CN109754813B - Variable step size echo cancellation method based on rapid convergence characteristic - Google Patents
Variable step size echo cancellation method based on rapid convergence characteristic Download PDFInfo
- Publication number
- CN109754813B CN109754813B CN201910231065.XA CN201910231065A CN109754813B CN 109754813 B CN109754813 B CN 109754813B CN 201910231065 A CN201910231065 A CN 201910231065A CN 109754813 B CN109754813 B CN 109754813B
- Authority
- CN
- China
- Prior art keywords
- signal
- filter
- calculating
- frequency domain
- step size
- 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
Images
Abstract
The invention discloses a variable step size echo cancellation method based on rapid convergence characteristics, which comprises the steps of calculating the distance between a far-end reference signal x of a current frame and a microphone receiving signal d,respectively carrying out Fourier transform to obtain corresponding far-end reference frequency domain signals X (k) and microphone receiving frequency domain signals D (k); calculating an a priori error E (k) according to the far-end reference frequency domain signal X (k) and the microphone receiving frequency domain signal D (k); calculating a transfer parameter T 'of a filter'i(ii) a Calculating the updating step size mu (k) of the filter, and updating the filter according to the updating step size mu (k); recalculating the current frame according to the updated filter coefficient to obtain the posterior error Ep(k) (ii) a A posterior error Ep(k) And (D) performing inverse Fourier transform, taking the second half data to obtain a time domain signal of the current frame, and returning to the step (A) to process the next frame. The invention has good application prospect in a real-time voice conversation system and a scene with higher real-time requirement.
Description
Technical Field
The invention relates to the technical field of acoustic echo cancellation, in particular to a variable step size echo cancellation method based on a rapid convergence characteristic.
Background
With the rapid development of intelligent voice technology, more and more intelligent hardware provides voice interaction function, and acoustic echo cancellation performance is a main index for detecting a voice interaction system. At present, the design of echo cancellation mainly includes that an echo cancellation chip is added on a hardware system, and on one hand, the echo cancellation chip acquires the voice input of the system from a microphone; and on the other hand, acquiring a signal output to the loudspeaker by the hardware control unit as a reference signal. Usually, the microphone signal is a signal formed by mixing a voice command given by a user and a sound played by the intelligent hardware through a loudspeaker, the reference signal is a signal played by the loudspeaker, and the echo canceller is used for eliminating the same part of the microphone signal as the reference signal.
A conventional approach to acoustic echo cancellation is to employ an adaptive filter to estimate the acoustic path between the speaker and the microphone. The time domain algorithm of the self-adaptive filter mainly comprises an NLMS algorithm, an AP algorithm, an RLS algorithm and various improved algorithms thereof. However, for a voice communication system with extremely high real-time requirements and voice terminal equipment with limited resources, the application of the NLMS in the frequency domain is wider. However, the fixed-step FDAF algorithm has a contradiction between the convergence rate and the steady-state detuning amount, and thus, the variable-step scheme is widely applied to a real-time voice conference system. The main idea of these schemes is to let the filter adaptively update the step size according to the convergence status: when the filter is in a transient state, the filter is updated by adopting a large step size to obtain a rapid convergence characteristic, and when the filter is in a steady state, the filter is updated by adopting a small step size to obtain a lower steady-state detuning amount.
Enzner proposes a scheme based on State-Space (SS) modeling of acoustic paths. The main idea of this kind of scheme is to approximate a time-varying acoustic path (LEM) using a first-order markov model, and at the same time, apply the classical filtering theory associated with it to the coefficient update of the adaptive filter. And (3) disassembling the frequency domain implementation of the Kalman filter into a Frequency Domain Adaptive Filter (FDAF) model, wherein the updating step length in the FDAF model is skillfully replaced by Kalman gain. The method has good balance among convergence rate, steady state detuning amount and robustness during double talk. Moreover, the method has no excessive parameter configuration, but the tracking capability of the path change and the steady state maladjustment are particularly outstanding contradictions. In the FDAF model, the larger the value of a system transmission parameter is, the smaller the steady state imbalance of the system is, but the tracking capability of a filter is obviously deteriorated when the path is suddenly changed, even the freezing phenomenon of the filter occurs, and only if the channel transmission parameter is set to be smaller, the sudden change path can be better tracked by the system, but the steady state imbalance of the sudden change path is increased.
