.This coding lag parameter portion 215 will be described in detail later.
Constant excitation code book 211 optionally output error minimizes the particular fixed excitation code vector that portion 209 is selected.Multiplier 216 multiply by the gain that gain code book 212 is provided with output, and the result is outputed to totalizer 206.
Totalizer 206 is asked the output sum of multiplier 213 and multiplier 216, and it is outputed to composite filter 204 as excitation vector.Simultaneously, output is fed back to adaptive codebook 210, and is accumulated.
Like this, the error signal that is stored in all excitation vectors in the adaptive codebook 210 is measured by error minimize portion 209, and will output to multiplexer 205 with output (P), the output (S) of constant excitation code book 211 and the output (G) of gain code book 212 from the corresponding 215b of coding lag parameter portion of the minimum value of the error signal of perceptual weighting portion 208.The linear predictor coefficient (L) of multiplexer 205 multiplexing quantifications and above-mentioned output (P), (S) and (G), and the result is outputed to the modulator 104 of Fig. 1.
The tone decoding portion 110 of radio communication device (Fig. 1) also comprises general CELP type Voice decoder, but omits its explanation herein.
Below, describe application coding lag parameter of the present invention portion 215 in detail.
Fig. 3 and Fig. 4 illustrate the structure of wanting portion of application coding lag parameter of the present invention portion 215, and wherein Fig. 3 illustrates the functional block of coding lag parameter portion end, and Fig. 4 illustrates the functional block of lag parameter lsb decoder end.This coding lag parameter portion is not limited to cell phone, but can be applicable to carry out all devices of audio coding/decoding.
As shown in Figure 3, coding lag parameter portion 215 comprises lag parameter code book 215a and the 215b of coding lag parameter portion that lagged value is encoded with reference to this lag parameter code book 215a.Lag parameter code book 215a is the table that storage has the output code of Input Hysteresis value and correspondence, is to produce according to certain rule in advance.
Equally, as shown in Figure 4, the lag parameter lsb decoder of Voice decoder comprises: lag parameter code book 215a is identical with above-mentioned coding lag parameter portion; With lag parameter lsb decoder 401, decode with reference to 215a pair of lag parameter corresponding of this lag parameter code book with the coding that receives/import.
Describe in detail below and have the as above coding lag parameter portion 215 of structure.
Lag parameter code book 215a illustrates the table that concerns between the code P of hysteresis parameter values T and correspondence.For example, if code book length is N, then storage and the corresponding lagged value T of code P (=0 to N-1).In addition, can also ask intermediate code P0 (0 to N-1) by the formula (1) that calculating formula, the ITU-T that for example mentions in the prior art recommend to be used for coding lag parameter in G.729 (8 kbpsCS-ACELP), and the mapping table of the storage final code P (=0 to N-1) corresponding with P0.
Lag parameter code book 215a of the present invention is characterised in that to have following structure, promptly, so produce code book, make have the decoding hysteresis parameter values of bit error approach not have the bit mistake decode value n doubly (comprising 1 times) or 1/n doubly the ratio of the lag parameter code of (n is an integer) improve.Its production method is with aftermentioned.
Lag parameter be be included in voice signal in the relevant parameter of pitch period.In some cases, because bit error etc. do not obtain correct lagged value.Yet the present inventor finds, if the n that wrong decoding lagged value approaches correct lagged value doubly (comprising 1 times) or 1/n (n is an integer) doubly, sensation worsens less relatively.This is because as long as satisfy above-mentioned situation, use the decoding of this mistake lagged value or the frequency spectrum of synthetic voice signal to comprise that the frequency component of correct pitch period is as its part.
As mentioned above, embodiment 1 has utilized following characteristic, promptly, value with lag parameter of bit error approached under n times (the comprising 1 times) or 1/n (n is an integer) situation doubly of the value of being correctly decoded, sensation worsens less, so can work as bit error when taking place, reduces the deterioration of feeling speech quality.
Below, the method for making above-mentioned lag parameter code book used among the present invention is described.This lag parameter code book is to produce like this, that is, make have the decoding hysteresis parameter values of bit error approach not have the bit mistake decode value n doubly (comprising 1 times) or 1/n doubly the ratio of the lag parameter code of (n is an integer) improve.
