CN111355855B - Echo processing method, device, equipment and storage medium - Google Patents
Echo processing method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN111355855B CN111355855B CN202010171540.1A CN202010171540A CN111355855B CN 111355855 B CN111355855 B CN 111355855B CN 202010171540 A CN202010171540 A CN 202010171540A CN 111355855 B CN111355855 B CN 111355855B
- Authority
- CN
- China
- Prior art keywords
- frequency domain
- echo
- signal
- output signal
- energy
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M9/00—Arrangements for interconnection not involving centralised switching
- H04M9/08—Two-way loud-speaking telephone systems with means for conditioning the signal, e.g. for suppressing echoes for one or both directions of traffic
- H04M9/082—Two-way loud-speaking telephone systems with means for conditioning the signal, e.g. for suppressing echoes for one or both directions of traffic using echo cancellers
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M7/00—Arrangements for interconnection between switching centres
- H04M7/006—Networks other than PSTN/ISDN providing telephone service, e.g. Voice over Internet Protocol (VoIP), including next generation networks with a packet-switched transport layer
Landscapes
- Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Telephone Function (AREA)
- Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)
Abstract
The embodiment of the invention discloses an echo processing method, an echo processing device, electronic equipment and a storage medium, wherein the method comprises the following steps: calculating the energy of residual echo contained in the frequency domain output signal according to the echo estimation signal and the frequency domain output signal; determining an update step length of the frequency domain adaptive filter based on the energy of the residual echo and the energy of the frequency domain output signal; updating the filter coefficient of the frequency domain adaptive filter by using the updating step length; echo cancellation is performed using the adaptive filter after updating the filter coefficients. The convergence speed of the frequency domain adaptive filter is adjusted according to the energy of the residual echo signal in the frequency domain signal and the energy of the output frequency domain signal, and the robustness of the frequency domain adaptive filter is improved.
Description
Technical Field
The present invention relates to the field of communications, and in particular, to an echo processing method, apparatus, device, and storage medium.
Background
In the echo cancellation (AEC) technique, the convergence speed of an Adaptive Filter (AF) is a key measure for performance. The step size in the filter coefficient of the adaptive filter affects the convergence speed of the frequency domain adaptive filter, and the larger the step size is, the faster the convergence speed is. The step size of the adaptive filter is typically a fixed value that is set. Wherein the adaptive filter comprises a frequency domain adaptive filter.
Currently, the frequency domain adaptive filter usually implements echo cancellation by increasing the step size to increase the convergence speed. However, due to the variability of the echo sound field, after the convergence rate of the frequency domain adaptive filter is increased, the risk of detuning is often encountered, and the robustness of the adaptive filter cannot be ensured.
Disclosure of Invention
The embodiment of the invention provides an echo processing method, device, equipment and storage medium, which determine the updating step length of a frequency domain adaptive filter according to the energy of residual echo in an output frequency domain signal and the energy of the output frequency domain signal, and update the filter of the adaptive frequency domain filter by using the updating step length so that the convergence speed of the frequency domain adaptive filter is adjusted according to the energy of the residual echo in the frequency domain signal and the energy of the output frequency domain signal, thereby improving the robustness of the frequency domain adaptive filter.
In a first aspect, an embodiment of the present application provides an echo processing method, where the method includes: calculating the energy of residual echo contained in the frequency domain output signal according to the echo estimation signal and the frequency domain output signal; the echo estimation signal is a frequency domain signal obtained by filtering with a frequency domain adaptive filter, and the frequency domain output signal is a frequency domain signal obtained by performing echo cancellation with the echo estimation signal; determining an update step size of the frequency domain adaptive filter based on the energy of the residual echo and the energy of the frequency domain output signal; updating the filter coefficients of the frequency domain adaptive filter using the update step size; performing echo cancellation using the adaptive filter after updating filter coefficients.
In an optional implementation manner, the calculating, according to the echo estimation signal and the frequency domain output signal, an energy of a residual echo included in the frequency domain output signal includes: calculating a power cross-correlation value of the echo estimation signal and the frequency domain output signal and a power auto-correlation value of the frequency domain output signal; calculating a linear regression coefficient of the power cross-correlation value and the power autocorrelation value as an echo leakage factor; and calculating the energy of a residual signal contained in the frequency domain output signal by using the echo leakage factor.
In an alternative implementation, the calculating the power cross-correlation value of the echo estimation signal and the frequency domain output signal and the power auto-correlation value of the frequency domain output signal includes: calculating a first statistical value of the power variation of the frequency domain output signal and a second statistical value of the power variation of the echo estimation signal; and calculating a power cross-correlation value of the echo estimation signal and the frequency domain output signal and a power auto-correlation value of the frequency domain output signal by using the first statistical value and the second statistical value.
In an alternative implementation, the determining the update step size of the frequency-domain adaptive filter based on the energy of the residual echo and the energy of the frequency-domain output signal includes: taking the ratio of the energy of the residual echo to the energy of the frequency domain output signal as the update step under the condition that the ratio is not larger than a step threshold; and taking the step threshold as the updating step when the ratio of the energy of the residual echo to the energy of the frequency domain output signal is larger than the step threshold.
In an optional implementation manner, after the calculating, according to the echo estimation signal and the frequency domain output signal, the energy of the residual echo included in the frequency domain output signal, and before the performing echo cancellation on the input signal by the adaptive filter, the method further includes: determining a correction coefficient of the frequency domain adaptive filter according to the energy of the residual echo and the energy of the frequency domain output signal; and updating the filter coefficient of the frequency domain adaptive filter by using the correction coefficient.
In an optional implementation, after performing echo cancellation on the input signal by the frequency-domain adaptive filter, the method further includes: and performing offset detection on the signal to be output obtained by performing echo cancellation.
