CN104751851B - It is a kind of based on the front and rear frame losing error concealment method and system to Combined estimator - Google Patents
It is a kind of based on the front and rear frame losing error concealment method and system to Combined estimator Download PDFInfo
- Publication number
- CN104751851B CN104751851B CN201310747005.6A CN201310747005A CN104751851B CN 104751851 B CN104751851 B CN 104751851B CN 201310747005 A CN201310747005 A CN 201310747005A CN 104751851 B CN104751851 B CN 104751851B
- Authority
- CN
- China
- Prior art keywords
- frame losing
- voice data
- current demand
- vowel
- demand signal
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 31
- 238000005562 fading Methods 0.000 claims abstract description 14
- 230000009467 reduction Effects 0.000 claims abstract description 11
- 238000004458 analytical method Methods 0.000 claims description 45
- 238000006243 chemical reaction Methods 0.000 claims description 9
- 230000007774 longterm Effects 0.000 claims description 9
- 230000000737 periodic effect Effects 0.000 claims description 9
- 230000004048 modification Effects 0.000 claims description 6
- 238000012986 modification Methods 0.000 claims description 6
- 125000002950 monocyclic group Chemical group 0.000 claims description 6
- 238000013213 extrapolation Methods 0.000 description 6
- 241000894007 species Species 0.000 description 6
- 230000008859 change Effects 0.000 description 4
- 230000015572 biosynthetic process Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 235000013399 edible fruits Nutrition 0.000 description 3
- 230000005284 excitation Effects 0.000 description 3
- 238000005086 pumping Methods 0.000 description 3
- 238000003786 synthesis reaction Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 206010002953 Aphonia Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 238000012856 packing Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000035807 sensation Effects 0.000 description 1
- 230000005236 sound signal Effects 0.000 description 1
Abstract
Present invention offer is a kind of based on the front and rear frame losing error concealment method and system to Combined estimator, including：Cache the two frame voice data played and the six frame voice data that will be played；When detecting generation frame losing, estimated to reduce frame losing voice data backward according to the voice data before frame losing；Estimated forward to reduce frame losing voice data according to the voice data after frame losing；Estimate according to the voice data of estimation reduction frame losing backward and that forward the voice data for reducing frame losing carries out the voice estimation release signal that cross-fading generation is lost.The present invention utilizes voice short-term stationarity feature, carries out Combined estimator to frame losing according to the voice messaging before and after frame losing, can more accurately estimate the speech frame of loss, the compensation to frame losing is more accurate.
Description
Technical field
It is more particularly to a kind of based on the front and rear frame losing error concealment side to Combined estimator the present invention relates to communication technique field
Method and system.
Background technology
Discarded Packets compensation (Packet Loss Concealment, PLC) algorithm, also referred to as frame error concealment (Frame
Erasure Concealment, FEC) algorithm.Since the problem of network quality may result in loss voice in voice communication
Data, the voice stream data for bulk occur disappears, so that the auditory response that voice can be caused to stop abruptly, causes not connecting for voice
It is continuous.The voice data of loss can be estimated when this come if compensating the data of these loss, then can just eliminate by
In the voice quality loss that loss of voice is brought.This is also the basic function of PLC, estimates the voice signal lost, then uses it
Fill the lack part of voice flow.
Discarded Packets compensation is mainly used for packing in coding, then by network transmission, in the decoded audio system of receiving terminal, such as
Fruit there are packet loss phenomenon, then carries out Discarded Packets compensation in this process.In many measured code excited linear predictive (Code
Exited Linear Prediction, CELP) audio coder ＆ decoder (codec), there is Discarded Packets compensation function.Such as：G.729, amr, efr
Deng.But for the codec that some non-CELP are encoded there is no Discarded Packets compensation algorithm, such as：G.711, G.722 etc..So
These codecs need to use Discarded Packets compensation algorithm.
Since voice has short-term stationarity, the voice letter of estimation loss can be carried out according to its neighbouring signal
Number.According to the viewpoint of CELP, voice has two parts：Sound channel and driving source.It is described mainly to be modeled by pole model,
Channel parameters are extracted by linear prediction.The pumping signal of the voice has two kinds：One kind is periodic excitation signal source, in addition
One kind is arbitrary excitation (white noise).Both signal constitutes information source excitation e (n).Composite filter is passed through by pumping signal
It is final to produce voice signal.The language of loss can be predicted by the short-term stationarity characteristic of this basic sonification model and voice
Sound signal part.