Fig. 5 illustrates the processing procedure that is used to make above-mentioned lag parameter code book.
At first, in step 501 initial codebook table (i) (i=0 to N-1 is set; N: code book length).Here, table (i) expression decode value (scalar value or vector value).If this code book is the lag parameter code book, then table (i) can be set to indicate the intermediate code P0 of code i, as described in the lag parameter code book 215a of embodiment 1.In addition, code in the initial codebook and the corresponding relation between the decode value can be determined arbitrarily.
Then, in step 502, be positioned at specific bit number (supposing that it is MB) so that (all combinations of the code of dH≤MB) are calculated the distortion of decode value between the code of these combinations one by one, and obtained summation D0 for Hamming distance dH in the table (i).
Here, the distortion of the decode value between the code depends on the parameter of code indication, but is to use Euclidean distance between the decode value etc.The method of expressing the lag parameter distortion measure is a characteristic of the present invention.This will further describe in embodiment 2.
Then, in step 503, from code book table (i), select at random Hamming distance dH surpass described specific bit number (dH≤MB) with interior code to ia and ib.In step 504, described code between mutually after the exchange decode value, calculate the summation D of Hamming distance decode value distortion between described specific bit number is with interior code.
Then in step 505, whether the summation D of distortion is less than the summation D0 of the distortion of calculating above in the determining step 504.If less than, then described code between exchange decode value, and upgrade the summation of distortion in step 506.
In step 507, judge the convergence of the summation D0 of described distortion, and repeat operation, until the summation convergence of described distortion from described step 503 to 507.
Use above-mentioned processing to make the lag parameter code book, can reduce the summation of the distortion measure of decode value between the code in the specific Hamming distance, decode value when making code generation bit error more approaches correct decode value, suppresses the deterioration of sensation speech quality.
Specifically, by with the minimization limits of the summation of distortion to Hamming distance between the specific bit number is with interior code, when still less the bit error of bit number takes place, can more effectively suppress to worsen.By selecting code that Hamming distance surpasses the specific bit number at random, can obtain higher efficient, and reduce the summation of distortion ia and ib.Like this, even bit error takes place, also can suppress to feel the deterioration of speech quality.
In above-mentioned steps 503, the code of selecting at random from code book table (i) is limited to Hamming distance to ia and ib and surpasses those of specific bit number, but the invention is not restricted to this.(embodiment 2)
Embodiment 2 realizes on the hardware and software identical with embodiment 1.Be the variation of distortion measure with the difference of the method for the used making lag parameter code book of embodiment 1.
Identical among the process of making code book and the Fig. 5 shown in the embodiment.Be with the difference of embodiment 1, with the distortion measured shown in the formula (2) as the decode value between step 502 and the 504 used codes.
D (fa, fb)=min (w1 * d0 (fb, fa), w2 * d0 (fb, 2 * fa), w3 * d0 (fb, 3 * fa)) is fa=Fs/Ta (Hz) wherein ... (2)
fb＝Fs/Tb(Hz)
fb≥fa
d0(fx，fy)＝|fx-fy|/(fx×fy)
1/2
Wherein, Ta and Tb are the decoding lagged value (units: sample) of object code ia and ib; Fa and fb are the frequency values (Hz) of Ta and Tb; Fs is sample frequency (Hz); And d (fa, fb) be code between the distortion of decode value.
Formula (2) is not just expressed the distortion of hysteresis parameter values by any thing that is similar to Euclidean distance.Formula (2) is to have considered that (w1, w2 and w3 are weighting constants for the example of definition of n (n is an integer) difference doubly of a lagged value and another lagged value, corresponding to n (n is an integer) distortion doubly from this value), and can use another definition that realizes similar notion.
Use this distortion measure, Hamming distance be positioned at the specific bit number with a decode value of code become n (n the is an integer) value doubly that approaches another decode value.As mentioned above, lag parameter be be included in voice signal in the relevant parameter of pitch period.If because bit error etc., doubly (comprising 1 times) or 1/n (n is an integer) be doubly for the n that the decoding lagged value approaches correct lagged value, then use this value and decode or frequency component that the frequency spectrum of synthetic voice signal comprises correct pitch period as a part, thereby worsen can be less relatively for sensation.
