CN104270539B - Based on the proportional affine projection echo cancel method of coefficient difference - Google Patents

Based on the proportional affine projection echo cancel method of coefficient difference Download PDF

Info

Publication number
CN104270539B
CN104270539B CN201410465261.0A CN201410465261A CN104270539B CN 104270539 B CN104270539 B CN 104270539B CN 201410465261 A CN201410465261 A CN 201410465261A CN 104270539 B CN104270539 B CN 104270539B
Authority
CN
China
Prior art keywords
current time
echo
sef
coefficient
adapting filter
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.)
Expired - Fee Related
Application number
CN201410465261.0A
Other languages
Chinese (zh)
Other versions
CN104270539A (en
Inventor
赵海全
郑宗生
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Southwest Jiaotong University
Original Assignee
Southwest Jiaotong University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Southwest Jiaotong University filed Critical Southwest Jiaotong University
Priority to CN201410465261.0A priority Critical patent/CN104270539B/en
Publication of CN104270539A publication Critical patent/CN104270539A/en
Application granted granted Critical
Publication of CN104270539B publication Critical patent/CN104270539B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)
  • Filters That Use Time-Delay Elements (AREA)

Abstract

Based on a proportional affine projection echo cancel method for coefficient difference, its step comprises: A, remote signaling are sampled; B, echo signal are estimated; C, echo signal are eliminated; D, filter tap weight coefficient upgrade; E, make n=n+1, repeat step B, C, D, real-time echo cancellor can be realized; Wherein D step tap weights coefficient update time, i-th tap weights coefficient w of the sef-adapting filter of current time ithe step-length g of (n) i(n), g i(n)=max{ ρ, | w i(n)-w i(kM) | }, namely within the time period that the time is M, the step-length of the sef-adapting filter tap weights coefficient of current time equals the difference of arbitrary tap weights coefficient of current time n and this tap weights coefficient of this time period initial time kM; The affine projection method of the input signal matrix that weight coefficient adopts the input signal vector in multiple moment to form when upgrading.The method is effective to echo cancellor, and environment self-adaption ability is strong, fast convergence rate, and steady-state error is little, and computation complexity is low simultaneously, and required hardware cost is low, and structure is simple, easily implements.