In above-mentioned estimation, two parts are usually estimated：First, linear predictor coefficient (Linear prediction
coefficient)；2nd, pumping signal.These can be estimated by neighbouring signal.
Most of in existing PLC algorithms is all to use extrapolation (Extrapolation), as shown in Figure 1, it is typical case
Extrapolation error concealment algorithm flow.Whole process is divided into two parts：Analysis part and comprehensive part.
The analysis part mainly estimates pitch period (T using the signal before frame losing0), linear predictor coefficient (A
(z)), residual signals (e (n)) and Modulation recognition (Class), for integrating when, estimate the voice messaging of loss.Wherein：
A (z)=1+a1z-1+a2z-2+...+a8z-8
Mainly the information of loss is estimated the comprehensive part.
First determine whether current demand signal is vowel, pitch period continuation if it is is carried out to residual signals e (n)；Such as
Fruit is not to modify to residual signals e (n) and carry out periodic extension again.
Then the signal after synthesis filter acquisition 1/A (z) synthesis is input to amended residual signals e ' (n) to obtain
y(n)。
Finally, according to the length computation fading gain g of current demand signal classification and frame losingmute, and to filter output signal
Carry out decay y ' (n)=y (n) gmute(n), the signal y ' (n) of frame losing estimation is finally obtained.
But since characteristic voice has short-term stationarity.General voice is stable in 10ms~30ms.Utilize extrapolation
The speech capability that method prediction is lost is very limited amount of, and hardly possible obtains the Long-term change trend of lost frames, such as：Fundamental tone changes, changes in amplitude
Deng.So the foreseeable frame number of extrapolation is considerably less, once prediction frame number becomes more, can cause to estimate inaccurate influence voice
Quality.
So most at present all can use fading gain to be used to decay to the voice signal after prediction, prevent
The frame number of prediction excessively causes forecasting inaccuracy below really to make voice quality degradation on the contrary.So existing Discarded Packets compensation algorithm energy
Power is extremely limited, is at best able to the speech frame of 2~3 frame of estimation (mono- frames of 20ms) loss, otherwise occurs since prediction error goes out
Cash dominant.The metallic sound refers to because frame losing is excessive, and when Discarded Packets compensation is done, due to the voice signal of estimation, its is same
A kind of signal short repetition duration in cycle is long, causes the phenomenon of Cool Metal sound on human auditory system.
The content of the invention
It is an object of the invention to provide a kind of based on the front and rear frame losing error concealment method and system to Combined estimator, with
It is very limited to solve the speech capability that technological prediction is lost, once prediction frame number becomes more, can cause to estimate that inaccuracy influences voice
The problem of quality.
In order to solve the above technical problems, present invention offer is a kind of based on the front and rear frame losing error concealment side to Combined estimator
Method, including：
Cache the two frame voice data played and the six frame voice data that will be played；
When detecting generation frame losing, estimated to reduce frame losing voice data backward according to the voice data before frame losing；According to
Voice data after frame losing is estimated to reduce frame losing voice data forward；
Intersection is carried out according to the voice data of estimation reduction frame losing backward and the forward voice data of estimation reduction frame losing to decline
Subtract the voice estimation release signal that generation is lost.
Preferably, described based in the front and rear frame losing error concealment method to Combined estimator, it is described according to frame losing before
Voice data estimate backward reduce frame losing voice data the step of include：
Linear prediction analysis filter obtains filter parameter；Estimate the pitch period of current demand signal；To current demand signal into
Row classification, the species of current demand signal include vowel state, voiceless sound state and conversion state；
Current analysis filter is obtained according to the filter parameter that the linear prediction analysis filter obtains；It is if current
Signal is vowel state, then repeats to estimate the residual signals of current demand signal by the pitch period, if current demand signal is non-vowel
State draws the residual signals of current demand signal after then modifying to the residual signals of current demand signal；Different classes of according to frame losing is adopted
The method to be decayed with piecewise linearity decays；
The voice data part of frame losing is estimated according to the residual signals of the current analysis filter and current demand signal；When
Voice signal after frame losing is not vowel, then decays, and the voice after frame losing is vowel, then without decay
Keep initial value, the final estimation signal for obtaining frame losing.
