CN101373595A - Self-adapting multi-velocity encoder with fixed velocity and coding method thereof - Google Patents

Abstract

The invention provides a self-adaptive multi-speed coding method capable of achieving multi-speed coding at a fixed speed and a device. The method can enable a coder and a decoder to keep consistent excitation signals for a long term, so that the sound signal synthesized by the coder can reduce the original sound more accurately. The method and the device can be used for the third generation mobile telecommunication system, i.e. universal mobile telecommunication system (UMTS), as the substitute for a self-adaptive variable-speed multi-speed coder.

Description

A kind of self-adapting multi-velocity encoder with fixed velocity and coding method

Technical field

The present invention relates to adaptive multi-rate coding device and coding method thereof, be specifically related to the adaptive multi-rate coding device finish in constant speed rate mode can many speed coding the time, the consistency problem of coding staff and decoding side.

Background technology

Code excited linear prediction coder has obtained using widely since 1985 are suggested.In the vocoder of CDMA (CDMA) and universal mobile telecommunications system (UMTS), all used the technology of code excited linear prediction coder.

Code Excited Linear Prediction has comprised linear prediction and quantification, self-adapting code book search and fixed codebook search.Because itself has quiet period voice, can be by reducing the transfer rate of the effective compressed voice data of data rate between these quiet period, the application number of Qualcomm is that the patent of 92104618.9 rate changeable vocoder is exactly a scheme about said method.

In UMTS, used adaptive multi-rate (AMR) voice coding, adaptive multi-rate (AMR) voice coding be 3GPP (3G (Third Generation) Moblie partner plan) formulate be applied to voice compression coding in the 3G (Third Generation) Moblie, adaptive multi-rate (AMR) voice coding is divided into self-adapting multi-rate narrowband (AMR-NB) voice coding, AMR-WB (AMR-WB) voice coding and AMR-WB modified (AMR-WB+) voice coding again, and these coding methods are all based on code book excitation linear linear forecasting technology.The code book excited linear prediction (CELP) coder that adopts in adaptive multi-rate (AMR) code encoding/decoding mode is divided into several subframes with a voice signal frame, carries out linear prediction and quantification, self-adapting code book search and quantification and fixed codebook search and quantification.12.2,10.2,7.95,7.40,6.70,5.90,5.15,4.75kb/s (kilobits/second) AMR-NB (self-adapting multi-rate narrowband) voice coding is supported the code rate of eight kinds of speed:, and the ground unrest code rate of low rate (1.80kb/s), the form 1 (Table1) of the chapters and sections 5 of the TS26.071-500 of 3GPP has provided the correspondence abbreviation of the speed of above-mentioned these self-adapting multi-rate narrowbands: AMR_12.20, AMR_10.20, AMR_7.95, AMR_7.40, AMR_6.70, AMR_5.90, AMR_5.15, AMR_4.75 and AMR_SID; AMR-WB (AMR-WB) voice coding is also supported the code rate of eight kinds of speed: 23.85,23.05,19.85,18.25,15.85,14.25,12.65,8.85,6.6kb/s, and the ground unrest code rate of low rate (1.75kb/s), the form 1 (Table1) of the chapters and sections 5 of the TS26.171-500 of 3GPP has provided the correspondence abbreviation of the speed of above-mentioned these AMR-WB: AMR-WB_23.85, AMR-WB_23.05, AMR-WB_19.85, AMR-WB_18.25, AMR-WB_15.85, AMR-WB_14.25, AMR-WB_12.65, AMR-WB_8.85, AMR-WB_6.60 and AMR-WB_SID.

Linear prediction and quantification have comprised: the voice signal frame that sampling is obtained or form a sequence through pretreated voice signal frame, take advantage of sample sound in this sequence with a window function, so that the voice data frame of a windowing to be provided; Voice data frame by described windowing calculates one group of coefficient of autocorrelation; Calculate one group of linear predictor coefficient with Lai Wenxun-Du Bin (Levinson-Durbin) algorithm by described coefficient of autocorrelation batch total: described linear predictor coefficient group is transformed into another spectrum domain; Quantize the described coefficient sets that is transformed on another spectrum domain according to the speed in the coded order, for example, one group of line frequency spectrum on 10 rank is to the value of (LSP), or one group of acoustic reactance on 16 rank is received the value of frequency spectrum to (ISP), about the line frequency spectrum to (LSP), in Song Hejiang is published in article " the line frequency spectrum is to (LSP) and speech data compression " in international language voice and the signal Processing meeting (ICASSP) ' 84, explanation is arranged, the application number of Qualcomm is in the patent of 92104618.9 rate changeable vocoder explanation to be arranged also, explanation is also all arranged in the C.S0014-A of the TS of 3GPP (technical manual) 26090 and 3GPP2, receive frequency spectrum to (ISP) about acoustic reactance, can receive frequency spectrum to acoustic reactance at 5.2.3 chapters and sections-linear predictor coefficient of the TS26190 of 3GPP and find explanation in to conversion (LP to ISP Conversion).

In the Qualcomm Code Excited Linear Prediction (QCELP) process, the best code book vector signal that self-adapting code book search and fixed codebook search obtain multiply by addition after separately the optimum gain, itself and be pumping signal.Pumping signal is must use in the cataloged procedure, and Qualcomm Code Excited Linear Prediction (QCELP) is the synthetic speech based on pumping signal of error minimum between search and the raw tone.

Self-adapting multi-rate narrowband (AMR-NB) and AMR-WB (AMR-WB) have all comprised the process of self-adapting code book search, the TS26.090 of 3GPP is described the self-adapting code book search of self-adapting multi-rate narrowband, for example, 5.6 of TS26.090-310 version joints; TS26.190 is described the self-adapting code book search of AMR-WB.Self-adapting code book search has comprised the calculating that pumping signal before closed loop pitch (pitch) search based on former pumping signal and the interpolation of being undertaken by selected integer and mark pitch delay after this obtains the self-adapting code book vector.

Closed loop pitch searcher is to finish by the minimizing of all square weighted errors between raw tone and the reconstruct voice, described minimizing need be found out minimum all square weighted error the pairing all square weighted error of each delay value in the hunting zone, and the pairing all square weighted error of each delay value is determined the response of former pumping signal by self-adapting code book ferret out signal (target signal) and weighted synthesis filter (weighted synthesis filter).Concerning self-adapting multi-rate narrowband, 5.6 in the TS26.090-310 version of 3GPP joint illustrates this, is exactly that the integer delay value k that finds the solution earlier by the characteristic item R (k) of following formula (1) expression when maximum obtains best integer delay,

$R\left(k\right)=\frac{{\mathrm{\Σ}}_{n=0}^{39}x\left(n\right)y_k\left(n\right)}{\sqrt{{\mathrm{\Σ}}_{n=0}^{39}{y}_k\left(n\right)y_k\left(n\right)}},---\left(1\right)$

X (n) is an echo signal, y _kBe that near the mark delay value best integer delay obtains by the normalized characteristic item R of interpolation (k) in the value through filtering of crossing deactivation signal at integer time-delay k place, the maximum mark delay value of search can obtain best score to postpone, deposit the pumping signal value be the excitation impact damper (u (n), n=-(143+11) ... 39,), the value of search phase (u (n), n=0 wherein, 1,, 39) be linear residual error (LP residual); Concerning AMR-WB, the joint of 5.7 in the TS26.190-510 version of 3GPP illustrates this, is exactly the characteristic item T that finds the solution earlier by following formula (2) expression _kInteger delay value k when maximum obtains best integer delay,

$T_k=\frac{{\mathrm{\Σ}}_{n=0}^{63}x\left(n\right)y_k\left(n\right)}{\sqrt{{\mathrm{\Σ}}_{n=0}^{63}{y}_k\left(n\right)y_k\left(n\right)}},---\left(2\right)$

Near best integer delay mark delay value also is by the normalized characteristic item T of interpolation _kObtain, the maximum mark delay value of search can obtain best score to postpone, deposit the pumping signal value be the excitation impact damper (u (n), n=-(231+17) ... 63), value (u (n), the n=0 of same search phase, 1 ..., 63) also be linear residual error (LPresidual).Pumping signal value (u (n), n＜0) before search phase in the excitation impact damper is the pumping signal value of former subframe.The pumping signal of each subframe is the signal that obtains after the self-adapting code book vector signal of current subframe amplifies by the self-adapting code book yield value that quantizes, obtain the signal resulting signal that superposes after amplifying by the fixed codebook gain value that quantizes with the fixed code book vector signal, about this point, can be referring to 5.9 joints of the TS26.090-310 version of 3GPP, its Chinese style (64) is the mathematical notation of pumping signal value; Also can be referring to 5.10 joints of TS26.190-510 version, its Chinese style (56) is the mathematical notation of pumping signal value.