In a second aspect, an embodiment of the present application provides an echo processing apparatus, including: a calculating unit, which calculates the energy of the residual echo contained in the frequency domain output signal according to the echo estimation signal and the frequency domain output signal; the echo estimation signal is a frequency domain signal obtained by filtering with a frequency domain adaptive filter, and the frequency domain output signal is a frequency domain signal obtained by performing echo cancellation with the echo estimation signal;
a determining unit, configured to determine an update step size of the frequency domain adaptive filter based on the energy of the residual echo and the energy of the frequency domain output signal;
an updating unit, configured to update the filter coefficients of the frequency-domain adaptive filter using the update step size;
a cancellation unit for performing echo cancellation using the adaptive filter after updating the filter coefficients.
In an optional implementation manner, the computing unit includes: a first calculating module, configured to calculate a power cross-correlation value of the echo estimation signal and the frequency domain output signal, and a power auto-correlation value of the frequency domain output signal; the determining module is used for calculating a linear regression coefficient of the power cross-correlation value and the power autocorrelation value as an echo leakage factor; and the second calculation module is used for calculating the energy of a residual signal contained in the frequency domain output signal by using the echo leakage factor.
In an optional implementation manner, the first calculation module is specifically configured to: calculating a first statistical value of the power variation of the frequency domain output signal and a second statistical value of the power variation of the echo estimation signal; and calculating a power cross-correlation value of the echo estimation signal and the frequency domain output signal and a power auto-correlation value of the frequency domain output signal by using the first statistical value and the second statistical value.
In an optional implementation manner, the determining unit is specifically configured to, when a ratio of the energy of the residual echo to the energy of the frequency domain output signal is not greater than a step threshold, use the ratio as the update step; and taking the step threshold as the updating step when the ratio of the energy of the residual echo to the energy of the frequency domain output signal is larger than the step threshold.
In an optional implementation manner, the determining unit is further configured to determine a correction coefficient of the frequency-domain adaptive filter according to the energy of the residual echo and the energy of the frequency-domain output signal; the determining unit is further configured to update the filter coefficients of the frequency-domain adaptive filter using the correction coefficients.
In an optional implementation, the apparatus further comprises: and the detection unit is used for carrying out offset detection on the signal to be output obtained by executing echo cancellation.
In a third aspect, an embodiment of the present application provides an electronic device, where the electronic device includes a receiver and a transmitter, and further includes: a processor adapted to implement one or more instructions; and a computer storage medium storing one or more instructions adapted to be loaded by the processor and to perform the method according to the first aspect as well as the optional implementations of the first aspect.
In a fourth aspect, embodiments of the present application provide a computer storage medium storing one or more instructions adapted to be loaded by a processor and to execute a method according to the first aspect and the optional implementation manner in the first aspect.
The embodiment of the invention provides an echo processing method, device, equipment and storage medium, which determine the updating step length of a frequency domain adaptive filter according to the energy of residual echo in an output frequency domain signal and the energy of the output frequency domain signal, and update the filter of the adaptive frequency domain filter by using the updating step length so that the convergence speed of the frequency domain adaptive filter is adjusted according to the energy of the residual echo in the frequency domain signal and the energy of the output frequency domain signal, thereby improving the robustness of the frequency domain adaptive filter.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments or the background art of the present application, the drawings required to be used in the embodiments or the background art of the present application will be described below.
Fig. 1 is a flowchart of an echo processing method according to an embodiment of the present application;
fig. 2 is a schematic diagram of a frequency-domain adaptive filter according to an embodiment of the present disclosure;
fig. 3 is a flowchart of another echo processing method according to an embodiment of the present application;
fig. 4 is a diagram of an adaptive echo cancellation processing architecture according to an embodiment of the present application;
fig. 5 is a diagram of another adaptive echo cancellation processing architecture according to an embodiment of the present application;
fig. 6 is a schematic structural diagram of an echo processing device according to an embodiment of the present application;
fig. 7 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
Detailed Description
The terms "first," "second," and "third," etc. in the description and claims of the present application and the above-described drawings are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. Furthermore, the terms "comprises" and "comprising," as well as any variations thereof, are intended to cover a non-exclusive inclusion, such as a list of steps or elements. A method, system, article, or apparatus is not necessarily limited to those steps or elements explicitly listed, but may include other steps or elements not explicitly listed or inherent to such process, system, article, or apparatus. "and/or" is used to indicate the selection of one or both between two objects to which it is connected. For example "A and/or B" means A, B or A + B.
The embodiment of the invention provides an echo processing method, and aims to more clearly describe the scheme of the invention. Some knowledge about the adaptive filter signal is described below.
In the process of real-time voice communication and Voice Over Internet Protocol (VOIP), a far-end signal emitted by a speaker of a near-end device is always picked up by a microphone of the near-end device, and if the voice signal collected by the microphone is not processed, a near-end user can hear the voice of speaking, namely echo, and experience is poor. In the field of man-machine interaction, because the sound emitted by the interactive terminal is picked up by the microphone and the speaking sound of the controller is picked up, if the sound emitted by the interactive terminal is not eliminated in the signal picked up by the microphone, the interactive terminal introduces strong interference when recognizing the speaking sound of the controller, the success rate of recognition is reduced, and finally interaction difficulty is caused.
To solve the above problems, the generation principle and characteristics of echo need to be analyzed in depth. Taking a mobile phone communication hands-free mode as an example, echo generation commonly originates from three ways, namely a direct path (direct path), early reflection (early reflection) and late reverberation (late reflection), wherein the energy of an echo signal in the direct path is the highest; the echo signals reflected in the early stage show a linear attenuation trend along with the time, and the time lasts for tens of milliseconds; the late reverberated echo signal is weak in energy and decays quickly to noise levels for over a hundred milliseconds. No matter the direct path, early reflection or late reverberation, the echo signal and the far-end signal are always linearly related, so that the echo is called linear echo and occupies more than 90% of the main energy of all echoes, and the adaptive filter is generally adopted in the industry to eliminate the echo. Naturally, the convergence speed and robustness become key indicators for overall performance measurement.