Preferably, described based in the front and rear frame losing error concealment method to Combined estimator, the linear prediction divides
It is 8 rank linear prediction analysis filters to analyse wave filter.
Preferably, described based in the front and rear frame losing error concealment method to Combined estimator, it is described according to frame losing after
Voice data estimate forward reduce frame losing voice data the step of include：
Classify to frame losing, the species of current demand signal includes vowel state, voiceless sound state and conversion state；Estimate current demand signal
Pitch period；
A monocyclic estimation signal is obtained, the frame losing information of estimation is obtained by periodic extension forward；
Judge whether the voice data after frame losing is vowel state, if vowel state, then open output control switch, if
It is not vowel state, then closes output control switch.
Preferably, described a single-revolution is being obtained based in the front and rear frame losing error concealment method to Combined estimator
In the step of estimation signal of phase, the frame losing information estimated by the acquisition of periodic extension forward, only to the voice after frame losing
Data are that the voice data of vowel state is estimated.
Correspondingly, the present invention also provides a kind of based on the front and rear frame losing error concealment system to Combined estimator, including：
Buffer part, for caching the two frame voice data played and the six frame voice data that will be played；
Back forecast part, for being estimated to reduce frame losing voice data backward according to the voice data before frame losing；
Forward prediction part, for being estimated forward to reduce frame losing voice data according to the voice data after frame losing；
Cross-fading part, for carrying out cross-fading life according to the back forecast part and the forward prediction part
Voice into loss estimates release signal.
Preferably, described based in the front and rear frame losing error concealment system to Combined estimator, the back forecast portion
Dividing includes：
Linear prediction analysis unit, filter parameter is obtained for linear prediction analysis filter；
Long-term prediction unit, for estimating the pitch period of current demand signal；
Signal classifier, for classifying to current demand signal, the species of current demand signal includes vowel state, voiceless sound state and turns
Change state；
Analysis filter, the filter parameter for being obtained according to the linear prediction analysis filter obtain current point
Analyse wave filter；
Pitch period repeats and residual signals modification unit, if being vowel state for current demand signal, passes through the fundamental tone
Cycle repeats to estimate the residual signals of current demand signal, and the residual signals of current demand signal are carried out if current demand signal is non-vowel state
The residual signals of current demand signal are drawn after modification；
Decay factor computing unit, is declined for the different classes of method to be decayed using piecewise linearity according to frame losing
Subtract；
Linear prediction integrates, for estimating frame losing according to the residual signals of the current analysis filter and current demand signal
Voice data part；
Binary decision unit, for frame losing after voice signal be not vowel, then decay, after frame losing
Voice signal be vowel, then without decay keep initial value；
Attenuation units, for generating the estimation signal of frame losing.
Preferably, described based in the front and rear frame losing error concealment system to Combined estimator, in linear prediction analysis
In unit, the linear prediction analysis filter is 8 rank linear prediction analysis filters.
Preferably, described based in the front and rear frame losing error concealment system to Combined estimator, the forward prediction portion
Dividing includes：
Signal classifier, for classifying to frame losing, the species of current demand signal includes vowel state, voiceless sound state and conversion
State；
Long-term prediction unit, for estimating the pitch period of current demand signal；
Fundamental tone repeats estimation unit, for obtaining a monocyclic estimation signal, is estimated by periodic extension forward
The frame losing information of meter；
Vowel decision unit, for judging frame losing after voice data whether be vowel state；
Output control switch, if vowel state, then open output control switch, if not vowel state, then close output control
System switch.
Preferably, repeat to estimate in fundamental tone based in the front and rear frame losing error concealment system to Combined estimator described
It is only that the voice data of vowel state is estimated to the voice data after frame losing in unit.
It is provided by the invention based on the front and rear frame losing error concealment method and system to Combined estimator, have below beneficial to effect
Fruit：The present invention utilizes voice short-term stationarity feature, and Combined estimator is carried out to frame losing according to the voice messaging before and after frame losing, can be more accurate
The true speech frame for estimating loss, the compensation to frame losing are more accurate.
Brief description of the drawings
Fig. 1 is existing extrapolation error concealment algorithm flow；
Fig. 2 is the signal based on the front and rear frame losing error concealment method and system to Combined estimator of the embodiment of the present invention
Figure；
Fig. 3 is that the decay based on the front and rear frame losing error concealment method and system to Combined estimator of the embodiment of the present invention is bent
Line schematic diagram；
Fig. 4-5 is the comparison diagram of the present invention and other embodiment.