In self-adapting multi-rate narrowband (AMR-NB) and AMR-WB (AMR-WB) the tone decoding process, each frame is all carried out LP (linear prediction) filter parameter decoding, thereby be formed for the LP filter coefficient of each subframe of the voice signal of each subframe of reconstruct; The building method of the pumping signal of each subframe is: the signal that obtains after the self-adapting code book vector signal is amplified by the self-adapting code book yield value, the signal that obtains after amplifying by the fixed codebook gain value with the fixed code book vector signal superposes, and self-adapting code book yield value here and fixed code book vector signal are the quantized values that the self-adapting code book gain index that obtains according to decoding and fixed code book index find from quantization table; Described self-adapting code book vector signal is based on the composite signal of the pumping signal on continuous 154 sample points of past in the excitation impact damper of the pumping signal that comprises a last subframe for AMR-NB, described self-adapting code book vector signal is based on the composite signal of the pumping signal on continuous 248 sample points of past in the excitation impact damper of the pumping signal that comprises a last subframe for AMR-WB, when the pitch delay of a last subframe during less than the time span of a subframe, the pumping signal that only needs a subframe, promptly, the self-adaption of decoding codebook index obtain integer and mark pitch delay, by described integer and mark pitch delay the pumping signal of a last subframe is carried out interpolation and obtains the self-adapting code book vector signal.

Self-adapting multi-rate narrowband (AMR-NB) and AMR-WB (AMR-WB) voice coding have all comprised the process that fixed codebook gain quantizes, fixed codebook gain quantizes to comprise: the prediction gain (or fixed code book prediction gain) that obtains based on the quantification energy predicting error (quantifiedprediction error) of former subframe, and the quantification of the modifying factor between fixed codebook gain and the described prediction gain (or fixed code book prediction gain).The quantification energy predicting error (quantifiedprediction error) of subframe is exactly the value after the logarithm of described modifying factor amplifies by fixed proportion.

The TS26.090 of 3GPP quantizes to be described to the fixed codebook gain of self-adapting multi-rate narrowband, for example, formula (54) and (56) in 5.8 joints of TS26.090-310 version, just Xia Mian formula (3) and (4) illustrate quantification energy predicting error how impact prediction gain

$\tilde{E}\left(n\right)=\underset{i=1}{\overset{4}{\mathrm{\Σ}}}{b}_i\hat{R}(n-i)---\left(3\right)$

$g_c^{\′}={10}^{0.05(\tilde{E}\left(n\right)+\stackrel{\‾}{E}-E_I)}---\left(4\right)$

Formula (3) is a n subframe prediction energy (predicted energy)

Definition, value is moving average (MA) predictive coefficient for [the b1 b2 b3 b4] of [0.68 0.58 0.340.19],

It is exactly the quantification energy predicting error of k subframe; Formula (4) is fixed code book prediction gain (predicted fixed-codebook gain) g ' _cDefinition, E upgrades the mean value of energy (innovation energy) and gets different constant values according to the difference of speed, is 36 decibels (dB) during 12.2kb/s for example, E _IBe on average to upgrade energy (mean innovation energy).Modifying factor between fixed codebook gain and the fixed code book prediction gain is the ratio of the former with the latter; And the formula (58) in 5.8 joints of TS26.090-310 version illustrate that energy predicting error R (n) 20 is multiplied by the logarithm of stating modifying factor, quantizes the energy predicting error and then is 20 and take advantage of the logarithm of quantification modifying factor.

TS26.190 quantizes to be described to the fixed codebook gain of AMR-WB, for example, and 5.9 joints of TS26.0190-510 version.In formula (50) and (52), just Xia Mian formula (5) and (6) illustrate and quantize the how impact prediction gain of energy predicting error,

$\tilde{E}\left(n\right)=\underset{i=1}{\overset{4}{\mathrm{\Σ}}}{b}_i\hat{R}(n-i)---\left(5\right)$

$g_c^{\′}={10}^{0.05(\tilde{E}\left(n\right)+\stackrel{\‾}{E}-E_I)}.---\left(6\right)$

Formula (5) is a n subframe prediction energy (predicted energy)

Definition, value is moving average (MA) predictive coefficient for [the b1 b2 b3 b4] of [0.5 0.4 0.3 0.2],

It is exactly the quantification energy predicting error of k subframe; Formula (6) is the definition of prediction gain (predicted gain) g ' c, and E is that value is the mean value of the renewal energy (innovation energy) of 30 decibels (dB), E _iBe on average to upgrade energy (mean innovation energy).Modifying factor between fixed codebook gain and the prediction gain is the ratio of the former with the latter; And the formula (53) in 5.9 joints of TS26.0190-510 version illustrate that energy predicting error R (n) 20 is multiplied by the logarithm of stating modifying factor, quantizes the energy predicting error and then is 20 and take advantage of the logarithm of quantification modifying factor.

The digital voice frame of sampled digital Speech frame through forming after the pre-service through linear prediction and quantification, self-adapting code book search and fixed codebook search after the resonance peak of formed synthetic digital Speech frame mainly determined by the employed linear prediction analysis of linear prediction (LPC), more definite, concerning AMR-NB, be exactly after LSP is converted to prediction (LP) coefficient, one 10 rank linear prediction synthesis filter (linear prediction synthesis filter) just can be determined d wherein by formula (7) _i(i=1 ..., m m=10) is prediction (LP) coefficient that has quantized; Concerning AMR-WB, be exactly after ISP is converted to prediction (LP) coefficient, one 16 rank linear prediction synthesis filter also can be determined d wherein by formula (7) _i(i=1 ..., m m=16) is prediction (LP) coefficient that has quantized.

$H\left(z\right)=\frac{1}{\hat{A}\left(z\right)}=\frac{1}{1+{\mathrm{\Σ}}_{i=1}^m{\hat{a}}_i{z}^{-i}},---\left(7\right)$

For AMR-NB and AMR-WB, is exactly synthetic digital Speech frame with pumping signal by the filtered output of linear prediction synthesis filter, so, the limit correspondence of linear prediction synthesis filter the frequency and the bandwidth of resonance peak of synthetic digital Speech frame, these resonance peaks are reflected on the intensity of the waveform on the time domain, and are very big to sense of hearing influence.

Summary of the invention

The technical matters that solves

By application number be one of the given information structuring of first embodiment in 200410068056.7 the patented claim as shown in Figure 1 generate the implementation of the adaptive multi-rate coding device of option automatically with quiet description (SID) frame, but adaptive multi-rate type (AMR-WB) scrambler of a variable Rate dual rate coding shown in Figure 1, have 2 voice coding modules 101 and 102, digital voice signal 1 (or through pretreated digital voice signal 1) is exported to the voice coding module 102 and the voice coding module 101 of the speech coding functional module 100 of being with voice activation to detect simultaneously, 102 output be speech coding frame 12,101 output be speech coding frame 19, the discontinuous transmission control of the speech coding functional module 100 that the band voice activation detects and operational module and voice coding module 101 be the many speed of output adaptive (AMR) frame type signal 11 and 16 respectively, the voice coding module 102 output speech coding frames 12 of the speech coding functional module 100 that the band voice activation detects, what voice coding module 101 was exported is speech coding frame 19, speech coding frame 12 and every frame bit number of 19 depend on code rate indicator signal 17 and 18 respectively, 2 tunnel outputs of code rate indicator signal 17 and 18 1 speed 1 tunnel that to be code rate control modules resolve into 2 speed in the rate command signal 9.Discontinuous transmission control and the many speed of operational module output adaptive (AMR) frame type signal 11 are to multi-path voice code multiplexing module and voice coding module, 11 indications of adaptive multi-rate (AMR) frame type signal: whether information bit 14 is efficient adaptive multi-speed coding frames, the speed of self-adaptive multi-speed coding frame, the speed of described self-adaptive multi-speed coding frame is the speed of active speech frame or the speed of silence description frames, when the speed of self-adaptive multi-speed coding frame was silence description frames, information bit 14 was quiet description (SID) frames 13 of comfort noise sending module output.Voice coding module 101 many rate speech of output adaptive coded frame 19 and adaptive multi-rate (AMR) frame type signal 16 thereof, the speed of adaptive multi-rate (AMR) frame type signal 16 expression adaptive multi-rate speech coding frames 19.Multi-path voice code multiplexing module will be merged into Speech frame coded sequence 2 from the information bit 14 and the adaptive multi-rate speech coding frame 19 of discontinuous transmission and operational module, the method that merges is: when adaptive multi-rate (AMR) frame type signal 11 indication information bits 14 were invalid, the speed that Speech frame coded sequence 2 is set was designated as invalid Speech frame coded sequence; When adaptive multi-rate (AMR) frame type signal 11 indication information bits 14 were silence description frames, the speed that Speech frame coded sequence 2 is set was designated as silence description frames, information bit 14 is put in the Speech frame coded sequence of Speech frame coded sequence 2; When adaptive multi-rate (AMR) frame type signal 11 indication information bits 14 are non-silence description frames, information bit 14 and 19 together is put in the Speech frame coded sequence of Speech frame coded sequence 2, the speed that Speech frame coded sequence 2 is set simultaneously is designated as the indicated speed of adaptive multi-rate (AMR) frame type signal 11 and 16.