Description

Based on the proportional affine projection echo cancel method of coefficient difference
Technical field
The invention belongs to the echo cancellation technology field in communication.
Background technology
The progressive active influence of information technology the every aspect of people's life, brings deep change to human society life.As the importance of information technology, the communication technology obtains significant progress, and the echo cancellor in communication process is one, this field has much attention rate and challenging focus.Sound can form echo through multiple reflections within the enclosed space, also can form echo in Signal transmissions because transmission medium middle impedance does not mate.Communication echo can carry out self adaptation elimination by System identification model: institute's identification system is echo channel, the output of System Discrimination is the estimation of echo signal, by subtracting each other the elimination that just can realize echo containing the voice signal of echo signal and the estimation of echo signal, the principle of Here it is self-adaptive echo eliminator.
Echo channel all has sparse characteristic, and namely most of coefficient of channel (system) is close to zero or equal zero, and only has a few coefficients to have larger amplitude.The adaptive filter method of existing echo cancellor, mainly contain: NLMS (normalization minimum mean-square) algorithm, it does not consider the structure of target impulse response, for all filter tap weight coefficients distribute identical step parameter, may be bigger than normal for its step-length used of little coefficient, can restrain after less iteration, but its little coefficient precision picked out is low; May be less than normal for its step-length used of large coefficient, need more iterations to restrain; This convergence of algorithm time and identification precision are all had much room for improvement.Proportional algorithm is that large coefficient distributes larger step parameter to accelerate the convergence rate of large coefficient, thus accelerates the global convergence speed of algorithm.Duttweiler first proposed proportional least mean square algorithm (PNLMS), the step parameter of current time is made to be proportional to the absolute value of the filter tap weight coefficient of current time by introducing step size controlling matrix, so, larger coefficient obtains larger step parameter, thus accelerates the global convergence speed of algorithm.But the openness requirement of PNLMS to algorithm is higher, when target impulse response is sparse not, convergence of algorithm speed ratio normalization minimum mean-square calculation (NLMS) algorithm also wants slow.Benesty, by the average of current for filter tap weight coefficient vector estimated value being added to the step parameter of each coefficient, proposes the proportional least mean square algorithm (IPNLMS) of improvement.This algorithm has the Fast Convergent performance same with PNLMS algorithm when processing sparse impulse response, and when processing non-sparse impulse response, its convergence rate is unlikely to deteriorate into the degree slower than NLMS algorithm simultaneously.
When processing correlated inputs signal (such as voice signal), affine projection algorithm (APA) has than NLMS algorithm convergence rate faster.APA algorithm can see the expansion of NLMS algorithm in time domain as.Each time in iteration, NLMS algorithm only uses current input signal vector, and APA algorithm uses P nearest input signal vector, decorrelation (or being called " albefaction ") has been carried out to input signal vector, partly eliminate the impact of input signal autocorrelation matrix characteristic value diffusion couple algorithm the convergence speed, so APA convergence of algorithm speed is faster than NLMS algorithm.Its computation complexity is relevant to projection order P, and uses less projection order namely can obtain satisfied result in generally applying.Just because of this, APA algorithm and multiple innovatory algorithm thereof are widely used in echo cancellor.
The same with traditional NLMS algorithm, APA algorithm its convergence rate when processing sparse impulse response is also corresponding slack-off.In order to accelerate algorithm the convergence speed, scholar is had to propose proportional affine projection algorithm by proportional thought is introduced APA algorithm.In the application of current echo cancellor, such algorithm of better performances has following two kinds:
(1) proportional affine projection echo cancel method
List of references " Double-talkrobustfastconvergingalgorithmsfornetworkechoc ancellation " (T. s.L.Gay, M.M.Sondhi, andJ.Benesty, IEEETrans.Speech, AudioProcess, vol.8, no.6, pp.656 – 663, Nov.2000.) the method the proportional thought in PNLMS algorithm directly introduced APA algorithm obtain, this algorithm is called proportional affine projection algorithm (PAPA).The same with PNLMS algorithm, PAPA algorithm can be that large coefficient distributes larger step-length, thus accelerate algorithm global convergence speed, but, when target impulse response is sparse not, and step size controlling matrix is appointed and so distributed step parameter by the absolute value of filter tap weight coefficient, the step-length that large coefficient obtains is excessive makes that convergence rate is slack-off, steady-state error increases, and algorithm performance worsens.Convergence of algorithm speed is also slower than APA algorithm.