The basic idea of the adaptive filtering algorithm is to adaptively adjust the coefficients of the filter according to the characteristics of the input signal, so as to realize optimal filtering. The adaptive filter automatically adjusts the filter parameter at the current moment by using the filter parameter obtained at the previous moment, and the error between the echo estimation signal output by the adaptive filter and the actual echo signal is minimized by a statistical method, so that the optimal filtering is realized. The filter parameters determine the convergence speed of the adaptive filter. Therefore, how to determine the filter parameters of the adaptive filter is the key for performing echo cancellation by the adaptive filter.
At present, the method for adjusting the convergence rate of the adaptive filter mainly includes three methods:
one way is to adjust the step size by detecting the state of the echo sound field, such as the methods proposed in patent applications CN105391879B and CN 108353107A. Patent application CN105391879B uses the convergence statistic parameter η (n) as the judgment feature, η (n) satisfying the formula:
Determining the filter when η (n) is less than a threshold minThe convergence rate of (2) needs to be increased, and the update step length is taken as mumax(ii) a If eta (n) is larger than the threshold value, the filter is considered to be in a stable state, and the updating step length is the normal updating speed mu (n). Patent application CN108353107A uses the updated energy of the AF coefficient as the detection feature, and considers that the AF coefficient is in the double-talk state when the energy is small, and considers that the AF coefficient is in the other state when the energy is large, so as to adjust the convergence speed of the filter differently. One relies on accurate detection of the echogenic sound field. Because there is no robustness guarantee measure in the filter itself, the output parameter characteristics do not have accuracy, and the state detection based on these characteristics has low reliability, which finally causes the adaptive filter coefficient to have a larger imbalance risk, and the robustness can not be reliably ensured.
The second way is to adjust the convergence speed of the adaptive filter by using a dual filter, such as the adaptive echo cancellation method based on the dual filter proposed in patent application CN 1095256C. The adaptive echo cancellation uses a first filter A to track the change of echo, a second filter B is used for carrying out optimal coefficient backup, quality evaluation is respectively carried out on the coefficients of the first filter A and the second filter B by using quality standards (qa and qp) after each coefficient update of the first filter A, if the quality of the first filter A is superior to that of the second filter B, the coefficient of the second filter B is updated to be the coefficient of the first filter A, and an echo canceller uses the latest updated coefficient of the first filter A to carry out the next operation; if the quality of the second filter B is better than that of the first filter a, the coefficient of the first filter a is restored to the optimal coefficient backed up by the second filter B, and the echo canceller still uses the optimal coefficient for operation. The second method has the disadvantage that when the first filter a is maladjusted, the update is regarded as invalid, and the convergence rate is slowed down to a certain extent. And in the second mode, two or more times of filtering and updating are needed, the calculated amount is doubled, the complexity is high, and the method is not suitable for a hardware system with high requirements on power consumption.
The third way is to adjust the convergence speed of the adaptive filter by adaptive optimal step size, for example, a frequency domain adaptive echo cancellation method proposed in patent application CN 101888455B. The method uses the error signal to estimate the optimal step size and the warping coefficient, thereby updating the coefficient of the filter. In the method, the optimal step length is as follows:
Alpha is linear regression coefficient and satisfies the formula
Wherein
The cross-correlation value of the error signal and the echo estimate signal,
is the autocorrelation value of the error signal.
The third way is disadvantageous in that its power cross-correlation
In the calculation process, the power of the residual signal and the power of the echo estimation signal are directly used for smooth calculation, and in the double-talk process, because the power of the residual signal and the power of the echo estimation signal are both large,
the linear regression coefficient alpha is a large positive value, so that the actual sound field change cannot be reflected by the linear regression coefficient alpha; accordingly, the filter is erroneously updated with high intensity in the double-talk state, and finally causes misadjustment.
The echo processing method provided by the embodiment of the present application is described in detail below. It should be noted that the filter mentioned in the embodiment of the present application is a frequency-domain adaptive filter.
Fig. 1 is a flowchart of an echo processing method according to an embodiment of the present application. As shown in fig. 1, the method may include:
101. and calculating the energy of the residual echo contained in the frequency domain output signal according to the echo estimation signal and the frequency domain output signal.
The echo processing device calculates the energy of the residual echo contained in the frequency domain output signal according to the echo estimation signal and the frequency domain output signal. The echo processing device may be a terminal device in real-time voice communication, a terminal device in VOIP, or an interactive terminal in human-computer interaction, and is not particularly limited.
As shown in fig. 2, the echo estimation signal is a frequency domain signal obtained by filtering the far-end signal with a frequency domain adaptive filter, and the echo processing device performs filtering processing on the far-end frequency domain signal with the adaptive filter to obtain an echo estimation signal in the frequency domain. The frequency domain output signal is a frequency domain signal obtained by performing echo cancellation using the echo estimation signal. The echo processing device subtracts the echo estimation signal from the frequency domain signal (near-end frequency domain signal) input by the microphone through the adder to obtain the frequency domain output signal. The frequency domain output signal is also referred to as an error signal or residual signal in the adaptive echo processing architecture.
Reference is made to an adaptive echo processing architecture as shown in fig. 4 or fig. 5. Optionally, a near-end frequency domain signal Dn(k) And a far-end frequency domain signal Xn(k) The frequency domain signals are obtained by performing short-time Fourier transform (STFT) on the near-end signal d (n) and the far-end signal x (n). Near-end frequency domain signal Dn(k) And a far-end frequency domain signal Xn(k) The following formula is satisfied: dn(k)＝STFT[d(n)]，Xn(k)＝STFT[x(n)]Where x (n) is the near-end signal collected by the microphone, and d (n) is the far-end signal. n is the frame index at which the signal is sampled and k is the frequency index. In other embodiments, the near-end frequency domain signal and the far-end frequency domain signalThe end frequency domain signal may also be a frequency domain signal obtained by a fast fourier transform.