Therefore, how to overcome the above problems and increase the effect of variable step size echo cancellation is currently needed to be solved.
Disclosure of Invention
The object of the present invention is to overcome the drawbacks of the prior art. The variable-step-size echo cancellation method based on the rapid convergence characteristic effectively improves the tracking performance of the system when the echo path is suddenly changed under the condition of ensuring small detuning amount, has small calculation amount, and has good application prospect in a real-time voice conversation system and a scene with higher real-time requirement.
In order to achieve the purpose, the invention adopts the technical scheme that:
a variable step size echo cancellation method based on fast convergence characteristics comprises the following steps,
respectively carrying out Fourier transform on a far-end reference signal x and a microphone receiving signal d of a current frame to obtain a corresponding far-end reference frequency domain signal X (k) and a corresponding microphone receiving frequency domain signal D (k), wherein k is a frequency point label and does not lose generality, and if N-point Fourier transform is adopted, the value of the frequency point label k is 0-N-1;
step (B), calculating a priori error E (k) according to the far-end reference frequency domain signal X (k) and the microphone receiving frequency domain signal D (k);
step (C) of calculating a transmission parameter T of the filteri′；
Step (D), calculating the updating step size mu (k) of the filter, and updating the filter according to the updating step size mu (k);
step (E), recalculating the current frame according to the updated filter coefficient to obtain the posterior error Ep(k)；
Step (F) of applying the posterior error Ep(k) And (4) performing inverse Fourier transform to obtain a time domain signal, taking the latter half data as a final output signal e (n) after echo cancellation processing, and returning to the step (A) to process the next frame.
In the foregoing method for echo cancellation with variable step size based on fast convergence, step (B) calculates a priori error e (k) according to the far-end reference frequency domain signal x (k) and the microphone receiving frequency domain signal d (k), as shown in formula (1),
wherein the content of the first and second substances,
for updated filter coefficients, G is a constraint factor which ensures the coincidence of the time domain and the frequency domain, and the expression is shown in formula (2)，
The step-length-variable echo cancellation method based on the fast convergence characteristic, step (C), calculates the transmission parameter T of the filteri' comprising the following steps,
(C1) calculating the cross-power spectrum S of the far-end reference signal and the error signal of the current framexeSelf-power spectral density S of remote reference signalxxSelf-power spectral density S of error signaleeAnd calculating an estimated signal
Cross power spectrum of received signals of microphone
Estimating a signal
Self-power spectral density of
Self-power spectral density S of microphone receiving signalddThe calculation process is shown in formulas (3.1) - (3.5),
Sxe＝αSxe+(1-α)X(k)E*(k)(3.1)
Sxx＝αSxx+(1-α)X(k)X*(k) (3.2)
See＝αSee+(1-α)E(k)E*(k) (3.3)
Sdd＝αSdd+(1-α)D(k)D*(k) (3.5)
wherein α is a smoothing factor with a value of 0.93,*in order to take the conjugate operation,
a frequency domain signal that is an estimated signal;
(C2) calculating a correlation measurement factor gamma of the remote reference signal and the error signal according to equation (4)xe(k) Calculating a correlation measurement factor of the estimated signal and the microphone received signal according to equation (5)
(C3) Selecting gamma according to the formula (6)xe(k) As the filter decision factor ζ,
wherein f ish＝fL+h(fH-fL)/H，[fL,fH]To determine the band range, fhIs the h frequency point, fLTo decide the lower limit of the frequency band, fHTo decide the upper limit of the frequency band;
(C4) mapping a filter decision factor zeta into a soft decision factor SD by adopting a Sigmoid function and a nonlinear truncation function, and limiting the value range to [ T [ ]min,η]The calculation is as shown in equation (7),
wherein, TminThe minimum value of the decision factor is η, the maximum value of the decision factor is β, and the compression factor of the Sigmoid function is β;
(C5) correcting the transmission parameters, at the beginning of filter completionAfter convergence begins, the ICflag state flag bit is set to 1, and the soft decision factor SD is used for correcting the transmission parameter T of the filteriWherein i represents the frequency domain block number, only if so to avoid the effect of double talk
Above a certain threshold τ, the soft decision factor SD is only applied to the transfer parameter TiWeighting to obtain a corrected transmission parameter Ti', as shown in the calculation formula (8),
wherein, ICflag is Initial convergence flag, that is, Initial convergence flag bit, and when the system completes Initial convergence, the value is automatically set to 1; the threshold τ is 0.93.