Can be little distortion by approaching this n (n is an integer) value defined doubly, and between the specific bit number is with interior code, make distortion minimization make code book to reduce the distortion summation by being restricted to Hamming distance.Therefore, if produce the lag parameter code book, then under the situation of bit error, even because the parameters such as lag parameter of mistake, the easy dislocation of decode value also can more effectively suppress to feel the deterioration of quality by said method.
As mentioned above, coding lag parameter method invention of the present invention is that lag parameter is carried out Methods for Coding, this lag parameter is to be used for parameter that voice signal is encoded, the present invention is owing to use the lag parameter code book of following generation that lag parameter is encoded, that is, make have the decoding hysteresis parameter values of bit error approach not have the bit mistake decode value n doubly (comprising 1 times) or 1/n doubly the ratio of the lag parameter code of (n is an integer) improve.
In lag parameter coding/decoding method of the present invention, use above-mentioned coding method to use the lag parameter code book identical to decode with coding side at the lag parameter of coding side coding.
As mentioned above, at code book by the following generation of use, promptly, make have the decoding hysteresis parameter values of bit error approach not have the bit mistake decode value n doubly (comprising 1 times) or 1/n doubly the ratio of the lag parameter code of (n is an integer) improve, and the decoding hysteresis parameter values that makes bit error approaches not have n doubly (comprising 1 times) or the 1/n doubly under the situation of the lag parameter code of (n is an integer) of the decode value of bit mistake, sensation worsens less, utilize above-mentioned characteristic, then when code generation bit error, can suppress to feel the deterioration of speech quality.
In addition, code book production method of the present invention invention is the method that is used for following making code book, that is, Hamming distance in the code book is minimized or near minimizing with the summation of the distortion of the decode value between the interior code at the specified bit number.Decode value during code generation bit error is set to the value near the value of being correctly decoded, can suppress to feel the deterioration of speech quality, and the summation of distortion is minimized target limit between the code of Hamming distance in the specific bit number, the deterioration of speech quality in the time of can more effectively suppressing bit generation bit error still less.
When making above-mentioned code book, can may further comprise the steps: the summation of in initial codebook, calculating Hamming distance distortion of decode value between the specified bit number is with interior code; From code book, select a code book right at random; Described code between after the exchange decode value, calculate the summation of Hamming distance distortion of decode value between described specified bit number is with interior code; If the summation of the distortion of described decode value then exchanges described decode value less than the summation of the described distortion of calculating in the past, and upgrade the summation of distortion; And the convergence of judging the summation of described distortion, wherein repeat the described code of selecting at random to, exchange decode value with upgrade the summation of distortion and the convergent step of judging the summation of distortion, until the summation convergence of described distortion.
And preferably use following distortion measure, promptly, the n of decoding hysteresis parameter values and this value doubly or between 1/n (n is an integer) another value doubly the distortion of measurement less, and use distortion measure with following characteristic, that is, the value with lag parameter of bit error approached under n times (the comprising 1 times) or 1/n (n is an integer) situation doubly of the value of being correctly decoded, and sensation worsens less, so in the time of can working as the bit error generation, reduce the deterioration of sensation speech quality.
In addition, the code book that can use above-mentioned coding method and coding/decoding method or be produced by a kind of method in the above-mentioned code book production method is realized a kind of coding lag parameter/coding/decoding method, is used for lag parameter is carried out coding/decoding.
Can also be a kind of speech coder with the invention process, comprise: code book, the parameter value of expression lag parameter and the corresponding relation between the code, this lag parameter is the coding parameter of voice signal; And the coding lag parameter device, use described code book that lag parameter is encoded.Can also be a kind of Voice decoder with the invention process, comprise and use code book the lag parameter demoder at coding side by the lag parameter code of above-mentioned encoder encodes decode identical with coding side.In addition, can also realize coding lag parameter device/demoder with single assembly.