Embodiment
Below in conjunction with the drawings and specific embodiments to proposed by the present invention hidden based on the front and rear frame losing mistake to Combined estimator
Method and system are hidden to be described in further detail.According to following explanation and claims, advantages and features of the invention will more
It is clear.It should be noted that attached drawing uses using very simplified form and non-accurate ratio, only to convenient, apparent
Ground aids in illustrating the purpose of the embodiment of the present invention.
Please refer to Fig.2, its be the embodiment of the present invention based on the front and rear frame losing error concealment method to Combined estimator and be
The schematic diagram of system.It is described that four parts are included based on the front and rear frame losing error concealment system to Combined estimator, buffer part, after
To predicted portions, forward prediction part and cross-fading part.
The buffer part is used to cache the 2 frame voice data played and caches the voice number that 6 frames will play
According to.When finding that the next frame data that will play are bad frames, then start error concealment mechanism immediately.Error concealment mechanism is divided into two
Part, Part I are that back forecast part is used to estimate reduction frame losing data backward, which utilizes the voice number before frame losing
Estimate the voice of loss backward according to voice-based short-term stationarity characteristic；Part II is that forward prediction part is used to estimate forward
Meter reduction frame losing data, the voice data after which is terminated using frame losing are estimated to reduce frame losing data forward.Error concealment
After reduction estimation is completed, the voice that cross-fading part carries out two parts of estimation voice data cross-fading generation missing is estimated
Release signal.
It is described to be specifically comprised the following steps based on the front and rear frame losing error concealment method to Combined estimator：
Step 1：Cache the two frame voice data played and the six frame voice data that will be played；The buffer area
Partial cache size is 8 frames (20ms/ frames).Wherein, the front cross frame of buffer area is the caching that has played, rear the six of buffer area
Frame is the voice data that will be played.When the next frame data generation frame losing (being bad frame/null frame) for finding to play, then stand
Start back forecast part and forward prediction part.
Step 2：Estimated to reduce frame losing voice data backward according to the voice data before frame losing；The back forecast part
Including：Linear prediction analysis unit, long-term prediction unit, signal classifier, analysis filter, pitch period repeat and residual error letter
Number modification unit, decay factor computing unit, linear prediction are comprehensive, binary decision unit and attenuation units；
First, linear prediction analysis filter A (z) obtains filter parameter using 8 rank linear prediction analysis filters, fixed
Justice is A (z)=1+a1z-1+a2z-2+...+a8z-8；
Linear prediction analysis is similar with the speech coder of general CELP to include two parts：Adding window and autocorrelation calculation and Du
Guest's algorithm (Levinson-Durbinalgorithm).Autocorrelation calculation includes the bandspreading of 60Hz and the white noise school of 40dB
Just.Wherein, linear prediction analysis window is an asymmetrical Hamming window：
Further, the pitch period of long-term prediction unit estimation current demand signal.
Caching head x (n), the n=-288 of frame losing ..., -1, (288 sampling points are maximum 2 times of pitch periods) passes through low pass filtered
Ripple device simultaneously carries out 4 times down-sampled, obtain bandwidth be 2khz it is down-sampled after signal t (n), n=-72 ..., -1.
Estimation pitch period T for the first timedCalculated by normalized crosscorrelation：
The T calculated according to first timedT=4T is found againdNeighbouring maximal correlation is all to obtain more accurate fundamental tone
Phase.
Further, signal classifier classifies current demand signal, the caching head of frame losing, and the species of current demand signal includes
Vowel state, voiceless sound state and conversion state；
Further, analysis filter obtains currently according to the filter parameter that the linear prediction analysis filter obtains
Analysis filter.Analysis filter is defined as A (z)=1+a1z-1+a2z-2+...+a8z-8, according to the filter of linear analysis acquisition
Ripple device parameter obtains current analysis filter, using to past signal x (n), n=-289 ..., and -1, it is filtered with regard to energy
Residual signals e (n), n=-289 are obtained ..., -1:
Further, pitch period repeats and the residual signals of phonological component are lost in residual signals modification unit generation.If
Current demand signal is vowel state, then repeats to estimate the residual signals of current demand signal, e (n)=e (n-T by the pitch period0)；
If current demand signal is non-vowel state, first to its amplitude limit,
Then estimate residual
Difference signal：E (n)=e (n-T0+(-1)n)
Further, decay factor computing unit, according to frame losing it is different classes of using piecewise linearity decay method into
Row decay.The loss voice estimated in order to prevent excessively causes estimation inaccurate, and the method to be decayed using piecewise linearity is declined
Subtract.Different attenuation curves is used according to the difference of Classification of Speech.Attenuation curve when in violent state as shown in figure 3, turn
It is dotted line to change fading gain curve；When being solid line in other state class attenuation state curves.