(2) the proportional affine projection echo cancel method improved
List of references " Ageneralizedproportionatevariablestep-sizealgorithmforfa stchangingacousticenvironments " (O.Hoshuyama, R.A.Goubran, andA.Sugiyama, IEEEICASSP, 2004, vol.4, pp.161-164, May2004.) the method the proportional thought improved in IPNLMS algorithm combined with APA algorithm by the inspiration of IPNLMS algorithm to obtain, and claims this algorithm to be the proportional affine projection algorithm (IPAPA) improved.The same with IPNLS algorithm, the average of current for filter tap weight coefficient vector estimated value is added to the step parameter of each coefficient by PAPA algorithm, have the Fast Convergent performance same with PNLMS algorithm when processing sparse impulse response, when processing non-sparse impulse response, its convergence rate is unlikely to deteriorate into the degree slower than NLMS algorithm simultaneously.But the essential one compromise form of taking up an official post so for PAPA algorithm and APA algorithm of IPAPA algorithm, its convergence rate can not promote further.Moreover the performance height of IPAPA algorithm depends on adjustment parameter, and adjusts the structure that parameter depends on target impulse response, can not self-adaptative adjustment.
Summary of the invention
Goal of the invention of the present invention is just to provide a kind of proportional affine projection echo cancel method based on coefficient difference, the method is effective to echo cancellor, environment self-adaption ability is strong, fast convergence rate, steady-state error is little, and computation complexity is low simultaneously, and required hardware cost is low, structure is simple, easily implements.
The present invention realizes the technical scheme that its goal of the invention adopts, a kind of proportional affine projection echo cancel method based on coefficient difference, and its step is as follows:
Based on a proportional affine projection echo cancel method for coefficient difference, its step is as follows:
A, remote signaling are sampled
Input signal x (n) of current time n is obtained by carrying out remote signaling sampling, by nearest L input signal x (n), x (n-1), ..., x (n-L+1) forms input signal vector X (n) of current time n, X (n)=[x (n), x (n-1), ..., x (n-L+1)] t; Subscript T represents vector transpose, and L is sef-adapting filter tap length, L=64,128,256,512;
B, echo signal are estimated
Input signal vector X (n) of current time n is obtained the estimated value of echo signal by sef-adapting filter namely
y ^ ( n ) = W T ( n ) X ( n )
Wherein, W (n)=[w 1(n), w 2(n), K, w i(n), Kw l(n)] tfor the sef-adapting filter tap weights coefficient vector of current time, its initial value is zero, w ii-th tap weights coefficient of n sef-adapting filter that () is current time;
C, echo signal are eliminated
Near end signal sampling is obtained near end signal d (n) of the current time being with echo, deducted the estimated value of the echo signal that step B obtains be eliminated useful signal e (n) of current time of echo, e ( n ) = d ( n ) - y ^ ( n ) ;
D, filter tap weight coefficient upgrade
By input signal vector X (n) in a nearest P moment, X (n-1), ..., X (n-P+1) forms input signal matrix U (n), U (n)=[X (n), X (n-1) ..., X (n-P+1)]; Wherein, P is affine projection exponent number, P=2 ~ 8;
By useful signal e (n) in a nearest P moment, e (n-1), K, e (n-P+1) forms useful signal vector E (n) of current time, E (n)=[e (n), e (n-1), K, e (n-P+1)] t;
Calculate i-th tap weights coefficient w of the sef-adapting filter of current time ithe step-length g of (n) i(n), g i(n)=max{ ρ, | w i(n)-w i(kM) | }; Wherein, max{} represents and gets maximum, || represent and take absolute value, ρ is a constant being greater than zero, and its span is 0.001 ~ 0.1; w i(kM) be i-th tap weights coefficient of the sef-adapting filter in kM moment, M is a constant, and its span is 100 ~ 300, represent downward round numbers;
By all tap weights coefficient w of the sef-adapting filter of current time ithe step-length g of (n) in () forms diagonalizable Step matrix G (n), G (n)=diag{g 1(n), g 2(n) ..., g i(n) ..., g l(n) }; Wherein, diag{} represents formation diagonalizable matrix;
Finally, draw sef-adapting filter tap weights coefficient vector W (n+1) of subsequent time by following formula,
W(n+1)＝W(n)+μG(n)U(n)[U T(n)G(n)U(n)+δI P] -1E(n)
Wherein, μ is the overall step parameter of sef-adapting filter, and its span is 0< μ <2, δ is the normal number preventing matrix inversion dyscalculia, and its value is generally 0.01, I pfor the unit matrix of P × P;
E, make n=n+1, repeat step B, C, D, real-time echo cancellor can be realized.
Compared with prior art, the invention has the beneficial effects as follows:
One, fast convergence rate, steady-state error is little
I-th tap weights coefficient w of the sef-adapting filter of current time of the present invention ithe step-length g of (n) i(n), g i(n)=max{ ρ, | w i(n)-w i(kM) | }, namely within the time period that the time is M, the step-length of the sef-adapting filter tap weights coefficient of current time equals the difference of arbitrary tap weights coefficient of current time n and this tap weights coefficient of this time period initial time kM, the difference calculated large coefficient time initial is also comparatively large, thus can obtain initial convergence speed faster; And diminish close to this difference during stable state, make the step-length of tap weights coefficient diminishing close to also corresponding during stable state, also namely target impulse response non-sparse time still can obtain convergence rate faster.In addition affine projection algorithm adopts the input signal matrix of the input signal vector in the multiple moment formation voice signal strong to relativity of time domain to carry out filter identification, also makes the present invention have convergence rate and lower steady-state error faster.