Can also realize above-mentioned coding method with computer software.More particularly, can construct a kind of system, comprise: computer-readable media; And programmed instruction portion, being used for instruction computer processor uses the lag parameter code book of following generation that lag parameter is encoded, promptly, make have the decoding hysteresis parameter values of bit error approach not have the bit mistake decode value n doubly (comprising 1 times) or 1/n doubly the ratio of the lag parameter code of (n is an integer) improve, wherein said programmed instruction portion is stored in the described medium with executable format, and is loaded in the computer memory with the operational computations machine when being carried out by described processor.
Certain above-mentioned decoding device also can be realized with computer software equally.
Can also use this encoding software in the various mediums by above-mentioned encoding software is stored into.It is the medium of stored program mechanical-readable, this programmed instruction computer processor uses the lag parameter code book of following generation that lag parameter is encoded, that is, make have the decoding hysteresis parameter values of bit error approach not have the bit mistake decode value n doubly (comprising 1 times) or 1/n doubly the ratio of the lag parameter code of (n is an integer) improve.Then, it is downloaded to computing machine, with the operational computations machine, thereby implements above-mentioned coding method.
Certain above-mentioned decoding software also can use by being stored to equally in the various mediums.
Can also be a kind of code book generation device with the invention process, comprise: computer-readable media; And programmed instruction portion, be used for the following generation code book of instruction computer processor, promptly, make in the code book summation of Hamming distance distortion of decode value between the specified bit number is with interior code be minimized or near minimizing, wherein said programmed instruction portion is stored in the described medium with executable format, and is loaded in the computer memory with the operational computations machine when being carried out by described processor.
When the present invention can be applied to be free from mistakes and detect, and can certainly be applied to error detection occurs the time.It can also be applied to carry out all voice coding/decoding methods of coding lag parameter.
The application is based on please No.HEI 10-29332 in the Japan Patent of submitting on January 27th, 1998, Its full content is contained in this as a reference. Applicability on the industry
Encoder of the present invention, decoder and Code And Decode method can be applicable to be furnished with widely voice The equipment of encoder and Voice decoder. Preferably the present invention is used in radio communication device, for example digital Cell phone is because it can suppress to feel the deterioration of speech quality effectively.
Claims (19)
1, a kind of coding lag parameter method, be used for lag parameter is encoded, this lag parameter is the coding parameter of voice signal, this method uses the lag parameter code book of following generation that lag parameter is encoded, that is, make have the decoding hysteresis parameter values of bit error approach not have the bit mistake decode value n doubly (comprising 1 times) or 1/n doubly the ratio of the lag parameter code of (n is an integer) improve.
2, a kind of lag parameter coding/decoding method uses the lag parameter code book identical with coding side to being decoded by the lag parameter code of the coding method coding of claim 1 at coding side.
3, a kind of code book production method, following generation code book promptly, makes in the code book summation of Hamming distance distortion of decode value between the specified bit number is with interior code be minimized or near minimizing.
4, code book production method as claimed in claim 3 may further comprise the steps:
In initial codebook, calculate the summation of Hamming distance distortion of decode value between the specified bit number is with interior code;
From code book, select a code book right at random;
Described code book between after the exchange decode value, calculate the summation of Hamming distance distortion of decode value between described specified bit number is with interior code;
If the summation of the distortion of described decode value then exchanges described decode value less than the summation of the described distortion of calculating in the past, and upgrade the summation of distortion; And
Judge the convergence of the summation of described distortion,
Wherein repeat the described code of selecting at random to, exchange decode value with upgrade the summation of distortion and the convergent step of judging the summation of distortion, until the summation convergence of described distortion.
5, code book production method as claimed in claim 3 uses following distortion measure, that is, the n of decoding hysteresis parameter values and this value doubly or 1/n doubly between another value of (n is an integer) distortion of measurement less.
6, code book production method as claimed in claim 4 uses following distortion measure, that is, the n of decoding hysteresis parameter values and this value doubly or 1/n doubly between another value of (n is an integer) distortion of measurement less.
7, a kind of coding lag parameter/coding/decoding method, the coding method of use claim 1 and the coding/decoding method of claim 2 carry out the coding/decoding of lag parameter.
8, a kind of coding lag parameter/coding/decoding method, use are carried out the coding/decoding of lag parameter by the code book of the code book production method generation of claim 3.