Can find, in such scheme, the voice coding module of carrying out the coding task of single active speech code rate (non-ground unrest code rate) needs independently to finish the quantification and the memory modification of linear prediction analysis and quantification, open-loop pitch analysis, impulse response calculating, echo signal calculating, self-adapting code book search, fixed codebook search, self-adapting code book and fixed codebook gain.

Now investigating at scrambler shown in Figure 1 all is under the situation of active speech from the judgement that original state starts and voice activation detects Speech frame, and the code rate of single 23.05kb/s is got back to single 23.05kb/s speed again after switching to the dual rate of 23.05kb/s and 8.85kb/s.

Consideration this switching of dual rate from single 23.05kb/s speed to 23.05kb/s and 8.85kb/s: originally voice coding module 102 and 101 all is operated in 23.05kb/s speed, after switching to 23.05kb/s and 8.85kb/s dual rate, voice coding module 102 is operated in 23.05kb/s speed, and voice coding module 101 is operated in 8.85kb/s speed.Voice coding module 101 and voice coding module 102 have same pumping signal and other state value, the value of the variable of the employed continuous variation of state value finger speech sound coding module cataloged procedure, the ISP parameter of previous frame for example, quantification energy predicting error of pumping signal and previous frame or the like, because the various state values of two voice coding modules this moments all are that same voice coding module is produced the last voice signal frame coding before switching with 23.05kb/s speed based on same original state value, voice coding module 102 and the 101 AMR-WB coded frame of encoding and producing 23.05kb/s and 8.85kb/s respectively based on consistent state value.

Consider again when the frame code rate of scrambler again from the switching of the dual rate receipt one 23.05kb/s speed of 23.05kb/s and 8.85kb/s: originally voice coding module 102 and 101 is operated in 23.05kb/s speed and 8.85kb/s speed respectively, after getting back to single-rate, the voice coding module 102 and 101 of scrambler all is operated under the 23.05kb/s speed, multi-path voice code multiplexing module can only both be put in the Speech frame coded sequence 2 for the 23.05kb/s speed coding frame of voice coding module 102 and 101 outputs with getting one, like this, first voice signal frame coding after switching back has 2 kinds of results because of the described restriction that both get one.These both get one selection relevant to the content that transmits on the influence of the side of decoding and the channel: after switching to the dual rate of 23.05kb/s and 8.85kb/s from single 23.05kb/s speed, being placed in what transmit on the channel in the Speech frame coded sequence 2 that comprises the dual rate frame that scrambler produces is the 23.05kb/s rate frame, after switching back 23.05kb/s speed from the dual rate of 23.05kb/s and 8.85kb/s, the 23.05kb/s speed coding frame of multi-path voice code multiplexing module selection voice coding module 102 is put into 2 li of Speech frame coded sequences can make code translator receive accurate coded frame so; After switching to the dual rate of 23.05kb/s and 8.85kb/s from single 23.05kb/s speed, if what transmit on the channel is 8.85kb/s rate frame among the dual rate speech coding frame sequence 2 of 23.05kb/s and 8.85kb/s, after switching back 23.05kb/s speed from the dual rate of 23.05kb/s and 8.85kb/s, the 23.05kb/s rate frame of multi-path voice code multiplexing module selection voice coding module 101 is put into 2 li of Speech frame coded sequences can make code translator receive accurate coded frame so.But multi-path voice code multiplexing module can't be distinguished 2 same speed rate frame, can only at will select, so the selectional restriction that both of the 23.05kb/s rate coding after switching back get one makes that the take over party might be inconsistent in the code rate conversion with transmit leg voice coding module.

Only when switching to single-rate, many speed do not have the coding and decoding both sides to use the problem of not quite identical speed, if scrambler moves in many speed mode always same problem is arranged also, when the frame code rate of AMR-WB (AMR-WB) scrambler is dual rate and the line output of 23.05kb/s and 8.85kb/s always, but AMR-WB (AMR-WB) scrambler of the dual rate of implementation as shown in Figure 1 coding, voice coding module 102 is the 23.05kb/s rate frame coding of voice signal frame, voice coding module 101 is the 8.85kb/s rate frame coding of voice signal frame, when the result of voice activation detection is active speech always, the output of scrambler is the Speech frame coded sequence that comprises 23.05kb/s and 8.85kb/s rate frame, be equivalent to, voice coding module 102 and voice coding module 101 are continued to use original rate coding separately.If transmit on the channel be on the coded frame of a voice signal frame be 23.05kb/s speed, for take over party's code translator, the coding of 102 pairs of current voice signal frames of voice coding module of 23.05kb/s speed is accurate; If but middle quilt to choose the coded frame of a last voice signal frame of issuing decoding side on the channel be the 8.85kb/s rate frame, for take over party's code translator, the coding of 102 pairs of current voice signal frames of voice coding module of 23.05kb/s speed is accurate inadequately.When the speed of the coded frame of choosing the voice signal frame of issuing decoding side on the channel constantly changes, for take over party's code translator, voice coding module 102 and 101 coding be out of true all, and it also just can't receive more accurate coded frame.

In sum, when adopting many speed of variable bit rate mode that the voice signal frame is carried out adaptive multi-rate coding,, the rate variation that is placed to the coded frame on the channel can cause the decoding error because of changing the speed of the coded frame that code translator receives.When this decoding error is transmitted frame by frame, bigger to the influence of speech acoustical quality.If it is very frequent to be placed to the rate variation of the self-adaptive multi-speed coding frame on the channel, speech quality can severe exacerbation.

For the constant speed rate multi-rate coding device that the present invention was devoted to, the function that it does not have the SID frame to generate automatically, and it is incorporated in the encoding function of ground unrest code rate and the encoding function of active speech speed in the module, as shown in Figure 2 not with the implementation of quiet period detection option AMR-WB (AMR-WB) scrambler, AMR-WB (AMR-WB) speech coder of realizing with this scheme serves as to investigate object, can find, not only need the voice coding module of fill order's rate coding task independently to finish linear prediction analysis and quantification, open-loop pitch is analyzed, impulse response is calculated, echo signal is calculated, the self-adapting code book search, fixed codebook search, the quantification of self-adapting code book and fixed codebook gain and memory modification also need to have at least a voice coding module can finish the whole cataloged procedure of quiet description (SID) frame.In the example specific to Fig. 2, but AMR-WB (AMR-WB) scrambler of a dual rate coding, have 2 voice coding modules 101 and 102, have 2 voice coding modules 101 and 102, they have had more the function of coding quiet description (SID) frame than the voice coding module among Fig. 1, and the voice coding module among Fig. 3 and Fig. 4 also has the function of coding quiet description (SID) frame in the back.Digital voice signal 1 (or through pretreated digital voice signal 1) is simultaneously to voice coding module 102 and 101 outputs, what voice coding module 102 was exported is AMR-WB coded frame 12,101 output be AMR-WB coded frame 19, voice coding module 102 and 101 is the many speed of output adaptive (AMR) frame type signal 11 and 16 respectively, adaptive multi-rate (AMR) frame type signal 11 and 16 is represented the speed of AMR-WB coded frame 12 and 19 respectively, AMR-WB coded frame 12 and every frame bit number of 19 depend on code rate indicator signal 17 and 18 respectively, 2 tunnel outputs of code rate indicator signal 17 and 18 1 speed 1 tunnel that to be code rate control modules resolve into the speed in the rate command signal 9.Multi-path voice code multiplexing module is merged into voice signal frame coded sequence 2 with AMR-WB coded frame 12 and 19, the method that merges is: when adapting to many speed (AMR) frame type signal 11 and the 16 coded frame speed of indicating not simultaneously, the information bit of AMR-WB coded frame 12 and 19 li together is put in the Speech frame coded sequence of Speech frame coded sequence 2, and the speed that Speech frame coded sequence 2 is set simultaneously is designated as the indicated speed of adaptive multi-rate (AMR) frame type signal 11 and 16; When 11 with the coded frame speed of 16 indications when identical, optional AMR-WB coded frame 12 and 19 one of them be put into 2 li of Speech frame coded sequences, the speed that Speech frame coded sequence 2 is set simultaneously is designated as described identical coded frame speed.