Echo estimation signal Yn(k) The following formula is satisfied:
wherein the content of the first and second substances,
is Xn(k) Conjugation of (a) Wn(k) Is the current filter coefficient of the frequency domain adaptive filter. Y (n) is an echo estimation signal in the time domain, which can be estimated by comparing Yn(k) And performing STFT (inverse STIFT) to obtain the product.
Output frequency domain signal En(k) The following formula is satisfied:
the echo processing device calculates the energy of the residual echo contained in the frequency domain output signal according to the echo estimation signal and the frequency domain output signal, and specifically comprises the following steps: the echo processing device calculates the power cross-correlation value of the echo estimation signal and the frequency domain output signal and the power auto-correlation value of the frequency domain output signal. And the echo processing device calculates the linear regression coefficient of the power cross-correlation value and the power autocorrelation value as an echo leakage factor. Finally, the echo processing device calculates the energy of the residual echo in the frequency domain output signal by using the echo leakage factor.
The echo processing device calculates a power cross-correlation value of the echo estimation signal and the frequency domain output signal, and a power auto-correlation value of the frequency domain output signal, and specifically includes: the echo processing device calculates a first statistical value of the power variation of the frequency domain output signal and a second statistical value of the power variation of the echo estimation signal. In some embodiments, the first statistical value PΔEn(k) The value satisfies the formula:
PΔEn(k)＝(1-α)PΔEn-1(k)+α(|En(k)|2-|En-1(k)|2) And α is a smoothing coefficient.
Second statistical value PΔYn(k) Satisfy the requirement ofFormula PΔYn(k)＝(1-α)PΔYn-1(k)+α(|Yn(k)|2-|Yn-1(k)|2)。
The echo processing device calculates the power cross-correlation value of the echo estimation signal and the frequency domain output signal and the power auto-correlation value of the frequency domain output signal by using the first statistical value and the second statistical value.
In some embodiments, the power cross-correlation value C of the echo estimation signal with the frequency domain output signalEYn(k) Satisfies the formula: cEYn(k)＝(1-β)CEYn-1(k)+βPΔEn(k)PΔYn(k) Wherein β is a smoothing coefficient.
Power autocorrelation value C of frequency domain output signalEEn(k) Satisfies the formula:
According to the above-mentioned power cross-correlation value CEYn(k) And a power autocorrelation value CEEn(k) In some embodiments, the echo leakage factor ηnSatisfies the formula:
leakage factor etanIs a statistical value, and thus in some embodiments the energy of the residual echo in the frequency domain output signal
Satisfies the formula:
in this embodiment, the power autocorrelation value, the power cross-correlation value, the echo leakage factor, and the energy of the residual echo are estimated values obtained by a statistical method. By using a statistical method, the correlation of the echo estimation signal and the output frequency domain signal along with the time change is calculated, and then the residual echo in the frequency domain output signal is determined through the echo leakage factor, so that the reliability of the calculated result can be improved.
102. And determining the updating step size of the frequency domain adaptive filter based on the energy of the residual echo and the energy of the frequency domain output signal.
The echo processing device determines an update step size of the frequency domain adaptive filter based on the energy of the residual echo and the energy of the frequency domain output signal. The echo processing device takes the ratio of the energy of the residual echo to the energy of the frequency domain output signal as an update step, in the case where the ratio is not greater than the step threshold. And the echo processing device takes the step threshold as an updating step under the condition that the ratio of the energy of the residual echo to the energy of the frequency domain output signal is greater than the step threshold.
Wherein the energy of the frequency domain output signal
Satisfies the formula:
where γ is a smoothing coefficient. The echo estimation signal calculates the energy of the frequency domain output signal through the variance of the frequency domain output signal to obtain a statistical value of the energy of the frequency domain output signal.
In some embodiments, the step size μ is updatedoptn(k) Satisfies the formula:
wherein, mumaxThe step size threshold is the maximum update step size set by the echo processing device. Mu.smaxIs a constant. In some embodiments, μmax1, based on the energy of the residual echo in the frequency domain output signal
Does not exceed the energy of the frequency domain output signal
The echo processing means sets the step size threshold to 1. In other embodiments, the step threshold may be other constant, for example, a maximum value 2 that the step size may take, and the step threshold may be adjusted according to practical situations, and is not limited herein.
103. And updating the filter coefficients of the frequency domain adaptive filter by using the updating step size.
The echo processing device updates the filter coefficients of the frequency domain adaptive filter using the update step size. Updated filter coefficient Wn+1(k) Satisfies the formula:
wherein, deltan(k) Are the correction coefficients in the filter coefficients. In some embodiments, δn(k) Is a constant. In other embodiments, δn(k) Is a variable of k.
104. Echo cancellation is performed using the adaptive filter after updating the filter coefficients.
The echo processing device uses the adaptive filter after updating the filter coefficient to execute echo cancellation, and a signal to be output is obtained. Frequency domain output signal En(k) Is the output signal in the frequency domain obtained by echo cancellation of the near-end signal without updating the filter coefficients. The time-domain output signal E (n) can be output by frequency-domain output signal En(k) And performing Fourier inverse transformation to obtain the target. The signal to be output is an output signal obtained by performing echo cancellation on the near-end signal after the filter coefficient is updated. The output signal may be a time domain signal or a frequency domain signal. Signal to be output E in the frequency domainn+1(k) Satisfies formula En+1(k)＝Dn+1(k)-Wn+1(k)X* n+1(k)。Dn+1(k) The near-end frequency domain signal after the filter coefficient update. X* n+1(k) For the far-end frequency domain signal X after the filter coefficient updatingn+1(k) Conjugation of (1). System for controlling a power supplyThe output signal is a time domain signal, and after the filter coefficient is updated, the system output signal E (n +1) can be obtained by comparing the signal E to be output on the frequency domainn+1(k) And performing Fourier inverse transformation to obtain the target.
According to the embodiment of the application, the updating step length of the frequency domain adaptive filter is determined according to the energy of the residual echo in the output frequency domain signal and the energy of the output frequency domain signal, and the filter of the adaptive frequency domain filter is updated by using the updating step length, so that the convergence speed of the frequency domain adaptive filter is adjusted according to the energy of the residual echo in the frequency domain signal and the energy of the output frequency domain signal. Through experiments, the power cross-correlation coefficient C is found in the case of the change of the echo propagation pathEYn(k) Is a continuously high value, CEEn(k) Gradually decreasing, update step size μ in the present applicationoptn(k) The increase, the convergence rate improves, the filter accelerates the renewal. In the case of doubletalk, the power cross-correlation coefficient CEYn(k) Decrease, CEEn(k) For a sustained high value, muoptn(k) And reducing, slowing down the convergence speed and slowing down and updating the filter. In the near-end monosyllabic or background noise scenario, PΔYn(k) Gradually approaches to 0, CEYn(k) Synchronization tends to be 0, muoptn(k) Take 0 and the filter is not updated. The updating portion length in the embodiment of the application can meet the requirement of the convergence speed of the scene with echo path change, double talk, single talk at the near end or only background noise. By using a statistical method, the convergence speed is adjusted according to the ratio of the energy of the residual echo in the output frequency domain signal to the energy of the output frequency domain signal, and the robustness of the frequency domain adaptive filter is improved.
Fig. 3 is a flowchart of an echo processing method according to an embodiment of the present application. The method shown in fig. 3 is a further refinement and refinement of the method shown in fig. 1, wherein the detailed parts of the same steps are not described in detail in the method. The method can comprise the following steps:
301. and calculating the energy of the residual echo contained in the frequency domain output signal according to the echo estimation signal and the frequency domain output signal.
The echo processing device estimates the signal and the frequency domain output according to the echoAnd outputting the signal, and calculating the energy of residual echo contained in the frequency domain output signal. Energy of residual echo
Satisfy the formula
Echo leakage factor etanSatisfies the formula:
CEYn(k) estimating a power cross-correlation value, C, of the signal with the frequency domain output signal for the echoEYn(k) Satisfies the formula:
CEYn(k)＝(1-β)CEYn-1(k)+βPΔEn(k)PΔYn(k) wherein β is a smoothing coefficient.
CEEn(k) Is the power autocorrelation value, C, of the frequency domain output signalEEn(k) Satisfies the formula:
PΔEn(k) A first statistical value, P, of the power variation of the frequency domain output signalΔEn(k) Satisfies the formula:
PΔEn(k)＝(1-α)PΔEn-1(k)+α(|En(k)|2-|En-1(k)|2) And α is a smoothing coefficient.
PΔYn(k) Estimating a second statistic, P, of the power variation of the signal for the echoΔYn(k) Satisfies the formula:
PΔYn(k)＝(1-α)PΔYn-1(k)+α(|Yn(k)|2-|Yn-1(k)|2) And α is a smoothing coefficient.
302. And determining the updating step size of the frequency domain adaptive filter based on the energy of the residual echo and the energy of the frequency domain output signal.
The echo processing device determines an update step size of the frequency domain adaptive filter based on the energy of the residual echo and the energy of the frequency domain output signal. Updating the step size muoptn(k) Satisfies the formula:
wherein, mumaxThe step size threshold is the maximum update step size set by the echo processing device.
For the energy of the frequency domain output signal, satisfy the formula
Gamma is a smoothing coefficient.
303. And determining a correction coefficient of the frequency domain adaptive filter according to the energy of the residual echo and the energy of the frequency domain output signal.
The echo processing device determines a correction coefficient of the frequency domain adaptive filter according to the energy of the residual echo and the energy of the frequency domain output signal.
Correction coefficient delta of frequency domain adaptive filtern(k) Satisfies the formula:
The present embodiment determines the correction coefficient of the frequency domain adaptive filter by the energy of the residual echo and the energy of the frequency domain output signal. To further improve the robustness of the frequency domain adaptive filter.
304. And updating the filter coefficient of the adaptive filter by using the updating step size and the correction coefficient.
The echo processing device updates the filter coefficients of the adaptive filter using the update step size and the correction coefficients. Updated filter coefficient Wn+1(k) Satisfies the formula:
305. echo cancellation is performed using the adaptive filter after updating the filter coefficients.
The echo processing device uses the adaptive filter after updating the filter coefficient to execute echo cancellation, and a signal to be output is obtained. The signal to be output may be a time domain signal or a frequency domain signal. Signal to be output E in the frequency domainn+1(k) Satisfies formula En+1(k)＝Dn+1(k)-Wn+1(k)X* n+1(k)。Dn+1(k) The near-end frequency domain signal after the filter coefficient update. X* n+1(k) For the far-end frequency domain signal X after the filter coefficient updatingn+1(k) Conjugation of (1).
306. And performing offset detection on the signal to be output obtained by performing echo cancellation.
The echo processing device carries out offset detection on the signal to be output obtained by executing echo cancellation, and carries out protection processing so that the value of the signal to be output after offset detection does not exceed the value of the near-end signal. Signal to be output En+1(k) After the detuning detection is performed, the formula is satisfied:
the system output signal E (n +1) after the filter coefficient is updated can pass through the signal E to be output after the detuning detection is carried outn+1(k) And performing Fourier inverse transformation to obtain the target.
Fig. 4 is a diagram of an adaptive echo cancellation processing architecture according to an embodiment of the present application. As shown in fig. 4, the far-end signal x (n) is emitted from a Speaker (SPK) and then collected by a Microphone (MIC) through an echo propagation path h (n). The near-end signal d (n) is composed of a human voice (voice), background noise (noise) and echo (echo) inputted from the near end. After the echo processing device performs STFT on a near-end signal d (n) and a received far-end signal x (n), echo cancellation is performed through an echo estimation signal Yn (k) to obtain a frequency domain output signal En (k), wherein En (k) meets the formula: en (k) ═ dn (k) — yn (k). The echo processing device further performs the detuning detection and the STIFT processing on the frequency domain output signal en (k) to obtain a system output signal e (n).
Fig. 5 is a diagram of another adaptive echo cancellation processing architecture according to an embodiment of the present application. In the processing architecture shown in fig. 5, the echo processing apparatus performs an STIFT transform on the echo estimation signal yn (k) to obtain an echo estimation signal y (n) in the time domain, and subtracts the echo estimation signal y (n) in the near-end signal d (n) by an adder to obtain a time-domain output signal e (n), where e (n) satisfies the following formula: e (n) ═ d (n) -y (n). The echo processing device then performs detuning detection on the time domain output signal e (n) to obtain a coefficient output signal.
The echo processing method provided by The application is also applicable to frequency domain adaptive filtering structures of other normalized Least Mean Square error algorithms (NLMS), Least Mean Square error algorithms (LMS), recursive Least Square algorithms (RLS), multi-block multi-delay-block frequency-domain algorithms (MDF) and FADF algorithms, and their variants.
Fig. 6 is a schematic structural diagram of an echo processing device according to an embodiment of the present application. As shown in fig. 6, the echo processing device includes:
calculating section 601 calculates the energy of the residual echo included in the frequency domain output signal from the echo estimation signal and the frequency domain output signal. The echo estimation signal is a frequency domain signal filtered by the frequency domain adaptive filter, and the frequency domain output signal is a frequency domain signal obtained by echo cancellation using the echo estimation signal.
A determining unit 602, configured to determine an update step size of the frequency-domain adaptive filter based on the energy of the residual echo and the energy of the frequency-domain output signal.
An updating unit 603 configured to update the filter coefficients of the frequency-domain adaptive filter with an update step size.
A cancellation unit 604 for performing echo cancellation using the adaptive filter after updating the filter coefficients.
In an optional implementation manner, the computing unit 601 includes: a first calculating module 6011, configured to calculate a power cross-correlation value between the echo estimation signal and the frequency domain output signal, and a power auto-correlation value of the frequency domain output signal; a second calculating module 6012, configured to calculate a linear regression coefficient of the power cross-correlation value and the power autocorrelation value as an echo leakage factor; a third calculating module 6013, configured to calculate the energy of the residual signal included in the frequency domain output signal by using the echo leakage factor.
In an optional implementation manner, the first computing module 6011 is specifically configured to: a first statistical value of the power variation of the frequency domain output signal and a second statistical value of the power variation of the echo estimation signal are calculated. And calculating a power cross-correlation value of the echo estimation signal and the frequency domain output signal and a power autocorrelation value of the frequency domain output signal by using the first statistical value and the second statistical value.
In an alternative implementation, the determining unit 602 is specifically configured to use the ratio as the update step size in a case that the ratio of the energy of the residual echo to the energy of the frequency domain output signal is not greater than the step size threshold. And taking the step threshold as an updating step under the condition that the ratio of the energy of the residual echo to the energy of the frequency domain output signal is larger than the step threshold.
In an alternative implementation, the determining unit 602 is further configured to determine the correction coefficient of the frequency-domain adaptive filter according to the energy of the residual echo and the energy of the frequency-domain output signal. The determining unit is further configured to update the filter coefficients of the frequency-domain adaptive filter using the correction coefficients.
In an optional implementation, the apparatus further comprises: a detecting unit 605, configured to perform offset detection on the signal to be output obtained by performing echo cancellation.
It should be understood that the division of each unit in the above echo processing device is only a logical division, and the actual implementation may be wholly or partially integrated into one physical entity, or may be physically separated. For example, the above units may be processing elements which are set up separately, or may be implemented by integrating the same chip, or may be stored in a storage element of the controller in the form of program codes, and a certain processing element of the processor calls and executes the functions of the above units. In addition, the units can be integrated together or can be independently realized. The processing element may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the method or the units above may be implemented by hardware integrated logic circuits in a processor element or instructions in software. The processing element may be a general-purpose processor, such as a Central Processing Unit (CPU), or may be one or more integrated circuits configured to implement the above method, such as: one or more application-specific integrated circuits (ASICs), one or more microprocessors (DSPs), one or more field-programmable gate arrays (FPGAs), etc.
Referring to fig. 7, an electronic device according to an embodiment of the present application is described below, where the electronic device includes:
fig. 7 is a schematic structural diagram of an electronic device according to an embodiment of the present application. As shown in fig. 7, the electronic device includes a processor 701, a memory 702, and a communication interface 703. The processor 701, the memory 702, and the communication interface 703 are connected to each other by a bus.
The memory 702 includes, but is not limited to, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), or a compact disc read-only memory (CDROM), and the memory 702 is used for related instructions and data. The communication interface 703 is used for receiving and transmitting data.
The processor 701 may employ a general-purpose Central Processing Unit (CPU), a microprocessor, an Application Specific Integrated Circuit (ASIC), or one or more Integrated circuits, for executing related programs to implement the echo Processing method provided by the foregoing embodiments.
The processor 701 may also be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the multi-connection method of the present application may be implemented by integrated logic circuits of hardware or instructions in the form of software in the processor 701. The processor 701 may also be a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, or discrete hardware components. The various methods, steps, and logic blocks disclosed in the embodiments of the present application may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present application may be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software module may be located in ram, flash memory, rom, prom, or eprom, registers, etc. storage media as is well known in the art. The storage medium is located in the memory 702, and the processor 701 reads information in the memory 702 and completes the echo processing method provided in the embodiment of the present application in combination with hardware thereof.
The communication interface 703 enables communication between the electronic device and other devices or a communication network using transceiver means such as, but not limited to, a transceiver. Bus 704 may include a pathway to transfer information between various components of the electronic device (e.g., memory 702, processor 701, communication interface 703).
The processor 701 in the electronic device is configured to read the program code stored in the memory 702 to implement the echo processing method provided by the foregoing embodiment.
In an embodiment of the present invention, there is provided a computer-readable storage medium storing a computer program which, when executed by a processor, implements: and calculating the energy of the residual echo contained in the frequency domain output signal according to the echo estimation signal and the frequency domain output signal. And determining the updating step size of the frequency domain adaptive filter based on the energy of the residual echo and the energy of the frequency domain output signal. And updating the filter coefficients of the frequency domain adaptive filter by using the updating step size. Echo cancellation is performed using the adaptive filter after updating the filter coefficients.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive various equivalent modifications or substitutions within the technical scope of the present invention, and these modifications or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (9)
1. An echo processing method, comprising:
calculating the energy of residual echo contained in the frequency domain output signal according to the echo estimation signal and the frequency domain output signal; the echo estimation signal is a frequency domain signal obtained by filtering with a frequency domain adaptive filter, and the frequency domain output signal is a frequency domain signal obtained by performing echo cancellation with the echo estimation signal;
determining an update step size of the frequency domain adaptive filter based on the energy of the residual echo and the energy of the frequency domain output signal;
updating the filter coefficients of the frequency domain adaptive filter using the update step size;
performing echo cancellation using the adaptive filter after updating filter coefficients.
2. The method of claim 1, wherein said calculating the energy of the residual echo contained in the frequency domain output signal based on the echo estimation signal and the frequency domain output signal comprises:
calculating a power cross-correlation value of the echo estimation signal and the frequency domain output signal and a power auto-correlation value of the frequency domain output signal;
calculating a linear regression coefficient of the power cross-correlation value and the power autocorrelation value as an echo leakage factor;
and calculating the energy of a residual signal contained in the frequency domain output signal by using the echo leakage factor.
3. The method of claim 2, wherein the calculating the power cross-correlation value of the echo estimation signal and the frequency domain output signal and the power auto-correlation value of the frequency domain output signal comprises:
calculating a first statistical value of the power variation of the frequency domain output signal and a second statistical value of the power variation of the echo estimation signal;
and calculating a power cross-correlation value of the echo estimation signal and the frequency domain output signal and a power auto-correlation value of the frequency domain output signal by using the first statistical value and the second statistical value.
4. The method of any of claims 1-3, wherein determining the update step size of the frequency-domain adaptive filter based on the energy of the residual echo and the energy of the frequency-domain output signal comprises:
taking the ratio of the energy of the residual echo to the energy of the frequency domain output signal as the update step under the condition that the ratio is not larger than a step threshold;
and taking the step threshold as the updating step when the ratio of the energy of the residual echo to the energy of the frequency domain output signal is larger than the step threshold.
5. The method according to any of claims 1-3, wherein after said calculating the energy of the residual echo contained in the frequency domain output signal from the echo estimation signal and the frequency domain output signal, before performing echo cancellation using the adaptive filter with updated filter coefficients, the method further comprises:
determining a correction coefficient of the frequency domain adaptive filter according to the energy of the residual echo and the energy of the frequency domain output signal;
and updating the filter coefficient of the frequency domain adaptive filter by using the correction coefficient.
6. The method of any of claims 1-3, wherein after echo canceling the input signal with the frequency domain adaptive filter, the method further comprises:
and performing offset detection on the signal to be output obtained by performing echo cancellation.
7. An echo processing device, comprising:
a calculating unit, which calculates the energy of the residual echo contained in the frequency domain output signal according to the echo estimation signal and the frequency domain output signal; the echo estimation signal is a frequency domain signal obtained by filtering with a frequency domain adaptive filter, and the frequency domain output signal is a frequency domain signal obtained by performing echo cancellation with the echo estimation signal;
a determining unit, configured to determine an update step size of the frequency domain adaptive filter based on the energy of the residual echo and the energy of the frequency domain output signal;
an updating unit, configured to update the filter coefficients of the frequency-domain adaptive filter using the update step size;
a cancellation unit for performing echo cancellation using the adaptive filter after updating the filter coefficients.
8. An electronic device comprising a receiver and a transmitter, characterized by further comprising:
a processor adapted to implement one or more instructions; and the number of the first and second groups,
a computer storage medium having stored thereon one or more instructions adapted to be loaded by the processor and to perform the method of any of claims 1 to 6.
9. A computer storage medium having stored thereon one or more instructions adapted to be loaded by a processor and to perform the method of any of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010171540.1A CN111355855B (en) | 2020-03-12 | 2020-03-12 | Echo processing method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010171540.1A CN111355855B (en) | 2020-03-12 | 2020-03-12 | Echo processing method, device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111355855A CN111355855A (en) | 2020-06-30 |
| CN111355855B true CN111355855B (en) | 2021-06-15 |
Family
ID=71196105
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010171540.1A Active CN111355855B (en) | 2020-03-12 | 2020-03-12 | Echo processing method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111355855B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111798827A (en) * | 2020-07-07 | 2020-10-20 | 上海立可芯半导体科技有限公司 | Echo cancellation method, apparatus, system and computer readable medium |
| CN116610903B (en) * | 2023-07-14 | 2023-10-10 | 国网山西省电力公司营销服务中心 | Electrical quantity vector estimation method, device and computer readable storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101917527A (en) * | 2010-09-02 | 2010-12-15 | 杭州华三通信技术有限公司 | Method and device of echo elimination |
| CN102065190A (en) * | 2010-12-31 | 2011-05-18 | 杭州华三通信技术有限公司 | Method and device for eliminating echo |
| CN104980600A (en) * | 2014-04-02 | 2015-10-14 | 想象技术有限公司 | Auto-tuning Of Non-linear Processor Threshold |
| US9613634B2 (en) * | 2014-06-19 | 2017-04-04 | Yang Gao | Control of acoustic echo canceller adaptive filter for speech enhancement |
| CN107026950A (en) * | 2017-05-04 | 2017-08-08 | 重庆第二师范学院 | A kind of frequency domain adaptive echo cancel method |
| CN109712637A (en) * | 2018-12-21 | 2019-05-03 | 珠海慧联科技有限公司 | A kind of Reverberation Rejection system and method |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7103177B2 (en) * | 2001-11-13 | 2006-09-05 | Oguz Tanrikulu | Reduced complexity transform-domain adaptive filter using selective partial updates |
| CN101119135B (en) * | 2007-07-04 | 2010-09-01 | 深圳市融创天下科技发展有限公司 | Step parameter regulation means and equipment for eliminating echo |
| US9479650B1 (en) * | 2015-05-04 | 2016-10-25 | Captioncall, Llc | Methods and devices for updating filter coefficients during echo cancellation |
| CN105491256B (en) * | 2015-12-09 | 2018-12-04 | 天津大学 | A kind of acoustic echo canceller startup stage steady step length regulating method |
| CN107134281A (en) * | 2017-05-04 | 2017-09-05 | 重庆第二师范学院 | Adaptive filter coefficient update method during a kind of adaptive echo is eliminated |
| CN109509482B (en) * | 2018-12-12 | 2022-03-25 | 北京达佳互联信息技术有限公司 | Echo cancellation method, echo cancellation device, electronic apparatus, and readable medium |
| CN109493878B (en) * | 2018-12-17 | 2021-08-31 | 嘉楠明芯(北京)科技有限公司 | Filtering method, device, equipment and medium for echo cancellation |
| CN109935238B (en) * | 2019-04-01 | 2022-01-28 | 北京百度网讯科技有限公司 | Echo cancellation method, device and terminal equipment |
-
2020
- 2020-03-12 CN CN202010171540.1A patent/CN111355855B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101917527A (en) * | 2010-09-02 | 2010-12-15 | 杭州华三通信技术有限公司 | Method and device of echo elimination |
| CN102065190A (en) * | 2010-12-31 | 2011-05-18 | 杭州华三通信技术有限公司 | Method and device for eliminating echo |
| CN104980600A (en) * | 2014-04-02 | 2015-10-14 | 想象技术有限公司 | Auto-tuning Of Non-linear Processor Threshold |
| US9613634B2 (en) * | 2014-06-19 | 2017-04-04 | Yang Gao | Control of acoustic echo canceller adaptive filter for speech enhancement |
| CN107026950A (en) * | 2017-05-04 | 2017-08-08 | 重庆第二师范学院 | A kind of frequency domain adaptive echo cancel method |
| CN109712637A (en) * | 2018-12-21 | 2019-05-03 | 珠海慧联科技有限公司 | A kind of Reverberation Rejection system and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111355855A (en) | 2020-06-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP4282260B2 (en) | Echo canceller | |
| CN105577961B (en) | Automatic tuning of gain controller | |
| US9426566B2 (en) | Apparatus and method for suppressing noise from voice signal by adaptively updating Wiener filter coefficient by means of coherence | |
| JP4463981B2 (en) | Signal processing apparatus and echo suppression apparatus | |
| US6928160B2 (en) | Estimating bulk delay in a telephone system | |
| US20090238373A1 (en) | System and method for envelope-based acoustic echo cancellation | |
| US20030007633A1 (en) | Double-talk detector suitable for a telephone-enabled PC | |
| CN111199748B (en) | Echo cancellation method, device, equipment and storage medium | |
| CN109273019B (en) | Method for double-talk detection for echo suppression and echo suppression | |
| WO2020181766A1 (en) | Voice signal processing method and device, apparatus, and readable storage medium | |
| CN111742541B (en) | Acoustic echo cancellation method, acoustic echo cancellation device and storage medium | |
| WO2010083641A1 (en) | Method and apparatus for detecting double talk | |
| WO2012158163A1 (en) | Non-linear post-processing for acoustic echo cancellation | |
| CN110995951B (en) | Echo cancellation method, device and system based on double-end sounding detection | |
| JP2003500936A (en) | Improving near-end audio signals in echo suppression systems | |
| CN111355855B (en) | Echo processing method, device, equipment and storage medium | |
| CN106448695A (en) | Double-end call robust variable-order variable-step-size affine projection method | |
| KR20060130067A (en) | Echo canceller having a series arrangement of adaptive filters with individual update control strategy | |
| CN106571147A (en) | Method for suppressing acoustic echo of network telephone | |
| CN110931032B (en) | Dynamic echo cancellation method and device | |
| CN106297816B (en) | Echo cancellation nonlinear processing method and device and electronic equipment | |
| CN111917926B (en) | Echo cancellation method and device in communication terminal and terminal equipment | |
| CN111445916B (en) | Audio dereverberation method, device and storage medium in conference system | |
| CN111989934B (en) | Echo cancellation device, echo cancellation method, signal processing chip, and electronic apparatus | |
| JP4888262B2 (en) | Call state determination device and echo canceller having the call state determination device |
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 |