The step-size-variable echo cancellation method based on the fast convergence characteristic, step (D), calculates the update step size μ (k) of the filter, as shown in equation (9),
wherein T is the transmission parameter of the filter and takes the value of [0.99, 1%]，XH(k) Is the conjugate transpose of x (k), p (k) is the variance of the state estimation error, as shown in equation (10),
Ψvvfor the estimation, as shown in equation (11),
ΨΔΔ(k)＝(1-T2)E[|W(k)|2](11)
wherein W (k) is a true echo path,
are continuously estimated and updated filter coefficients.
The foregoing variable step size echo cancellation method based on fast convergence characteristics updates the filter according to the update step size μ (k), as shown in equation (12),
the step-length-variable echo cancellation method based on the fast convergence characteristic includes step (E), recalculating the current frame according to the updated filter coefficient to obtain the posterior error Ep(k) As shown in the formula (13),
the invention has the beneficial effects that: the variable-step-size echo cancellation method based on the rapid convergence characteristic effectively improves the tracking performance of the system when the echo path is suddenly changed under the condition of ensuring small detuning amount, has small calculation amount, and has good application prospect in a real-time voice conversation system and a scene with higher real-time requirement.
Has the following advantages:
(1) by combining the state information of the filter and mapping and weighting the state information to the transmission parameters of the filter, the transmission parameters T of the filter are corrected according to the soft decision factor SD when the acoustic path changes under the condition of ensuring the detuning amount, so that the filter can effectively and quickly track, and the phenomenon of locking the filter is avoided;
(2) when modifying the transfer parameter of the filter, a correlation measurement factor of the estimated signal and the received signal of the microphone is introduced
Only when
When the value is larger than a certain threshold, the soft decision value is only used for the parameter TiAre weighted at
The strategy canThe influence of the double-talk state on the filter transmission parameter is effectively avoided;
(3) compared with the original scheme, the method has small added calculation and is suitable for a real-time conference system or low-power-consumption equipment. The balance among the steady-state detuning amount, the convergence tracking performance and the calculation complexity is proper.
Drawings
Fig. 1 is a flow chart of the variable step size echo cancellation method based on the fast convergence characteristic of the present invention.
Detailed Description
The invention will be further described with reference to the accompanying drawings.