The code rate of the single 23.05kb/s that scrambler shown in Figure 2 starts from original state is got back to single 23.05kb/s speed again after switching to the dual rate of 23.05kb/s and quiet description (SID) frame.

Consideration is from single 23.05kb/s speed to 23.05kb/s with this switching of dual rate of ground unrest code rate (AMR-WB_SID): originally voice coding module 102 and 101 all is operated in 23.05kb/s speed, after switching to 23.05kb/s and AMR-WB_SID dual rate, voice coding module 102 is operated in 23.05kb/s speed, and voice coding module 101 is operated in ground unrest code rate (AMR-WB_SID) speed; Before switching, voice coding module 101 and voice coding module 102 have same pumping signal and other state value, the ISP parameter of previous frame for example, quantification energy predicting error of pumping signal and previous frame or the like, because the various state values of two voice coding modules this moments all are that same voice coding module is produced the same last voice signal frame coding before switching with 23.05kb/s speed by same original state value, voice coding module 102 and 101 is encoded based on consistent state value and to be produced the AMR-WB coded frame of 23.05kb/s and quiet description (SID) frame respectively.

Consider again when the frame code rate of scrambler again from the switching of the dual rate receipt one ground unrest code rate (AMR-WB_SID) of 23.05kb/s and ground unrest code rate (AMR-WB_SID): originally voice coding module 102 and 101 is operated in 23.05kb/s speed and ground unrest code rate (AMR-WB_SID) respectively, after getting back to single-rate, the voice coding module 102 and 101 of scrambler all is operated under the 23.05kb/s speed, multi-path voice code multiplexing module can only both be put in the Speech frame coded sequence 2 for the 23.05kb/s speed coding frame of voice coding module 102 and 101 outputs with getting one, like this, first voice signal frame coding after switching back has 2 kinds of results because of the described restriction that both get one.These both get one selection relevant to the content that transmits on the influence of the side of decoding and the channel: switch to the dual rate of 23.05kb/s and ground unrest code rate (AMR-WB_SID) from single 23.05kb/s speed after, being placed in what transmit on the channel in the Speech frame coded sequence 2 that comprises the dual rate frame that scrambler produces is the 23.05kb/s rate frame, after switching back 23.05kb/s speed from the dual rate of 23.05kb/s and ground unrest code rate (AMR-WB_SID), the 23.05kb/s speed coding frame of multi-path voice code multiplexing module selection voice coding module 102 is put into 2 li of Speech frame coded sequences can make code translator receive accurate coded frame so; Switch to the dual rate of 23.05kb/s and ground unrest code rate (AMR-WB_SID) from single 23.05kb/s speed after, if what transmit on the channel is quiet description (SID) frame rate frame among the dual rate speech coding frame sequence 2 of 23.05kb/s and quiet description (SID) frame, after switching back 23.05kb/s speed from the dual rate of 23.05kb/s and ground unrest code rate (AMR-WB_SID), the 23.05kb/s rate frame of multi-path voice code multiplexing module selection voice coding module 101 is put into 2 li of Speech frame coded sequences can make code translator receive accurate coded frame so.But multi-path voice code multiplexing module can't be distinguished 2 same speed rate frame, can only at will select, so the selectional restriction that both of the 23.05kb/s rate coding after switching back get one makes that the take over party might be inconsistent in the code rate conversion with transmit leg voice coding module.

Only when switching to single-rate, many speed do not have the coding and decoding both sides to use the problem of not quite identical speed, if scrambler moves in many speed mode always same problem is arranged also, when the frame code rate of AMR-WB (AMR-WB) scrambler is the dual rate and the line output of 23.05kb/s and ground unrest code rate (AMR-WB_SID) always, but AMR-WB (AMR-WB) scrambler with the dual rate of implementation shown in Figure 2 coding, voice coding module 102 is the 23.05kb/s rate frame coding of voice signal frame, voice coding module 101 is ground unrest code rate (AMR-WB_SID) coding of voice signal frame, the output of scrambler is the Speech frame coded sequence that comprises 23.05kb/s and quiet description (SID) frame rate frame, be equivalent to, voice coding module 102 and voice coding module 101 are continued to use original rate coding separately.If transmit on the channel be on the coded frame of a voice signal frame be 23.05kb/s speed, for take over party's code translator, the coding of 102 pairs of current voice signal frames of voice coding module of 23.05kb/s speed is accurate; If but middle quilt to choose the coded frame of a last voice signal frame of issuing decoding side on the channel be quiet description (SID) frame, for take over party's code translator, the coding of 102 pairs of current voice signal frames of voice coding module of 23.05kb/s speed is very coarse.When the speed of the coded frame of choosing the voice signal frame of issuing decoding side on the channel constantly changes, for take over party's code translator, voice coding module 102 and 101 coding be out of true all, and it also just can't receive accurate coded frame.

In sum, when adopting many speed of constant speed rate mode that the voice signal frame is carried out adaptive multi-rate coding, equally can cause the decoding error together with variable Rate multi-rate coding device because of the rate variation that is placed to the coded frame on the channel, and this rate variation has also related to ground unrest code rate (AMR-WB_SID), this is because its each voice coding module is encoded in strict accordance with specified coding speed, the key state value that does not just resemble the scrambler of many speed of variable Rate mode when VAD detects quiet period all voice coding modules resets, and makes all key state values of coding and decoding side get back to unanimity.And it is when in a single day this decoding error is transmitted frame by frame, bigger to the influence of speech acoustical quality.

No matter be variable bit rate or constant speed rate mode, AMR-WB (AMR-WB) scrambler of the self-adapting multi-rate narrowband of many speed (AMR-NB) scrambler or many speed all will be placed to the influence of the code rate rate variation of the coded frame of giving code translator on the channel, with respect to same voice signal, the code rate of the coded frame that the code rate of its voice coding module and code translator are received is inconsistent in conversion, and the present invention will suppress described code rate inconsistent to infringement that voice signal caused in conversion.

Technical scheme

Solution to the problems described above of the present invention is exactly: although the conversion of the code rate of the conversion of the code rate of each the voice coding module in the scrambler and the handled coded frame of code translator can be inconsistent, guarantee that each the voice coding module in the scrambler can be with reference to consistent pumping signal with code translator.

Can know from the introduction of background technology: by integer and mark pitch delay the pumping signal that comprises a last subframe in the excitation impact damper is carried out interpolation and obtain the self-adapting code book vector signal, this self-adapting code book vector signal is again by the signal that obtains after the self-adapting code book yield value amplification that quantizes, the signal that obtains after amplifying by the fixed codebook gain value that quantizes with the fixed code book vector signal superposes, and resulting signal is exactly a pumping signal.

It is that the decoded device of self-adaptive multi-speed coding frame that present frame produces is received that a voice coding module must be arranged in the scrambler, this voice coding module with the 4th subframe of previous frame and before pumping signal in the pumping signal of the synthetic present frame of pumping signal the time parameter used in used consistent the having of code translator: integer and mark pitch delay, the quantification gain of self-adapting code book vector, the modifying factor of fixed code book vector and quantification, because they are coded directly onto self-adaptive multi-speed coding frame and have suffered, therefore as long as this voice coding module of scrambler is also consistent in the quantification gain of another parameter-fixed code book vector with code translator, the pumping signal that this voice coding module and code translator synthesize based on consistent pumping signal is also in full accord, as long as zero defect takes place in the communication process between scrambler and the code translator, the voice coding module of scrambler and the consistance of the pumping signal between the code translator are transmitted frame by frame.