Further, linear prediction synthesis is estimated according to the residual signals of the current analysis filter and current demand signal
The voice data part of frame losing：
Further, the voice signal after the frame losing of binary decision unit judges is vowel, if not vowel, then
Decay, i.e.,：gmute(n)=1.If vowel then keeps initial value without decay.
Finally, the estimation signal of attenuation units generation frame losing：y′1(n)=gmute(n)×y1(n)
Step 3：Estimated forward to reduce frame losing voice data according to the voice data after frame losing；The forward prediction part
Including：Signal classifier, long-term prediction unit, fundamental tone repeat estimation unit, vowel decision unit and output control switch.
First, signal classifier classifies current demand signal, the caching tail of frame losing, and the species of current demand signal includes vowel
State, voiceless sound state and conversion state；
Further, the pitch period of long-term prediction unit estimation current demand signal；But with the length of the back forecast part
When predicting unit further it is accurate calculate pitch period when position it is different.
Caching tail z (n), the n=0 of frame losing, -1 ... 287 (288 sampling points are maximum 2 times of pitch periods) pass through low pass filtered
Ripple device simultaneously carries out 4 times down-sampled, obtain bandwidth be 2khz it is down-sampled after signal t (n), n=1 ..., 71.
Estimation pitch period T for the first timedCalculated by normalized crosscorrelation：
The T calculated according to first timedT=4T is found againdNeighbouring maximal correlation is all to obtain more accurate fundamental tone
Phase.
Further, fundamental tone repeats estimation unit for the effect being optimal, and forward direction reduction estimation is only after the loss
The forward estimation of error signal is carried out when voice is vowel.First, a monocyclic estimation signal is obtained：y2(n)=z
(T0- L+n), n=L-1-T0,…,L-1；The frame losing information of estimation is obtained by periodic extension forward again：y2(n)=y2(T0-n),
N=0 ..., L-1-T0。
Further, vowel decision unit judges whether the voice data after frame losing is vowel state；
Finally, output control switch, if vowel state, then open output control switch, if not vowel state, then close defeated
Go out controlling switch.
Step 4：According to the voice data of estimation reduction frame losing backward and the voice data of estimation reduction frame losing carries out forward
The voice estimation release signal that cross-fading generation is lost.Cross-fading is used to merge between frame and frame using quarter window,
N=0 ..., N-1
N is the length to forward lap, and general length is 80.
Base this, the present invention utilize voice short-term stationarity feature, frame losing is combined according to the voice messaging before and after frame losing
Estimation, can more accurately estimate the speech frame of loss, and the compensation to frame losing is more accurate.
As shown in Figure 4 and Figure 5, it is using the present invention and hidden using other methods mistake when not introducing linear attenuation
Hide the contrast of algorithm.When using the present invention, same sound duration is ofer short duration, metallic sound on such human auditory system
Sensation can weaken significantly, error concealment algorithm performance can be lifted.
Foregoing description is only the description to present pre-ferred embodiments, not to any restriction of the scope of the invention, this hair
Any change, the modification that the those of ordinary skill in bright field does according to the disclosure above content, belong to the protection of claims
Scope.