The variable step echo cancellation method based on the rapid convergence characteristic is characterized in that a first-order Markov model is used for representing a time-varying acoustic path, and Kalman gain is used for replacing the updating step of a filter; calculating an amplitude square coherence function of the reference signal and the residual error, and taking the amplitude square coherence function as filter state information; mapping the state information of the filter into a soft decision value, and weighting the channel transmission parameter; the acoustic path is tracked by using the weighted channel transfer parameters, so as to quickly complete the convergence process of the adaptive filter, as shown in fig. 1, specifically including the following steps,
respectively carrying out Fourier transform on a far-end reference signal and a microphone receiving signal d of a current frame to obtain a corresponding far-end reference frequency domain signal X (k) and a corresponding microphone receiving frequency domain signal D (k), wherein k is a frequency point label and does not lose generality, and if N-point Fourier transform is adopted, the value of the frequency point label k is 0-N-1;
step (B), calculating a priori error E (k) according to the far-end reference frequency domain signal X (k) and the microphone receiving frequency domain signal D (k), as shown in formula (1),
wherein the content of the first and second substances,
to be updatedFilter coefficients, G is a constraint factor which ensures the coincidence of the time domain and the frequency domain, and the expression is shown in formula (2),
wherein F is a discrete Fourier transform matrix, ILA unit array representing a size of L × L;
step (C) of calculating a transmission parameter T of the filteri' comprising the following steps,
(C1) calculating the cross-power spectrum S of the far-end reference signal and the error signal of the current framexeSelf-power spectral density S of remote reference signalxxSelf-power spectral density S of error signaleeAnd calculating an estimated signal
And cross power spectral density of microphone received signal
Estimating a signal
Self-power spectral density of
Self-power spectral density S of microphone receiving signalddThe calculation process is shown in formulas (3.1) - (3.5),
Sxe＝αSxe+(1-α)X(k)E*(k) (3.1)
Sxx＝αSxx+(1-α)X(k)X*(k) (3.2)
See＝αSee+(1-α)E(k)E*(k) (3.3)
Sdd＝αSdd+(1-α)D(k)D*(k) (3.5)
the cross-power spectral density of the microphone received signal is calculated by the formula:
to estimate a signal
α is a smoothing factor, with a value of 0.93,*in order to take the conjugate operation,
a frequency domain signal that is an estimated signal;
(C2) calculating a correlation measurement factor gamma of the remote reference signal and the error signal according to equation (4)xe(k) Calculating a correlation measurement factor of the estimated signal and the microphone received signal according to equation (5)
(C3) Selecting gamma according to the formula (6)xe(k) As the filter decision factor ζ,
wherein f ish＝fL+h(fH-fL)/H，[fL,fH]To determine the band range, fhIs the h frequency point, fLFor deciding the lower frequency bandLimit, fHTo decide the upper limit of the frequency band;
(C4) mapping a filter decision factor zeta into a soft decision factor SD by adopting a Sigmoid function and a nonlinear truncation function, and limiting the value range to [ T [ ]min,η]The calculation is as shown in equation (7),
wherein, TminThe minimum value of the decision factor is η, the maximum value of the decision factor is β, and the compression factor of the Sigmoid function is β;
(C5) correcting transmission parameter, setting ICflag state flag bit to 1 after the filter completes initial convergence, and using soft decision factor SD to correct transmission parameter T of filteriWherein i represents the frequency domain block number, only if so to avoid the effect of double talk
Above a certain threshold τ, the soft decision factor SD is only applied to the transfer parameter TiWeighting to obtain a corrected transmission parameter Ti', as shown in the calculation formula (8),
wherein, ICflag (initial convergence flag) is an initial convergence flag bit, and when the system finishes initial convergence, the value is automatically set to 1; the value of the threshold tau is 0.93;
step (D), calculating the updating step size mu (k) of the filter, and updating the filter according to the updating step size mu (k), wherein the updating step size mu (k) of the filter is calculated, as shown in formula (9),
where T is the filter's transfer parameter, typically [0.99, 1%]，XH(k) Is a conjugate transpose of X (k), P (k) is a state estimation errorThe variance, as shown in equation (10),
ΨΔΔto estimate, ΨΔΔ(k) As the variance of the filter coefficients, as shown in equation (11),
ΨΔΔ(k)＝(1-T2)E[|W(k)|2](11)
wherein W (k) represents the true echo path,
is the updated filter coefficients;
the filter is updated according to the update step size mu (k), as shown in equation (12),
step (E), recalculating the current frame according to the updated filter coefficient to obtain the posterior error Ep(k) As shown in the formula (13),
step (F) of applying the posterior error Ep(k) And (3) performing inverse Fourier transform to obtain a time domain signal, taking the latter half data as a final output signal e (n) after echo cancellation processing, wherein in an ideal case, e (n) only contains a near-end signal and does not contain a coupling signal of a loudspeaker, and returning to the step (A) to process the next frame.