For the AMR-WB scrambler following technical scheme is arranged,

A kind of can it is characterized in that by constant speed rate AMR-WB (AMR-WB) coding method of a plurality of code rate codings:

Receive the appointment indication of the code rate of the previous input signal frame of encoding by a plurality of code rates;

Code rate according to described appointment indication is selected a pumping signal from the pumping signal of a plurality of speed of continuous 248 sample points that comprise the 4th subframe of the previous input signal frame of described appointment indication;

From the quantification energy predicting error of a plurality of speed of the previous input signal frame of described appointment indication, select the quantification energy predicting error of four subframes of a speed according to the code rate of described appointment indication;

Generate coding output according to described pumping signal of selecting and quantification energy predicting error.

For the scrambler of self-adapting multi-rate narrowband (AMR-NB), following technical scheme is arranged,

A kind of can it is characterized in that by constant speed rate self-adapting multi-rate narrowband (AMR-NB) coding method of a plurality of code rate codings:

Receive the appointment indication of the code rate of the previous input signal frame of encoding by a plurality of code rates;

Code rate according to described appointment indication is selected a pumping signal from the pumping signal of a plurality of speed of continuous 154 sample points that comprise the 4th subframe of the previous input signal frame of described appointment indication;

From the quantification energy predicting error of a plurality of speed of the previous input signal frame of described appointment indication, select the quantification energy predicting error of four subframes of a speed according to the code rate of described appointment indication;

Generate coding output according to described pumping signal of selecting and quantification energy predicting error.

All voice coding modules of scrambler have consistent original state value with code translator when starting, and therefore, their pumping signal and quantification energy predicting error are all in full accord.If scrambler fill order rate coding on this basis, all voice coding modules of scrambler be all with this single-rate coding, they produce same coded frame has the pumping signal that same state value and code translator be consistent and quantizes the energy predicting error.

Adaptive multi-rate (AMR) scrambler is coded directly onto the quantification gain and the fixed code book vector of integer and mark pitch delay, self-adapting code book vector in adaptive multi-rate (AMR) coded frame, directly fixed codebook gain is not encoded, but, also to calculate fixed code book prediction gain (or prediction gain) g ' after the modifying factor that decoder for decoding obtains quantizing to the modifying factor quantization encoding between fixed codebook gain and the fixed code book prediction gain (or prediction gain) _c, code translator is with fixed code book prediction gain (or prediction gain) g ' _cTake advantage of the modifying factor of quantification just to be fixed the quantification gain of codebook vector.

For the self-adapting multi-rate narrowband vocoder, the fixed code book prediction gain g ' shown in the formula (4) in front _cCalculating formula in have only the prediction energy (predicted energy) of subframe

Depend on the relevant state value-quantification energy predicting error of coding with the subframe of front, the value of mean value E of upgrading energy is only relevant with current code rate, on average upgrades ENERGY E _IOnly relevant with the fixed code book vector, about this point, the formula (55) in 5.8 joints of TS26.090-310 version has provided explanation.

For the AMR-WB vocoder, the prediction gain g ' shown in the formula (6) in front _cCalculating formula in correlated variables, have only the prediction energy (predicted energy) of subframe

Depend on the relevant state value-quantification energy predicting error of coding with the subframe of front, the value of mean value E of upgrading energy is constant, on average upgrades ENERGY E _iOnly relevant with the fixed code book vector, the formula (51) in 5.9 joints of the TS26.190-510 version of 3GPP has provided explanation.

The coded frame decoding that the self-adapting multi-rate narrowband code translator produces the voice coding module is because at the one and same coding frame, so the mean value E of code translator and these voice coding module both sides' renewal energy and the average ENERGY E of upgrading _IIn full accord, if use the quantification energy predicting error of four subframes of same previous frame to calculate the prediction energy of subframe

Both sides' fixed code book prediction gain g ' _cIn full accord.

Equally, the coded frame decoding that the AMR-WB code translator produces the voice coding module is because at the one and same coding frame, so the mean value E of code translator and these voice coding module both sides' renewal energy and the average ENERGY E of upgrading _iIn full accord, if use the quantification energy predicting error of four subframes of same previous frame, both sides' prediction gain g ' _cIn full accord.

When scrambler during for the first time with the mode coded speech signal frame of odd encoder speed, pumping signal that each voice coding module energy reference in the scrambler is consistent with code translator and quantification energy predicting error, and, the coded frame that always has a voice coding module to produce in the scrambler wants decoded device to receive, code translator directly obtains the pitch delay consistent with this voice coding module, consistent self-adapting code book quantification gain, consistent fixed code book vector and consistent quantification modifying factor from the coded frame of receiving, with fixed code book prediction gain (or prediction gain) g ' _cTake advantage of the modifying factor of consistent quantification to obtain quantizing fixed codebook gain; Because code translator and this voice coding module be with reference to consistent quantification energy predicting error and at same coded frame, their prediction gain (or fixed code book prediction gain) g ' _cIn full accord, so their quantification fixed codebook gain is also in full accord; Code translator is according to the pumping signal and the synthetic consistent self-adapting code book vector of pitch delay of the previous frame subframe of unanimity, and with self-adapting code book vector and fixed code book vector multiply by separately quantification gain back addition and as the pumping signal of new subframe, the pumping signal of the pumping signal of new subframe and this voice coding module is in full accord; Scrambler has the voice coding module of concord receiving the pumping signal that can't determine before specifying indication with the new subframe of code translator, does not wait these not need the encoding operation of pumping signal but do not influence its execution linear prediction and analysis and quantification; After scrambler obtains its appointment indication of the voice signal frame of multi-rate coding for the first time, the voice coding module of agreeing with the pumping signal of the new subframe of code translator is determined, and every other voice coding module is all with reference to the pumping signal of the subframe of this voice coding module.This process constantly repeats, so, scrambler can synthesize the pumping signal of the next frame subframe of new unanimity under the condition of the pumping signal of using the previous frame subframe consistent with code translator, the consistance of pumping signal also can be transmitted frame by frame, and the consistance of pumping signal also can obtain the maintenance of long period.

The renewal that produces in the time of should be able to using last subframe of coding previous frame during the coded speech signal frame to the pumping signal in all excitation impact dampers, promptly, voice coding module 101 among Fig. 1 and 102 should be able to be used the pumping signal in the other side's the excitation impact damper, the pumping signal of used last subframe of previous frame can freely be selected when making the coded speech signal frame, a pumping signal of last subframe that produces when the voice coding module is only used the last voice signal frame of self coding in this and the current single-rate scrambler has great difference, this be because, leave the pumping signal of excitation in the impact damper in for what produce in the process that is coded in a last adaptive multi-rate (AMR) frame that transmits on the channel, the excitation impact damper of depositing it can belong to any one the voice coding module in the scrambler.

The processing procedure with coding and decoding both sides when switching to 23.05kb/s speed again after single 23.05kb/s speed switches to the dual rate of 23.05kb/s and 8.85kb/s shown in Figure 1 in the technical matters that will solve with the front is an example, the pumping signal that is placed in the excitation impact damper that the 23.05kb/s of AMR-WB (AMR-WB) scrambler and the dual rate of 8.85kb/s coding produce has 2 kinds, concerning the next voice signal frame of encoding with the 23.05kb/s single-rate, these 2 kinds of pumping signals all can be used, by preamble before the output of this single-rate coding, have as can be known one both get the restriction of one, if before the search of the self-adapting code book after switching to 23.05kb/s speed again, the 8.85kb/s rate frame that scrambler receives in the dual rate coding output of 23.05kb/s and 8.85kb/s is placed to the information (specifying indication) that transmits on the channel, voice coding module 101 just is selected by pumping signal and the quantification energy predicting error that the 8.85kb/s rate coding is produced in above-mentioned 2 kinds of pumping signals, as effective pumping signal and effective energy predicting error that quantizes, effectively pumping signal and effective energy predicting error that quantizes are exactly that the take over party deciphers employed pumping signal and quantizes the energy predicting error, voice coding module 102 and 101 is all carried out the 23.05kb/s rate coding based on the quantification energy predicting error of the pumping signal of this 8.85kb/s speed, for code translator, the coded frame of 102 and 101 outputs all is based on same pumping signal and quantizes the energy predicting error, thereby multi-path voice code multiplexing module is got, and any one can in 102 and 101 coded frame of exporting.

According to technique scheme, but the rate variation because of being placed to the coded frame on the channel of the adaptive multi-rate coding device of multi-rate coding has reduced the error that code translator causes, this be because: the pumping signal of the subframe of the previous frame that all voice coding module is can reference consistent with code translator in the scrambler; The synthetic pumping signal of the parameter that provides in the self-adaptive multi-speed coding frame that code translator is received by the quantification energy predicting sum of errors of unanimity, the self-adaptive multi-speed coding frame that code translator is received is that certain the voice coding module in the scrambler is sent, the synthetic used parameter of pumping signal of this voice coding module just has it to be encoded to parameter in the self-adaptive multi-speed coding frame that code translator receives, so the synthetic pumping signal of code translator still is consistent with the pumping signal that this voice coding module generates.

It is worthy of note that above-mentioned these schemes are applicable to that not merely a plurality of speed in the coded command all are the situations of non-ground unrest code rate, the situation that comprises the ground unrest code rate for a plurality of speed in the coded command is suitable equally, do not comprise pitch delay in quiet description (SID) frame, the self-adapting code book gain, these parameters of fixed code book vector and modifying factor, but carry out the voice coding module of quiet description (SID) frame coding and the code translator of quiet description (SID) frame coding is all reset to initial value with separately pumping signal and quantification energy predicting error.So, although quiet description (SID) frame coding does not need the pumping signal of previous frame and quantizes the energy predicting error, the coding of next active speech frame of quiet description (SID) frame is needed this active speech frame previous frame pumping signal and quantize the energy predicting error.

After the dual rate coding of ground unrest code rate and active speech code rate is finished, next frame coding is to use the pumping signal that is reset because of quiet description (SID) frame coding and quantizes the energy predicting error and also is to use the pumping signal that produces because of active speech code rate coding and quantizes the energy predicting error equally by specifying the indication decision, because the appointment indication of the previous voice signal frame of encoding with ground unrest code rate and the such dual rate mode of active speech code rate can provide the code rate of the coded frame of this voice signal frame that transmits on the channel, this code rate be exactly next frame coding institute based on pumping signal and the speed of quantification energy predicting error, in this article, when the speed of pumping signal is meant voice signal frame coding is produced this pumping signal institute according to code rate, when the speed that quantizes the energy predicting error is meant this quantification energy predicting error of voice signal frame coding generation according to code rate.

The processing procedure with coding and decoding both sides when switching to 23.05kb/s speed again after single 23.05kb/s speed switches to the dual rate of 23.05kb/s and AMR-WB_SID shown in Figure 1 in the technical matters that will solve with the front is an example, the pumping signal that is placed in the excitation impact damper that the 23.05kb/s of AMR-WB (AMR-WB) scrambler and the dual rate of AMR-WB_SID coding produce has 2 kinds, concerning the next voice signal frame of encoding with the 23.05kb/s single-rate, these 2 kinds of pumping signals all can be used, by preamble before the output of this single-rate coding, have as can be known one both get the restriction of one, if before the search of the self-adapting code book after switching to 23.05kb/s speed again, the SID frame that scrambler receives in the dual rate coding output of 23.05kb/s and AMR-WB_SID is placed to the information (specifying indication) that transmits on the channel, in above-mentioned 2 kinds of pumping signals voice coding module 101 by the AMR-WB_SID rate coding produced reset after pumping signal and quantize the energy predicting error and just be selected, as effective pumping signal and effective energy predicting error that quantizes, effectively pumping signal and effective energy predicting error that quantizes are exactly that the take over party deciphers employed pumping signal and quantizes the energy predicting error, voice coding module 102 and 101 is all carried out the 23.05kb/s rate coding based on the quantification energy predicting error of the pumping signal of this AMR-WB_SID speed, for code translator, the coded frame of 102 and 101 outputs all are based on same pumping signal and quantize the energy predicting error, thus multi-path voice code multiplexing module get 102 with the coded frame of 101 outputs in any one can (because code translator also use pumping signal and quantification energy predicting error after consistent key state value resets).

Because the present invention is the fixing encoder scheme of code rate of non-variable bit rate, the SID frame of voice signal frame generates in yard device of not being on the permanent staff automatically and realizes, therefore scrambler can be as the quiet description of the automatic generation of variable rate coder (SID) frame, has only when the code rate in the coded command is the ground unrest code rate scrambler just can export quiet description (SID) frame.

When comprising ground unrest code rate and other code rate in the coded command simultaneously, the coding that outputs to the wireless access part has comprised active speech coded frame and ground unrest coded frame (quiet description (SID) frame), wireless access part can adopt the coded frame of which code rate according to the synthetic video signal frame decision after the decoding of the coded signal frame of non-ground unrest code rate, and such way is to the minimum that influences of speech quality.Therefore, encode by non-ground unrest code rate needed linear prediction and quantification, self-adapting code book search and the resulting pumping signal of fixed codebook search externally can be exported by the synthetic digital audio signal that the determined linear prediction synthesis filter of linear prediction is produced.

Content below in the technical scheme of AMR-WB scrambler, increasing:

Export the synthetic digital voiced frame of non-ground unrest code rate when in described a plurality of code rates, comprising the ground unrest code rate.

Content below in the technical scheme of self-adapting multi-rate narrowband scrambler, increasing:

Export the synthetic digital voiced frame of non-ground unrest code rate when in described a plurality of code rates, comprising the ground unrest code rate.

The technical scheme of the device about the multi-rate coding that comprises the ground unrest code rate of the present invention is as follows, for the scrambler of AMR-WB type (AMR-WB) and self-adapting multi-rate narrowband (AMR-NB) technology following technical scheme is arranged:

A kind of constant speed rate AMR-WB (AMR-WB) scrambler of pressing ground unrest code rate and non-ground unrest code rate coding, it is characterized in that it comprises: reception is by the device of the appointment indication of the code rate of the previous input signal frame of two code rates codings;

Thereby the current input signal frame is carried out the device that linear prediction, self-adapting code book search and fixed codebook search generate synthetic digital audio signal frame by the non-ground unrest code rate in described two code rates;

From the pumping signal of the 4th subframe of the previous input signal frame of described appointment indication and two speed before thereof, select the device of a pumping signal according to the code rate of described appointment indication;

From the quantification energy predicting error of two speed of the previous input signal frame of described appointment indication, select the device of quantification energy predicting error of four subframes of a speed according to the code rate of described appointment indication, identical before the quantification energy predicting sum of errors that it may be noted that its four subframes after AMR-WB_SID frame coding is finished is here encoded;

Generate the device of coding output according to described pumping signal of selecting and the quantification energy predicting error of selecting.

A kind of constant speed rate self-adapting multi-rate narrowband (AMR-NB) scrambler of pressing ground unrest code rate and non-ground unrest code rate coding, it is characterized in that it comprises: reception is by the device of the appointment indication of the code rate of the previous input signal frame of two code rates codings;

Thereby the current input signal frame is carried out the device that linear prediction, self-adapting code book search and fixed codebook search generate synthetic digital audio signal frame by the non-ground unrest code rate in described two code rates;

From the pumping signal of the 4th subframe of the previous input signal frame of described appointment indication and two speed before thereof, select the device of a pumping signal according to the code rate of described appointment indication;

From the quantification energy predicting error of two speed of the previous input signal frame of described appointment indication, select the device of quantification energy predicting error of four subframes of a speed according to the code rate of described appointment indication;

Generate the device of coding output according to described pumping signal of selecting and the quantification energy predicting error of selecting.

Beneficial effect

But any one pumping signal in the pumping signal that the adaptive multi-rate coding device that the invention enables multi-rate coding produces in the time of can encoding to a last voice signal frame respectively with its a plurality of voice coding modules and the quantification energy predicting error of any one speed are that present frame carries out adaptive multi-rate coding, and in case by specifying indication to receive the speed that is placed in the self-adaptive multi-speed coding frame that sends code translator on the channel in this previous frame multi-rate coding output sequence, just will be pumping signal that this rate coding produced immediately as effective pumping signal, immediately for quantification energy predicting error that this rate coding produced as effective quantification energy predicting error.Like this, a voice coding module in take over party's code translator and the scrambler, effective pumping signal of the previous frame subframe of reference unanimity respectively and effective energy predicting error that quantizes, parameter in the coded frame of receiving on the one side use channel, the opposing party uses and oneself is encoded to the parameter of going in this coded frame, generate the pumping signal and the synthetic speech of subframe separately respectively, thus take over party's code translator synthetic pumping signal and the described voice coding module pumping signal of synthesizing in full accord.The synthetic speech that produces based on the decoder for decoding of the pumping signal of unanimity can reduce original sound preferably.The invention enables scrambler to use and the on all four pumping signal of code translator, thereby farthest eliminated the influence of the decoding error that causes because of the code rate variation that is placed to the coded frame on the channel.The present invention also makes constant speed rate scrambler to use quiet description (SID) frame to suppress to decipher the transmission frame by frame of error in quiet period, and this is extremely important for the scrambler that does not use silence description frames to generate automatically.

Description of drawings

The theory diagram of AMR-WB (AMR-WB) scrambler that but Fig. 1 is the dual rate of variable Rate to encode.

But Fig. 2 is the theory diagram of AMR-WB (AMR-WB) scrambler of constant speed rate dual rate coding.

But Fig. 3 is the theory diagram that has the self-adapting multi-velocity encoder with fixed velocity of the dual rate coding of specifying the indication receiving interface, is applicable to broadband (AMR-WB) scrambler and arrowband (AMR-NB) scrambler.

But the theory diagram of constant speed rate AMR-WB (AMR-WB) scrambler of the dual rate that the band speech quiet period that Fig. 4 is detects coding, it also is applicable to arrowband (AMR-NB) scrambler.

Embodiment

Embodiment 1, but AMR-WB (AMR-WB) scrambler of a dual rate coding, as shown in Figure 3, the voice signal frame 1 of band frame number is simultaneously to pattern 1 coding module of voice coding module 102 and the pattern 2 coding modules output of voice coding module 101, pattern 1 coding module together sends to multi-path voice code multiplexing module with adaptive multi-rate (AMR) coded frame 12 and adaptive multi-rate (AMR) frame type signal 11, and adaptive multi-rate (AMR) frame type signal 11 is sent to excitation and energy error indicating module separately; Pattern 2 coding modules together send to multi-path voice code multiplexing module with adaptive multi-rate (AMR) coded frame 19 and adaptive multi-rate (AMR) frame type signal 16, and adaptive multi-rate (AMR) frame type signal 16 is sent to excitation and energy error indicating module separately.The speed and the frame number of adaptive multi-rate (AMR) frame type signal 11 indication self-adaptive multi-speed coding frames 12, the speed and the frame number of adaptive multi-rate (AMR) frame type signal 16 indication self-adaptive multi-speed coding frames 19, self-adaptive multi-speed coding frame 12 and 19 speed depend on code rate indicator signal 17 and 18 respectively, 2 tunnel output signals that send to the voice coding module of code rate indicator signal 17 and 18 1 speed 1 tunnel that to be code rate control modules resolve into the speed in the code rate command signal 9.The adaptive multi-rate speech coding frame 12 and 19 that multi-path voice code multiplexing module is identical with frame number is merged into and comprises frame number, the Speech frame coded sequence 2 of speed indication and information bit, the method that merges is: the frame number that Speech frame coded sequence 2 is set is the identical frame number in adaptive multi-rate (AMR) the frame type signal 11 and 16, and, when the speed of adaptive multi-rate (AMR) frame type signal 11 and 16 expressions is inequality, self-adaptive multi-speed coding frame 12 and 19 together is put in the information bit of Speech frame coded sequence 2, the speed with adaptive multi-rate (AMR) frame type signal 11 and 16 is put in the speed indication of Speech frame coded sequence 2 together simultaneously; When adaptive multi-rate (AMR) frame type signal 11 is identical with 16 speed of representing, in the optional self-adaptive multi-speed coding frame 12 and 19 one is put in the information bit of Speech frame coded sequence 2, and the speed that Speech frame coded sequence 2 is set simultaneously is designated as described identical speed.Pattern 1 coding module and pattern 2 coding modules are respectively in the excitation impact damper of the voice coding module of output drive signal 31 and 34 under separately, the excitation impact damper of voice coding module 102 in voice coding module 102 pattern 1 coding module and pattern 2 coding modules in the voice coding module 101 pumping signal 33 is provided.The excitation impact damper of voice coding module 101 in voice coding module 101 pattern 2 coding modules and pattern 1 coding module in the voice coding module 102 pumping signal 35 is provided.The pattern 1 coding module output of voice coding module 102 quantizes the quantification energy predicting error buffer of energy predicting error 30 to voice coding module 102, the output of pattern 2 coding modules of voice coding module 101 quantizes the quantification energy predicting error buffer of energy predicting error 36 to voice coding module 101, voice coding module 102 quantification energy predicting error buffer provide to pattern 2 coding modules of pattern 1 coding module of voice coding module 102 and voice coding module 101 and quantize energy predicting error 32.The quantification energy predicting error buffer of voice coding module 101 provides quantification energy predicting error 37 to pattern 2 coding modules of voice coding module 101 and pattern 1 coding module of voice coding module 102, excitation and energy error indicating module receive the appointment indicator signal 20 of the code rate of previous input signal frame, the appointment indicator signal 20 of the code rate of previous input signal frame has comprised frame number and speed, excitation and energy error indicating module are with the frame number in same adaptive multi-rate (AMR) the frame type signal of receiving in the past 11 and 16 of the frame number in the indicator signal 20 and speed and speed ratio, when the frame number in the indicator signal 20 and speed during with identical in adaptive multi-rate (AMR) the frame type signal of receiving in the past 11, send excitation and energy error selection instruction 21 and 22, indicate pattern 1 coding module and pattern 2 coding modules in the voice coding module 101 in the voice coding module 102 to encode respectively based on pumping signal in the excitation impact damper in the voice coding module 102 and the quantification energy predicting error amount that quantizes in the energy predicting error buffer; When the frame number in the indicator signal 20 and speed during with identical in adaptive multi-rate (AMR) the frame type signal of receiving in the past 16, send excitation and energy error selection instruction 21 and be clipped to voice coding module 102 and 101 in 22 minutes, indicate pattern 1 coding module and pattern 2 coding modules in the voice coding module 101 in the voice coding module 102 to encode respectively based on pumping signal in the excitation impact damper in the voice coding module 101 and the quantification energy predicting error amount that quantizes in the energy predicting error buffer.

Scrambler shown in Figure 3 begins coding from original state, and first voice signal frame is with the 23.05kb/s rate coding, and second voice signal frame and later frame all are the dual rate codings with 23.05kb/s and 8.85kb/s.Coding output for second voice signal frame---comprise the coded sequence of 23.05kb/s and two kinds of speed coding frames of 8.85kb/s, have only the coded frame of 8.85kb/s speed to be sent on the channel; For the 3rd and afterwards voice signal frame coding output---comprise the coded sequence of 23.05kb/s and two kinds of speed coding frames of 8.85kb/s equally, the coded frame of 23.05kb/s speed is sent on the channel.In the present embodiment, use the wireless access part of Speech frame coded sequence 2 before the voice coding module of scrambler begins the self-adapting code book search, just the appointment indication of the code rate of a frame on previous to be sent to scrambler.

Form 1

Coding implementation status when form 1 has illustrated above-mentioned first frame to the, four frames, wherein the content of each row the 3rd row " excitation and energy error selection instruction " is relatively to produce the back according to content of last row " record (voice coding module; frame number; speed) in excitation and the energy error indicating module " of lastrow and the content that secondary series " is specified the content (frame number; speed) of indication ", this just encourage and the energy error indicating module according to the adaptive multi-rate frame type signal that receives with specify the performed operation in indication back.Here, the code rate control module is when receiving the dual rate coded order, sending code rate indicator signal 17 allows voice coding module 102 carry out the coding task of 23.05kb/s, send code rate indicator signal 18 and allow voice coding module 101 carry out the coding task of 8.85kb/s, but this is not unique selection of code rate control module.

Voice coding module in the scrambler is when each subframe of each voice signal frame of coding, all to use the pumping signal of a subframe, voice coding module 102 produces identical coded frame and state value with same code rate 23.05kb/s to first frame coding with identical original state with 101, and code translator all is consistent with any one pumping signal and quantification energy predicting error in two voice coding modules.Then, voice coding module 102 produces coded frame and pumping signal with 23.05kb/s to second frame coding, voice coding module 101 produces coded frame and pumping signal with 8.85kb/s to second frame coding, because coded frame the 3rd frame that only has voice coding module 101 to produce in the output of the coding of second frame is placed on the channel, so till second frame, code translator and voice coding module 101 are in full accord on rate transition, and pumping signal between the two and quantification energy predicting error are in full accord.During the 3rd frame coding, there are the pumping signal and the quantification energy predicting error of the subframe of 23.05kb/s and 8.85kb/s second frame to use in the scrambler, after the appointment of receiving indication provides pumping signal and quantification energy predicting error with the 8.85kb/s of second frame, just can be to select the 4th subframe of second frame of 23.05kb/s and 8.85kb/s rate coding and the pumping signal before thereof with the 4th subframe of second frame of 8.85kb/s rate coding and pumping signal before thereof, the voice coding module 102 that is operated in 23.05kb/s calculates integer and mark pitch delay according to this pumping signal (pumping signal that produces during voice coding module 101 codings second frame), and this pumping signal is carried out the self-adapting code book vector signal that interpolation obtains first subframe of the 3rd frame by described integer and mark pitch delay, and then calculate self-adapting code book gain and quantize, then, carry out fixed codebook search, finally be fixed the fixed codebook gain of codebook vector and quantification; Above-mentioned integer and mark pitch delay, self-adapting code book quantize gain, the fixed code book vector directly is encoded in the AMR-WB coded frame of voice coding module 102 outputs, and the fixed codebook gain of quantification then is encoded in the AMR-WB coded frame indirectly with the form of the modifying factor that quantizes; The voice coding module 101 that is operated in 8.85kb/s is carried out similar operation according to same pumping signal (pumping signal that produces during voice coding module 101 codings second frame).Find out that from form 1 what be placed in the coding output of the 3rd voice signal frame that channel transmits is the coded frame of the AMR-WB 23.05kb/s speed of voice coding module 102 outputs, but direct decoding obtained integer and mark pitch delay after code translator was received this AMR-WB coded frame, self-adapting code book quantizes gain, with the fixed code book vector, and the fixed codebook gain of taking advantage of prediction gain to obtain quantizing with the modifying factor that direct decoding obtains, the quantification energy predicting error amount of four subframes of second frame that the quantification energy predicting error amount of the subframe of four second frames that code translator uses and voice coding module 101 produce when finishing second frame coding is identical, the quantification energy predicting error amount of the subframe of four second frames that use during just with voice coding module 102 codings the 3rd frame is identical, according in the technical scheme to the explanation of the calculating of prediction gain, the fixed codebook gain of the quantification of using when the modifying factor that obtains of code translator direct decoding is taken advantage of fixed codebook gain that prediction gain obtains quantizing and voice coding module 102 codings as can be seen is identical, code translator with and the consistent pumping signal of pumping signal of used second frame the 4th subframe of voice coding module 102 and 8.85kb/s before thereof synthesize the pumping signal of the 3rd frame first subframe according to these parameters, based on pumping signal reconstruct synthetic speech again, the pumping signal of other subframes of the 3rd frame and synthetic speech also just can draw by the pumping signal of last subframe and the calculation of parameter in the coded frame.So the pumping signal of synthetic the 3rd frame of code translator and the pumping signal of voice coding module 102 Synthetic 2 3.05kb/s have also obtained consistent.The quantification energy predicting error of four subframes of the 3rd frame of code translator is determined that by the modifying factor that direct decoding obtains the quantification energy predicting error that code translator generates quantification energy predicting sum of errors voice coding module 102 generations of the 3rd frame has also obtained unanimity.So the pumping signal of code translator all the time with scrambler in a voice coding module pumping signal and quantize the energy predicting error and be consistent, code translator and some voice coding modules use the synthetic speech of the linear prediction analysis quantization parameter reconstruct in the same coded frame just very approaching based on consistent pumping signal respectively, so code translator can reduce the original speech signal preferably.

But embodiment 2-band shown in Figure 4 synthesizes self-adapting multi-rate narrowband (AMR-NB) scrambler of the dual rate coding of digital voice signal, and it is the synthetic digital voice signal 8 of the submodule prediction synthesis filter module output drive signal 33 of pattern 1 coding module through filtered generation than difference shown in Figure 3.The input of the voice signal frame 1 of code rate control module receiving belt frame number and the input of rate command signal 9, when comprising AMR_SID speed in the rate command signal 9, the output of code rate control module allows prediction synthesis filter export the order 39 of synthetic digital signal 8.

Claims (6)

1. one kind can is characterized in that by the constant speed rate self-adapting multi-rate narrowband coding method of a plurality of code rate codings:

Receive the appointment indication of the code rate of the previous input signal frame of encoding by a plurality of code rates;

From the pumping signal of a plurality of code rates of continuous 154 sample points that comprise the 4th subframe of the previous input signal frame of described appointment indication, select the pumping signal of a code rate according to the code rate of described appointment indication;

From the quantification energy predicting error of a plurality of speed of the previous input signal frame of described appointment indication, select the quantification energy predicting error of four subframes of a code rate according to the code rate of described appointment indication;

Generate coding output according to described pumping signal of selecting and the described quantification energy predicting error of selecting.

2. according to the method for claim 1, it is characterized in that:

Detect described a plurality of code rate and whether comprise the ground unrest code rate, if testing result is for being then to export the synthetic digital signal frame of non-ground unrest code rate.

3. one kind can is characterized in that by the constant speed rate AMR-WB coding method of a plurality of code rate codings:

Receive the appointment indication of the code rate of the previous input signal frame of encoding by a plurality of code rates;

From the pumping signal of a plurality of code rates of continuous 248 sample points that comprise the 4th subframe of the previous input signal frame of described appointment indication, select the pumping signal of a code rate according to the code rate of described appointment indication;

From the quantification energy predicting error of a plurality of code rates of the previous input signal frame of described appointment indication, select the quantification energy predicting error of four subframes of a code rate according to the code rate of described appointment indication;

Generate coding output according to described pumping signal of selecting and the described quantification energy predicting error of selecting.

4. according to the method for claim 3, it is characterized in that:

Detect described a plurality of code rate and whether comprise the ground unrest code rate, if testing result is for being then to export the synthetic digital signal frame of non-ground unrest code rate.

5. constant speed rate self-adapting multi-rate narrowband scrambler of pressing ground unrest code rate and non-ground unrest code rate coding, it is characterized in that it comprises: reception is by the device of the appointment indication of the code rate of the previous input signal frame of two code rates codings;

Thereby the current input signal frame is carried out the device that linear prediction, self-adapting code book search and fixed codebook search generate synthetic digital audio signal frame by the non-ground unrest code rate in described two code rates;

From the pumping signal of two code rates of continuous 154 sample points that comprise the 4th subframe of the previous input signal frame of described appointment indication, select the device of the pumping signal of a code rate according to the code rate of described appointment indication;

From the quantification energy predicting error of two code rates of the previous input signal frame of described appointment indication, select the device of quantification energy predicting error of four subframes of a code rate according to the code rate of described appointment indication;

Generate the device of coding output according to described pumping signal of selecting and the quantification energy predicting error of selecting.

6. constant speed rate AMR-WB scrambler of pressing ground unrest code rate and non-ground unrest code rate coding, it is characterized in that it comprises: reception is by the device of the appointment indication of the code rate of the previous input signal frame of two code rates codings;

Thereby the current input signal frame is carried out the device that linear prediction, self-adapting code book search and fixed codebook search generate synthetic digital audio signal frame by the non-ground unrest code rate in described two code rates;

From the pumping signal of two code rates of continuous 248 sample points that comprise the 4th subframe of the previous input signal frame of described appointment indication, select the device of the pumping signal of a code rate according to the code rate of described appointment indication;

From the quantification energy predicting error of two code rates of the previous input signal frame of described appointment indication, select the device of quantification energy predicting error of four subframes of a code rate according to the code rate of described appointment indication;

Generate the device of coding output according to described pumping signal of selecting and the quantification energy predicting error of selecting.

Images