Claims (8)
- It is 1. a kind of based on the front and rear frame losing error concealment method to Combined estimator, it is characterised in that including：Cache the two frame voice data played and the six frame voice data that will be played；When detecting generation frame losing, estimated to reduce frame losing voice data backward according to the voice data before frame losing；According to frame losing Voice data afterwards is estimated forward to reduce frame losing voice data；According to the voice data of estimation reduction frame losing backward and the voice data of estimation reduction frame losing carries out cross-fading life forward Voice into loss estimates release signal；The voice data according to before frame losing is estimated to include the step of reducing frame losing voice data backward：Linear prediction analysis filter obtains filter parameter；Estimate the pitch period of current demand signal；Current demand signal is divided Class, the species of current demand signal include vowel state, voiceless sound state and conversion state；Current analysis filter is obtained according to the filter parameter that the linear prediction analysis filter obtains；If current demand signal It is vowel state, then repeats to estimate the residual signals of current demand signal by the pitch period, if current demand signal is non-vowel state The residual signals of current demand signal are drawn after modifying to the residual signals of current demand signal；Divided according to different classes of use of frame losing The method of section linear attenuation decays；The voice data part of frame losing is estimated according to the residual signals of the current analysis filter and current demand signal；When described Voice after frame losing is not vowel, then decays, and the voice after the frame losing is vowel, then without decay Keep initial value, the final estimation signal for obtaining frame losing.
- 2. as claimed in claim 1 based on the front and rear frame losing error concealment method to Combined estimator, it is characterised in that the line Property forecast analysis wave filter is 8 rank linear prediction analysis filters.
- 3. as claimed in claim 1 based on the front and rear frame losing error concealment method to Combined estimator, it is characterised in that described Estimate forward to include the step of reducing frame losing voice data according to the voice data after frame losing：Classify to frame losing, the species of current demand signal includes vowel state, voiceless sound state and conversion state；Estimate the fundamental tone of current demand signal Cycle；A monocyclic estimation signal is obtained, the frame losing information of estimation is obtained by periodic extension forward；Judge whether the voice data after frame losing is vowel state, if vowel state, then open output control switch, if not Vowel state, then close output control switch.
- 4. as claimed in claim 3 based on the front and rear frame losing error concealment method to Combined estimator, it is characterised in that obtaining In the step of one monocyclic estimation signal, the frame losing information estimated by the acquisition of periodic extension forward, only frame losing is terminated Voice data afterwards is that the voice data of vowel state is estimated.
- It is 5. a kind of based on the front and rear frame losing error concealment system to Combined estimator, it is characterised in that including：Buffer part, for caching the two frame voice data played and the six frame voice data that will be played；Back forecast part, for being estimated to reduce frame losing voice data backward according to the voice data before frame losing；Forward prediction part, for being estimated forward to reduce frame losing voice data according to the voice data after frame losing；Cross-fading part, loses for carrying out cross-fading generation according to the back forecast part and the forward prediction part The voice estimation release signal of mistake；The back forecast part includes：Linear prediction analysis unit, filter parameter is obtained for linear prediction analysis filter；Long-term prediction unit, for estimating the pitch period of current demand signal；Signal classifier, for classifying to current demand signal, the species of current demand signal includes vowel state, voiceless sound state and conversion State；Analysis filter, the filter parameter for being obtained according to the linear prediction analysis filter obtain current analysis filter Ripple device；Pitch period repeats and residual signals modification unit, if being vowel state for current demand signal, passes through the pitch period The residual signals of estimation current demand signal are repeated, are modified if current demand signal is non-vowel state to the residual signals of current demand signal The residual signals of current demand signal are drawn afterwards；Decay factor computing unit, is decayed for the different classes of method to be decayed using piecewise linearity according to frame losing；Linear prediction integrates, for estimating the language of frame losing according to the residual signals of the current analysis filter and current demand signal Sound data portion；Binary decision unit, is not vowel for the voice after the frame losing, then decays, when the frame losing terminates Voice afterwards is vowel, then keeps initial value without decay；Attenuation units, for generating the estimation signal of frame losing.
- 6. as claimed in claim 5 based on the front and rear frame losing error concealment system to Combined estimator, it is characterised in that linear In forecast analysis unit, the linear prediction analysis filter is 8 rank linear prediction analysis filters.
- 7. as claimed in claim 5 based on the front and rear frame losing error concealment system to Combined estimator, it is characterised in that before described Include to predicted portions：Signal classifier, for classifying to frame losing, the species of current demand signal includes vowel state, voiceless sound state and conversion state；Long-term prediction unit, for estimating the pitch period of current demand signal；Fundamental tone repeats estimation unit, and for obtaining a monocyclic estimation signal, estimation is obtained by periodic extension forward Frame losing information；Vowel decision unit, for judging frame losing after voice data whether be vowel state；Output control switch, if vowel state, then open output control switch, if not vowel state, then close output control and open Close.