Two, computation complexity is low, and adaptive ability is strong
Compare with IPAPA algorithm with the PAPA algorithm of routine, the present invention does not have the normalization step of redundancy, significantly can reduce computation complexity, makes its hardware implementation cost low yet, and structure is simple, easily implements.And different from IPAPA, the present invention does not adjust parameter, strong to the adaptive ability of environment.
Accompanying drawing explanation
Fig. 1 is PAPA algorithm, IPAPA algorithm and normalization steady output rate curve of the present invention.
Embodiment
The following detailed description of the present invention's performing step in actual applications.
Embodiment
A kind of embodiment of the present invention is, a kind of proportional affine projection echo cancel method based on coefficient difference, and its step is as follows:
A, remote signaling are sampled
Input signal x (n) of current time n is obtained by carrying out remote signaling sampling, by nearest L input signal x (n), x (n-1), ..., x (n-L+1) forms input signal vector X (n) of current time n, X (n)=[x (n), x (n-1), ..., x (n-L+1)] t; Subscript T represents vector transpose, and L is sef-adapting filter tap length, L=64,128,256,512;
B, echo signal are estimated
Input signal vector X (n) of current time n is obtained the estimated value of echo signal by sef-adapting filter namely
y ^ ( n ) = W T ( n ) X ( n )
Wherein, W (n)=[w 1(n), w 2(n), K, w i(n), Kw l(n)] tfor the sef-adapting filter tap weights coefficient vector of current time, its initial value is zero, w ii-th tap weights coefficient of n sef-adapting filter that () is current time;
C, echo signal are eliminated
Near end signal sampling is obtained near end signal d (n) of the current time being with echo, deducted the estimated value of the echo signal that step B obtains be eliminated useful signal e (n) of current time of echo, e ( n ) = d ( n ) - y ^ ( n ) ;
D, filter tap weight coefficient upgrade
By input signal vector X (n) in a nearest P moment, X (n-1), ..., X (n-P+1) forms input signal matrix U (n), U (n)=[X (n), X (n-1) ..., X (n-P+1)]; Wherein, P is affine projection exponent number, P=2 ~ 8;
By useful signal e (n) in a nearest P moment, e (n-1), K, e (n-P+1) forms useful signal vector E (n) of current time, E (n)=[e (n), e (n-1), K, e (n-P+1)] t;
Calculate i-th tap weights coefficient w of the sef-adapting filter of current time ithe step-length g of (n) i(n), g i(n)=max{ ρ, | w i(n)-w i(kM) | }; Wherein, max{} represents and gets maximum, || represent and take absolute value, ρ is a constant being greater than zero, and its span is 0.001 ~ 0.1; w i(kM) be i-th tap weights coefficient of the sef-adapting filter in kM moment, M is a constant, and its span is 100 ~ 300, represent downward round numbers;
By all tap weights coefficient w of the sef-adapting filter of current time ithe step-length g of (n) in () forms diagonalizable Step matrix G (n), G (n)=diag{g 1(n), g 2(n) ..., g i(n) ..., g l(n) }; Wherein, diag{} represents formation diagonalizable matrix;
Finally, draw sef-adapting filter tap weights coefficient vector W (n+1) of subsequent time by following formula,
W(n+1)＝W(n)+μG(n)U(n)[U T(n)G(n)U(n)+δI P] -1E(n)
Wherein, μ is the overall step parameter of sef-adapting filter, and its span is 0< μ <2, δ is the normal number preventing matrix inversion dyscalculia, and its value is generally 0.01, I pfor the unit matrix of P × P;
E, make n=n+1, repeat step B, C, D, real-time echo cancellor can be realized.
Emulation experiment
In order to verify the validity of the proportional affine projection echo cancel method based on coefficient difference, carry out emulation experiment, and do performance comparison with the PAPA algorithm of current better performances and IPAPA algorithm.
In emulation experiment, sef-adapting filter tap length L is 64, the input signal employing limit of far-end is first-order autoregression (AR (the 1)) signal of 0.95, be long 6.25m in room, wide 3.75m, high 2.5m, temperature 20 DEG C, in the quiet closed room of humidity 50%, near-end is by the remote signaling that receives after loud speaker is play, and be 8000Hz with microphone by sample frequency in a room, sampling order L is 64 near end signals d (n) picking up out 3000 moment point altogether.
By using the present invention and PAPA algorithm, IPAPA algorithm to carry out echo cancellor to above-mentioned signal respectively, in experiment, the concrete value of the parameter of each algorithm is as following table.
The optimized parameter of each algorithm simulating experiment is similar to value
PAPA μ＝0.2，δ＝0.01，ρ＝0.01，δ P＝0.01
IPAPA μ＝0.2，δ＝0.01，α＝0
The present invention μ＝0.2，δ＝0.01，ρ＝0.005，M＝2L
Simulation result is on average obtained for 20 times by independent operating.Fig. 1 is the normalization steady output rate curve of PAPA algorithm, IPAPA algorithm and simulation result of the present invention.As can be seen from Figure 1 when steady-state error is identical, IPAPA convergence of algorithm speed is faster than PAPA algorithm, and convergence rate of the present invention is obviously faster than PAPA algorithm and IPAPA algorithm.