9, a kind of coding lag parameter/coding/decoding method, use are carried out the coding/decoding of lag parameter by the code book of the code book production method generation of claim 4.
10, a kind of coding lag parameter/coding/decoding method, use are carried out the coding/decoding of lag parameter by the code book of the code book production method generation of claim 5.
11, a kind of coding lag parameter/coding/decoding method, use are carried out the coding/decoding of lag parameter by the code book of the code book production method generation of claim 6.
12, a kind of coding lag parameter device comprises:
Code book, the parameter value of expression lag parameter and the corresponding relation between the code, this lag parameter is the coding parameter of voice signal; And
The coding lag parameter device uses described code book that lag parameter is encoded,
Wherein said code book is following generation, that is, make have the decoding hysteresis parameter values of bit error approach not have the bit mistake decode value n doubly (comprising 1 times) or 1/n doubly the ratio of the lag parameter code of (n is an integer) improve.
13, a kind of lag parameter demoder comprises and uses code book the lag parameter demoder at coding side by the lag parameter code of the encoder encodes of claim 12 decoded identical with coding side.
14, a kind of coding lag parameter device/demoder, the scrambler of use claim 12 and the demoder of claim 13 carry out the coding/decoding of lag parameter.
15, a kind of speech coders/decoders comprises the coding lag parameter device/demoder of claim 14.
16, a kind of radio communication device comprises the speech coders/decoders of claim 15.
17, a kind of scrambler comprises:
Computer-readable media; And
Programmed instruction portion, being used for instruction computer processor uses the lag parameter code book of following generation that lag parameter is encoded, promptly, make have the decoding hysteresis parameter values of bit error approach not have the bit mistake decode value n doubly (comprising 1 times) or 1/n doubly the ratio of the lag parameter code of (n is an integer) improve
Wherein said programmed instruction portion is stored in the described medium with executable format, and is loaded in the computer memory with the operational computations machine when being carried out by described processor.
18, a kind of demoder comprises:
Computer-readable media; And
Programmed instruction portion, being used for instruction computer processor uses the lag parameter code book of following generation that lag parameter is decoded, promptly, make have the decoding hysteresis parameter values of bit error approach not have the bit mistake decode value n doubly (comprising 1 times) or 1/n doubly the ratio of the lag parameter code of (n is an integer) improve
Wherein said programmed instruction portion is stored in the described medium with executable format, and is loaded in the computer memory with the operational computations machine when being carried out by described processor.
19, a kind of code book generation device comprises:
Computer-readable media; And
Programmed instruction portion is used for the following generation code book of instruction computer processor, that is, make in the code book summation of Hamming distance distortion of decode value between the specified bit number is with interior code be minimized or near minimizing,
Wherein said programmed instruction portion is stored in the described medium with executable format, and is loaded in the computer memory with the operational computations machine when being carried out by described processor.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2933298 | 1998-01-27 | ||
| JP29332/1998 | 1998-01-27 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1256001A true CN1256001A (en) | 2000-06-07 |
Family
ID=12273290
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN99800072A Pending CN1256001A (en) | 1998-01-27 | 1999-01-26 | Method and device for coding lag parameter and code book preparing method |
Country Status (6)
| Country | Link |
|---|---|
| EP (1) | EP0971338A1 (en) |
| KR (1) | KR20010005669A (en) |
| CN (1) | CN1256001A (en) |
| AU (1) | AU2075199A (en) |
| CA (1) | CA2283203A1 (en) |
| WO (1) | WO1999038157A1 (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103474075A (en) * | 2013-08-19 | 2013-12-25 | 安徽科大讯飞信息科技股份有限公司 | Method and system for sending voice signals, and method and system for receiving voice signals |