In summary, the variable-step-size echo cancellation method based on the fast convergence characteristic of the present invention effectively improves the tracking performance of the system when the echo path is suddenly changed under the condition of ensuring that the amount of detuning is small, has a small amount of calculation, has a good application prospect in a real-time voice dialog system and a scene with a high real-time requirement, and does not have the following advantages:
(1) by combining the state information of the filter and mapping and weighting the state information to the transmission parameters of the filter, the transmission parameters T of the filter are corrected according to the soft decision factor SD when the acoustic path changes under the condition of ensuring the detuning amount, so that the filter can effectively and quickly track, and the phenomenon of locking the filter is avoided;
(2) when modifying the transfer parameter of the filter, a correlation measurement factor of the estimated signal and the received signal of the microphone is introduced
Only when
When the value is larger than a certain threshold, the soft decision value is only used for the parameter TiAre weighted at
The strategy can effectively avoid the influence of the double-talk state on the transmission parameters of the filter;
(3) compared with the original scheme, the method has small added calculation and is suitable for a real-time conference system or low-power-consumption equipment. The balance among the steady-state detuning amount, the convergence tracking performance and the calculation complexity is proper.
The foregoing illustrates and describes the principles, general features, and advantages of the present invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only to illustrate the principle of the present invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the present invention, which fall within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (4)
1. The variable step size echo cancellation method based on the rapid convergence characteristic is characterized in that: comprises the following steps of (a) carrying out,
respectively carrying out Fourier transform on a far-end reference signal x and a microphone receiving signal d of a current frame to obtain a corresponding far-end reference frequency domain signal X (k) and a corresponding microphone receiving frequency domain signal D (k), wherein k is a frequency point label and does not lose generality, and if N-point Fourier transform is adopted, the value of the frequency point label k is 0-N-1;
step (B), calculating a priori error E (k) according to the far-end reference frequency domain signal X (k) and the microphone receiving frequency domain signal D (k);
step (C) of calculating a transmission parameter T of the filteri′；
Step (D), calculating the updating step size mu (k) of the filter, and updating the filter according to the updating step size mu (k);
step (E), recalculating the current frame according to the updated filter coefficient to obtain the posterior error Ep(k)；
Step (F) of applying the posterior error Ep(k) Performing inverse Fourier transform to obtain time domain signal, taking the latter half data as final output signal e (n) after echo cancellation, returning to step (A), processing the next frame,
wherein, in the step (B), the prior error E (k) is calculated according to the far-end reference frequency domain signal X (k) and the microphone receiving frequency domain signal D (k), as shown in the formula (1),
wherein the content of the first and second substances,
for updated filter coefficients, G is a constraint factor that ensures the coincidence of the time domain and the frequency domain, and the expression is shown in formula (2),
step (C), calculating a transfer parameter T 'of the filter'iThe method comprises the following steps of,
(C1) calculating the cross-power spectrum S of the far-end reference signal and the error signal of the current framexeSelf-power spectral density S of remote reference signalxxSelf-power spectral density S of error signaleeAnd calculating an estimated signal
Cross power spectrum of received signals of microphone
Estimating a signal
Self-power spectral density of
Self-power spectral density S of microphone receiving signalddThe calculation process is shown in formulas (3.1) - (3.5),
Sxe＝αSxe+(1-α)X(k)E*(k) (3.1)
Sxx＝αSxx+(1-α)X(k)X*(k) (3.2)
See＝αSee+(1-α)E(k)E*(k) (3.3)
Sdd＝αSdd+(1-α)D(k)D*(k) (3.5)
wherein α is a smoothing factor with a value of 0.93,*in order to take the conjugate operation,
a frequency domain signal that is an estimated signal;
(C2) calculating a correlation measurement factor gamma of the remote reference signal and the error signal according to equation (4)xe(k) Calculating a correlation measurement factor of the estimated signal and the microphone received signal according to equation (5)
(C3) Selecting gamma according to the formula (6)xe(k) As the filter decision factor ζ,
wherein f ish＝fL+h(fH-fL)/H，[fL，fH]To determine the band range, fhIs the h frequency point, fLTo decide the lower limit of the frequency band, fHTo decide the upper limit of the frequency band;
(C4) mapping a filter decision factor zeta into a soft decision factor SD by adopting a Sigmoid function and a nonlinear truncation function, and limiting the value range to [ T [ ]min，η]The calculation is as shown in equation (7),
wherein, TminThe minimum value of the decision factor is η, the maximum value of the decision factor is β, and the compression factor of the Sigmoid function is β;
(C5) correcting transmission parameter, after the filter completes initial convergence, setting ICflag state flag bit to 1, soft decision factor SD used for correcting transmission parameter T of filteriWherein i represents the frequency domain block number, only if so to avoid the effect of double talk
Above a certain threshold τ, the soft decision factor SD is only applied to the transfer parameter TiAre weighted to obtainTo a corrected transfer parameter T'iAs shown in the calculation formula (8),
wherein, ICflag is Initial convergence flag, that is, Initial convergence flag bit, and when the system completes Initial convergence, the value is automatically set to 1; the threshold τ is 0.93.