Claims (1)

1., based on a proportional affine projection echo cancel method for coefficient difference, its step is as follows:
A, remote signaling are sampled
Input signal x (n) of current time n is obtained by carrying out remote signaling sampling, by nearest L input signal x (n), x (n-1), ..., x (n-L+1) forms input signal vector X (n) of current time n, X (n)=[x (n), x (n-1), ..., x (n-L+1)] t; Subscript T represents vector transpose, and L is sef-adapting filter tap length, L=64,128,256,512;
B, echo signal are estimated
Input signal vector X (n) of current time n is obtained the estimated value of echo signal by sef-adapting filter namely
y ^ ( n ) = W T ( n ) X ( n )
Wherein, W (n)=[w 1(n), w 2(n) ..., w i(n) ... w l(n)] tfor the sef-adapting filter tap weights coefficient vector of current time, its initial value is zero, w ii-th tap weights coefficient of n sef-adapting filter that () is current time;
C, echo signal are eliminated
Near end signal sampling is obtained near end signal d (n) of the current time being with echo, deducted the estimated value of the echo signal that step B obtains be eliminated useful signal e (n) of current time of echo, e ( n ) = d ( n ) - y ^ ( n ) ;
D, filter tap weight coefficient upgrade
By input signal vector X (n) in a nearest P moment, X (n-1), ..., X (n-P+1) forms input signal matrix U (n), U (n)=[X (n), X (n-1) ..., X (n-P+1)]; Wherein, P is affine projection exponent number, P=2 ~ 8;
By useful signal e (n) in a nearest P moment, e (n-1), ..., e (n-P+1) forms useful signal vector E (n) of current time, E (n)=[e (n), e (n-1) ..., e (n-P+1)] t;
Calculate i-th tap weights coefficient w of the sef-adapting filter of current time ithe step-length g of (n) i(n), g i(n)=max{ ρ, | w i(n)-w i(kM) | }; Wherein, max{} represents and gets maximum, || represent and take absolute value, ρ is a constant being greater than zero, and its span is 0.001 ~ 0.1; w i(kM) be i-th tap weights coefficient of the sef-adapting filter in kM moment, M is a constant, and its span is 100 ~ 300, k is round parameter downwards, represent downward round numbers;
By all tap weights coefficient w of the sef-adapting filter of current time ithe step-length g of (n) in () forms diagonalizable Step matrix G (n), G (n)=diag{g 1(n), g 2(n) ..., g i(n) ..., g l(n) }; Wherein, diag{} represents formation diagonalizable matrix;
Finally, draw sef-adapting filter tap weights coefficient vector W (n+1) of subsequent time by following formula,
W(n+1)＝W(n)+μG(n)U(n)[U T(n)G(n)U(n)+δI P] -1E(n)
Wherein, μ is the overall step parameter of sef-adapting filter, and its span is 0 < μ < 2, δ is the normal number preventing matrix inversion dyscalculia, and its value is generally 0.01, I pfor the unit matrix of P × P;
E, make n=n+1, repeat step B, C, D, real-time echo cancellor can be realized.
CN201410465261.0A 2014-09-12 2014-09-12 Based on the proportional affine projection echo cancel method of coefficient difference Expired - Fee Related CN104270539B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410465261.0A CN104270539B (en) 2014-09-12 2014-09-12 Based on the proportional affine projection echo cancel method of coefficient difference

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410465261.0A CN104270539B (en) 2014-09-12 2014-09-12 Based on the proportional affine projection echo cancel method of coefficient difference

Publications (2)

Publication Number Publication Date
CN104270539A CN104270539A (en) 2015-01-07
CN104270539B true CN104270539B (en) 2016-03-02

Family

ID=52162027

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410465261.0A Expired - Fee Related CN104270539B (en) 2014-09-12 2014-09-12 Based on the proportional affine projection echo cancel method of coefficient difference

Country Status (1)