| CN110999088A (en) * | 2017-07-25 | 2020-04-10 | 日本电信电话株式会社 | Encoding device, decoding device, data structure of code string, encoding method, decoding method, encoding program, and decoding program |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8793557B2 (en) | 2011-05-19 | 2014-07-29 | Cambrige Silicon Radio Limited | Method and apparatus for real-time multidimensional adaptation of an audio coding system |
| US8819523B2 (en) * | 2011-05-19 | 2014-08-26 | Cambridge Silicon Radio Limited | Adaptive controller for a configurable audio coding system |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3250367B2 (en) * | 1994-04-15 | 2002-01-28 | 松下電器産業株式会社 | Encoded signal decoding method and apparatus |
| JPH09261070A (en) * | 1996-03-22 | 1997-10-03 | Sony Corp | Digital audio signal processing unit |
| JPH10200580A (en) * | 1997-01-16 | 1998-07-31 | Matsushita Electric Ind Co Ltd | Method for reproducing voice packet |
| JP3287543B2 (en) * | 1997-01-24 | 2002-06-04 | 日本電信電話株式会社 | Error correction encoding method and decoding method |
-
1999
- 1999-01-26 EP EP99901171A patent/EP0971338A1/en not_active Withdrawn
- 1999-01-26 CA CA002283203A patent/CA2283203A1/en not_active Abandoned
- 1999-01-26 KR KR1019997008737A patent/KR20010005669A/en active IP Right Grant
- 1999-01-26 WO PCT/JP1999/000294 patent/WO1999038157A1/en not_active Application Discontinuation
- 1999-01-26 AU AU20751/99A patent/AU2075199A/en not_active Abandoned
- 1999-01-26 CN CN99800072A patent/CN1256001A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103474075A (en) * | 2013-08-19 | 2013-12-25 | 安徽科大讯飞信息科技股份有限公司 | Method and system for sending voice signals, and method and system for receiving voice signals |
| CN103474075B (en) * | 2013-08-19 | 2016-12-28 | 科大讯飞股份有限公司 | Voice signal sending method and system, method of reseptance and system |
| CN110999088A (en) * | 2017-07-25 | 2020-04-10 | 日本电信电话株式会社 | Encoding device, decoding device, data structure of code string, encoding method, decoding method, encoding program, and decoding program |
Also Published As
| Publication number | Publication date |
|---|---|
| CA2283203A1 (en) | 1999-07-29 |
| WO1999038157A1 (en) | 1999-07-29 |
| KR20010005669A (en) | 2001-01-15 |
| EP0971338A1 (en) | 2000-01-12 |
| AU2075199A (en) | 1999-08-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1143265C (en) | Transmission system with improved speech encoder | |
| CN1123866C (en) | Dual subframe quantization of spectral magnitudes | |
| CN1192356C (en) | Decoding method and systme comprising adaptive postfilter | |
| CN1225723C (en) | Noise suppression | |
| CN1154086C (en) | CELP transcoding | |
| CN1235190C (en) | Method for improving the coding efficiency of an audio signal | |
| CN1218295C (en) | Method and system for speech frame error concealment in speech decoding | |
| RU2509379C2 (en) | Device and method for quantising and inverse quantising lpc filters in super-frame | |
| US7978101B2 (en) | Encoder and decoder using arithmetic stage to compress code space that is not fully utilized | |
| CN1241170C (en) | Method and system for line spectral frequency vector quantization in speech codec | |
| CN1221169A (en) | Coding method and apparatus, and decoding method and apparatus | |
| CN101061535A (en) | Method and device for the artificial extension of the bandwidth of speech signals | |
| CN1109697A (en) | Vector quantizer method and apparatus | |
| CN1623185A (en) | Efficient improvement in scalable audio coding | |
| CN1432176A (en) | Method and appts. for predictively quantizing voice speech | |
| US6928406B1 (en) | Excitation vector generating apparatus and speech coding/decoding apparatus | |
| CN1167048C (en) | Speech coding apparatus and speech decoding apparatus | |
| CN101080767A (en) | Method and device for low bit rate speech coding | |
| CN1161750C (en) | Speech encoding and decoding method and apparatus, telphone set, tone changing method and medium | |
| US10607624B2 (en) | Signal codec device and method in communication system | |
| CN1167046C (en) | Vector encoding method and encoder/decoder using the method | |
| CN1192357C (en) | Adaptive criterion for speech coding | |
| CN1145925C (en) | Transmitter with improved speech encoder and decoder | |
| CN101572093B (en) | Transcoding method and device | |
| CN1256001A (en) | Method and device for coding lag parameter and code book preparing method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| 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 |