2. The fast convergence property based variable step size echo cancellation method according to claim 1, wherein: step (D), calculating the updating step size mu (k) of the filter, as shown in formula (9),
wherein T is the transmission parameter of the filter and takes the value of [0.99, 1%]，XH(k) Is the conjugate transpose of x (k), p (k) is the variance of the state estimation error, as shown in equation (10),
as shown in the formula (11),
ΨΔΔ(k)＝(1-T2)E[|W(k)|2](11)
wherein W (k) is a true echo path,
are updated filter coefficients.
3. The fast convergence property based variable step size echo cancellation method according to claim 2, wherein: the filter is updated according to the update step size mu (k), as shown in equation (12),
4. the fast convergence property based variable step size echo cancellation method according to claim 1, wherein: step (E), recalculating the current frame according to the updated filter coefficient to obtain the posterior error Ep(k) As shown in the formula (13),
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910231065.XA CN109754813B (en) | 2019-03-26 | 2019-03-26 | Variable step size echo cancellation method based on rapid convergence characteristic |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910231065.XA CN109754813B (en) | 2019-03-26 | 2019-03-26 | Variable step size echo cancellation method based on rapid convergence characteristic |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109754813A CN109754813A (en) | 2019-05-14 |
| CN109754813B true CN109754813B (en) | 2020-08-25 |
Family
ID=66409340
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910231065.XA Active CN109754813B (en) | 2019-03-26 | 2019-03-26 | Variable step size echo cancellation method based on rapid convergence characteristic |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109754813B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110767245B (en) * | 2019-10-30 | 2022-03-25 | 西南交通大学 | Voice communication self-adaptive echo cancellation method based on S-shaped function |
| CN110992923B (en) * | 2019-12-19 | 2022-06-07 | 浙江大华技术股份有限公司 | Echo cancellation method, electronic device, and storage device |
| CN111028856B (en) * | 2020-01-08 | 2022-01-28 | 西南交通大学 | Echo cancellation method with variable step length |
| CN111246037B (en) * | 2020-03-16 | 2021-11-16 | 北京字节跳动网络技术有限公司 | Echo cancellation method, device, terminal equipment and medium |
| CN111445917B (en) * | 2020-03-17 | 2023-02-10 | 浙江大华技术股份有限公司 | Echo cancellation method, device and computer storage medium |
| CN111916099B (en) * | 2020-10-13 | 2020-12-29 | 南京天悦电子科技有限公司 | Adaptive echo cancellation device and method for variable-step hearing aid |
| CN112802487B (en) * | 2020-12-31 | 2023-05-12 | 北京梧桐车联科技有限责任公司 | Echo processing method, device and system |
| CN113194385B (en) * | 2021-01-14 | 2023-03-10 | 四川湖山电器股份有限公司 | Subband self-adaptive feedback elimination method and system based on step size control |
| CN113393857A (en) * | 2021-06-10 | 2021-09-14 | 腾讯音乐娱乐科技（深圳）有限公司 | Method, device and medium for eliminating human voice of music signal |
Family Cites Families (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9100090B2 (en) * | 2013-12-20 | 2015-08-04 | Csr Technology Inc. | Acoustic echo cancellation (AEC) for a close-coupled speaker and microphone system |
| US9729968B2 (en) * | 2014-03-05 | 2017-08-08 | Texas Instruments Incorporated | Method and system for acoustic echo cancellation using cascaded kalman filtering |
| CN105391879B (en) * | 2015-12-09 | 2018-06-12 | 天津大学 | A kind of acoustic echo removing method of echoless residual dual end communication robust |
| CN106448695B (en) * | 2016-09-28 | 2019-09-03 | 天津大学 | A kind of change rank variable step affine projection method of dual end communication robust |