Country Link
CN (1) CN104270539B (en)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104683614B (en) * 2015-03-24 2016-03-02 西南交通大学 Based on the proportional illumination-imitation projection self-adoptive echo cancel method of memory that M estimates
CN105070296A (en) * 2015-07-10 2015-11-18 西南交通大学 Active factor set membership proportional sub band self-adaption echo cancellation method
CN106254698B (en) * 2016-08-19 2019-03-22 西南交通大学 A kind of zero norm subband acoustic echo removing method
CN106331402B (en) * 2016-08-25 2019-03-22 西南交通大学 One kind being based on the proportional subband convex combination adaptive echo null method of coefficient difference
CN107105111B (en) * 2017-03-15 2019-08-02 西南交通大学 A kind of proportional affine projection echo cancel method of combination step-length
CN109040497B (en) * 2018-07-24 2020-12-25 西南交通大学 Proportional affine projection self-adaptive echo cancellation method based on M estimation
CN113225045B (en) * 2021-03-25 2023-06-23 苏州大学 Sparse-facilitated affine projection adaptive filter with low computational complexity

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6744886B1 (en) * 1999-01-06 2004-06-01 Lucent Technologies Inc. Adaptive filter utilizing proportional affine projection algorithm
KR100547765B1 (en) * 2004-05-03 2006-01-31 삼성전자주식회사 Method and apparatus for removing acoustic echo according to simultaneous call section detection
DE102008035074A1 (en) * 2008-07-28 2010-02-18 Siemens Aktiengesellschaft Method for measuring the current strength of an alternating current
CN103457887A (en) * 2012-06-04 2013-12-18 西安秦码软件科技有限公司 Ultra wide band channel equalizer model established based on affine projection algorithm
CN103323651B (en) * 2013-07-09 2016-03-02 西南交通大学 Based on the variable step affine projection harmonic current detecting method that time coherence is average
CN103561185B (en) * 2013-11-12 2015-08-12 沈阳工业大学 A kind of echo cancel method of sparse path

Also Published As

Publication number Publication date
CN104270539A (en) 2015-01-07

Similar Documents

Publication Publication Date Title
CN104270539B (en) Based on the proportional affine projection echo cancel method of coefficient difference
CN104683614B (en) Based on the proportional illumination-imitation projection self-adoptive echo cancel method of memory that M estimates
CN101917527B (en) Method and device of echo elimination
CN107105111B (en) A kind of proportional affine projection echo cancel method of combination step-length
CN109817209A (en) A kind of intelligent speech interactive system based on two-microphone array
CN103369162B (en) A kind of listener&#39;s echo self adaptive elimination method of low complex degree
CN105721729B (en) Based on the sparse proportional echo cancel method for reusing weight coefficient affine projection of block
CN105070296A (en) Active factor set membership proportional sub band self-adaption echo cancellation method
CN105810202B (en) It is a kind of to drop hypoechoic method, apparatus and communication apparatus
CN107333017A (en) One kind reuses weight coefficient exponential type error adaptive echo cancellation method
CN110191245A (en) A kind of adaptive echo cancellation method based on time-varying parameter
Niwa et al. Post-filter design for speech enhancement in various noisy environments
CN108877824B (en) Combined step echo cancellation method with high tracking performance
CN110572525B (en) Self-adaptive communication echo cancellation method for voice communication
CN104144269B (en) A kind of proportional self adaptation listener&#39;s echo removing method based on decorrelation
CN106533500A (en) Method for optimizing convergence characteristic of acoustic echo canceller
CN109040497A (en) A kind of proportional class illumination-imitation projection self-adoptive echo cancel method based on M estimation
CN107610714A (en) The echo cancel method of the minimum cube absolute value attracted based on a norm zero
CN102760461A (en) Volume adaptive control audio playing equipment and method
CN104158512A (en) self-adaptive sparse system identification method based on impact-interference-resistance of independent activity factor
CN106161820B (en) A kind of interchannel decorrelation method for stereo acoustic echo canceler
CN109151237B (en) Affine projection self-adaptive echo cancellation method based on zero attraction
CN101568058B (en) Digital hearing aid echo path estimation method based on weighted subgradient projection
CN101789771B (en) Pulse noise active control method based on logarithm conversion
CN106157965A (en) A kind of zero norm collection person&#39;s illumination-imitation projection self-adoptive echo cancel method reused based on weight vector

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20160302

Termination date: 20180912

CF01 Termination of patent right due to non-payment of annual fee