- 8. as claimed in claim 7 based on the front and rear frame losing error concealment system to Combined estimator, it is characterised in that in fundamental tone Repeat in estimation unit, be only that the voice data of vowel state is estimated to the voice data after frame losing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310747005.6A CN104751851B (en) | 2013-12-30 | 2013-12-30 | It is a kind of based on the front and rear frame losing error concealment method and system to Combined estimator |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310747005.6A CN104751851B (en) | 2013-12-30 | 2013-12-30 | It is a kind of based on the front and rear frame losing error concealment method and system to Combined estimator |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104751851A CN104751851A (en) | 2015-07-01 |
| CN104751851B true CN104751851B (en) | 2018-04-27 |
Family
ID=53591410
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310747005.6A Active CN104751851B (en) | 2013-12-30 | 2013-12-30 | It is a kind of based on the front and rear frame losing error concealment method and system to Combined estimator |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104751851B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110278436A (en) * | 2019-06-28 | 2019-09-24 | 瓴盛科技有限公司 | Picture frame mistake hidden method and device |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105788601B (en) * | 2014-12-25 | 2019-08-30 | 联芯科技有限公司 | The shake hidden method and device of VoLTE |
| CN106788876B (en) * | 2015-11-19 | 2020-01-21 | 电信科学技术研究院 | Method and system for compensating voice packet loss |
| CN110366029B (en) * | 2019-07-04 | 2021-08-24 | 中国科学院深圳先进技术研究院 | Method and system for inserting image frame between videos and electronic equipment |
| CN111371534B (en) * | 2020-06-01 | 2020-09-18 | 腾讯科技（深圳）有限公司 | Data retransmission method and device, electronic equipment and storage medium |
| CN113035205B (en) * | 2020-12-28 | 2022-06-07 | 阿里巴巴（中国）有限公司 | Audio packet loss compensation processing method and device and electronic equipment |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101120400A (en) * | 2005-01-31 | 2008-02-06 | 索诺瑞特公司 | Method for generating concealment frames in communication system |
| CN101207468A (en) * | 2006-12-19 | 2008-06-25 | 华为技术有限公司 | Method, system and apparatus for missing frame hide |
| CN101325631A (en) * | 2007-06-14 | 2008-12-17 | 华为技术有限公司 | Method and apparatus for implementing bag-losing hide |
| CN101471073A (en) * | 2007-12-27 | 2009-07-01 | 华为技术有限公司 | Package loss compensation method, apparatus and system based on frequency domain |
| CN101833954A (en) * | 2007-06-14 | 2010-09-15 | 华为终端有限公司 | Method and device for realizing packet loss concealment |
| CN102833037A (en) * | 2012-07-18 | 2012-12-19 | 华为技术有限公司 | Speech data packet loss compensation method and device |
| CN103065636A (en) * | 2011-10-24 | 2013-04-24 | 中兴通讯股份有限公司 | Voice frequency signal frame loss compensation method and device |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101008508B1 (en) * | 2006-08-15 | 2011-01-17 | 브로드콤 코포레이션 | Re-phasing of decoder states after packet loss |
-
2013
- 2013-12-30 CN CN201310747005.6A patent/CN104751851B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101120400A (en) * | 2005-01-31 | 2008-02-06 | 索诺瑞特公司 | Method for generating concealment frames in communication system |
| CN101207468A (en) * | 2006-12-19 | 2008-06-25 | 华为技术有限公司 | Method, system and apparatus for missing frame hide |
| CN101325631A (en) * | 2007-06-14 | 2008-12-17 | 华为技术有限公司 | Method and apparatus for implementing bag-losing hide |
| CN101833954A (en) * | 2007-06-14 | 2010-09-15 | 华为终端有限公司 | Method and device for realizing packet loss concealment |
| CN101471073A (en) * | 2007-12-27 | 2009-07-01 | 华为技术有限公司 | Package loss compensation method, apparatus and system based on frequency domain |
| CN103065636A (en) * | 2011-10-24 | 2013-04-24 | 中兴通讯股份有限公司 | Voice frequency signal frame loss compensation method and device |
| CN102833037A (en) * | 2012-07-18 | 2012-12-19 | 华为技术有限公司 | Speech data packet loss compensation method and device |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110278436A (en) * | 2019-06-28 | 2019-09-24 | 瓴盛科技有限公司 | Picture frame mistake hidden method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104751851A (en) | 2015-07-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104751851B (en) | It is a kind of based on the front and rear frame losing error concealment method and system to Combined estimator | |
| KR100964402B1 (en) | Method and Apparatus for determining encoding mode of audio signal, and method and appartus for encoding/decoding audio signal using it | |
| KR102237718B1 (en) | Device and method for reducing quantization noise in a time-domain decoder | |
| JP4658596B2 (en) | Method and apparatus for efficient frame loss concealment in speech codec based on linear prediction | |
| US7472059B2 (en) | Method and apparatus for robust speech classification | |
| KR100883656B1 (en) | Method and apparatus for discriminating audio signal, and method and apparatus for encoding/decoding audio signal using it | |
| KR100367267B1 (en) | Multimode speech encoder and decoder | |
| JP6671439B2 (en) | Method and apparatus for voice activity detection | |
| EP1747554B1 (en) | Audio encoding with different coding frame lengths | |
| EP2535893B1 (en) | Device and method for lost frame concealment | |
| ES2380962T3 (en) | Procedure and apparatus for coding low transmission rate of high performance deaf speech bits | |
| US10706858B2 (en) | Error concealment unit, audio decoder, and related method and computer program fading out a concealed audio frame out according to different damping factors for different frequency bands | |
| KR20080103113A (en) | Signal encoding | |
| WO2008067719A1 (en) | Sound activity detecting method and sound activity detecting device | |
| KR20020052191A (en) | Variable bit-rate celp coding of speech with phonetic classification | |
| US10937432B2 (en) | Error concealment unit, audio decoder, and related method and computer program using characteristics of a decoded representation of a properly decoded audio frame | |
| US9293143B2 (en) | Bandwidth extension mode selection | |
| KR101794149B1 (en) | Noise filling without side information for celp-like coders | |
| CN107818789B (en) | Decoding method and decoding device | |
| US6564182B1 (en) | Look-ahead pitch determination | |
| EP2608200B1 (en) | Estimation of speech energy based on code excited linear prediction (CELP) parameters extracted from a partially-decoded CELP-encoded bit stream | |
| RU2707144C2 (en) | Audio encoder and audio signal encoding method | |
| Faúndez-Zanuy | Adaptive hybrid speech coding with a MLP/LPC structure | |
| US20220180884A1 (en) | Methods and devices for detecting an attack in a sound signal to be coded and for coding the detected attack | |
| Thimmaraja et al. | Enhancements in encoded noisy speech data by background noise reduction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20150701 Assignee: Shanghai Li Ke Semiconductor Technology Co.,Ltd. Assignor: LEADCORE TECHNOLOGY Co.,Ltd. Contract record no.: 2018990000159 Denomination of invention: Before and after combined estimation based frame loss error hiding method and system Granted publication date: 20180427 License type: Common License Record date: 20180615 |
|
| EE01 | Entry into force of recordation of patent licensing contract | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20180824 Address after: 201206 Pudong New Area, Shanghai, China (Shanghai) free trade pilot area, 1258 A406 3 fourth story room. Patentee after: Chen core technology Co.,Ltd. Address before: 201206 4 building, No. 333, No. 41, Qinjiang Road, Shanghai, Xuhui District Patentee before: LEADCORE TECHNOLOGY Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20221121 Address after: 266500 No. 687, Binhai street, Huangdao District, Qingdao, Shandong Patentee after: Chenxin Technology Co.,Ltd. Patentee after: Qingdao Weixuan Technology Co.,Ltd. Address before: 201206 Pudong New Area, Shanghai, China (Shanghai) free trade pilot area, 1258 A406 3 fourth story room. Patentee before: Chen core technology Co.,Ltd. |
|
| CP03 | Change of name, title or address | ||
| CP03 | Change of name, title or address |
Address after: Room 102, Building 16, No. 1699, the Pearl River Road, Huangdao District, Qingdao, Shandong 266499 Patentee after: Chenxin Technology Co.,Ltd. Patentee after: Qingdao Weixuan Technology Co.,Ltd. Address before: 266500 No. 687, Binhai street, Huangdao District, Qingdao, Shandong Patentee before: Chenxin Technology Co.,Ltd. Patentee before: Qingdao Weixuan Technology Co.,Ltd. |