| CN108111704B (en) * | 2016-11-24 | 2019-12-24 | 中国科学院声学研究所 | Non-contact self-adaptive echo cancellation device and echo cancellation method |
| CN106533500B (en) * | 2016-11-25 | 2019-11-12 | 上海伟世通汽车电子系统有限公司 | A method of optimization Echo Canceller convergence property |
| US10170134B2 (en) * | 2017-02-21 | 2019-01-01 | Intel IP Corporation | Method and system of acoustic dereverberation factoring the actual non-ideal acoustic environment |
| US10229698B1 (en) * | 2017-06-21 | 2019-03-12 | Amazon Technologies, Inc. | Playback reference signal-assisted multi-microphone interference canceler |
| CN108172233B (en) * | 2017-12-12 | 2019-08-13 | 天格科技（杭州）有限公司 | The echo cancel method of signal and error signal regression vectors is estimated based on distal end |
| CN108806709B (en) * | 2018-06-13 | 2022-07-12 | 南京大学 | Self-adaptive acoustic echo cancellation method based on frequency domain Kalman filtering |
| CN109087665B (en) * | 2018-07-06 | 2022-11-11 | 南京时保联信息科技有限公司 | Nonlinear echo suppression method |
-
2019
- 2019-03-26 CN CN201910231065.XA patent/CN109754813B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109754813A (en) | 2019-05-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109754813B (en) | Variable step size echo cancellation method based on rapid convergence characteristic | |
| CN102461205B (en) | Acoustic multi-channel echo cancellation device and method for cancelling acoustic multi-channel echo | |
| US9426279B2 (en) | Systems and methods for echo cancellation and echo suppression | |
| US9768829B2 (en) | Methods for processing audio signals and circuit arrangements therefor | |
| US9699554B1 (en) | Adaptive signal equalization | |
| CN111445917B (en) | Echo cancellation method, device and computer storage medium | |
| CN104050971A (en) | Acoustic echo mitigating apparatus and method, audio processing apparatus, and voice communication terminal | |
| WO2006049260A1 (en) | Signal processing method, signal processing device, and signal processing program | |
| US9343073B1 (en) | Robust noise suppression system in adverse echo conditions | |
| CN106448695B (en) | A kind of change rank variable step affine projection method of dual end communication robust | |
| CN106657507A (en) | Acoustic echo cancellation method and device | |
| CN110992923B (en) | Echo cancellation method, electronic device, and storage device | |
| CN111145771A (en) | Voice signal processing method, processing device, terminal and storage medium thereof | |
| CN111742541A (en) | Acoustic echo cancellation method and apparatus | |
| CN110265054A (en) | Audio signal processing method, device, computer readable storage medium and computer equipment | |
| CN109559756B (en) | Filter coefficient determining method, echo eliminating method, corresponding device and equipment | |
| CN110956975A (en) | Echo cancellation method and device | |
| US11417351B2 (en) | Multi-channel echo cancellation with scenario memory | |
| CN112929506B (en) | Audio signal processing method and device, computer storage medium and electronic equipment | |
| TWI234941B (en) | Echo canceler, article of manufacture, and method and system for canceling echo | |
| Mayyas | Stereophonic acoustic echo cancellation using lattice orthogonalization | |
| CN106409307A (en) | Affine projection method with selective evolution affine projection orders | |
| Fukui et al. | Acoustic echo and noise canceller for personal hands-free video IP phone | |
| CN112185404B (en) | Low-complexity double-end detection method based on subband signal-to-noise ratio estimation | |
| US8144863B2 (en) | Method and apparatus for echo cancellation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |