Filter coefficient determination method, echo elimination method and device

Abstract

The present application provides a filter coefficient determination method, an echo cancellation method and a device. The filter coefficient determination method includes: constructing multiple sub-remote-end signals based on a far-end signal; taking the error signal obtained by the next filtering of each filter equal to the near-end signal as a calculation condition, calculating the step length of each filter based on the sub-remote-end signal corresponding to each filter, wherein the product of the filter coefficient lengths of all filters is equal to the length of the far-end signal; and calculating the next filter coefficient of each filter based on the step length of each filter. The present application can improve the filter convergence speed.

Description

Technical Field

[0001] The present application relates to the field of filtering technology, and in particular to a filter coefficient determination method, an echo elimination method and a device. Background Art

[0002] In real life, when people use their mobile phones to turn on the hands-free phone or video conference terminals for video conferences, the sound played by the speakers will be picked up again by the microphones due to the external speakers, resulting in echo problems. Generally speaking, echo signals will seriously affect the call quality and reduce the accuracy of speech recognition, so adaptive filters are needed to eliminate the generated echoes. However, the existing adaptive filters have a slow convergence speed. Summary of the invention

[0003] The present application provides a filter coefficient determination method, an echo elimination method and a device to solve the problem of slow convergence speed of the existing adaptive filter.

[0004] To solve the above problems, the present application provides a method for determining filter coefficients, the method comprising:

[0005] constructing a plurality of sub-remote-end signals based on the remote-end signal;

[0006] The error signal obtained by the next filtering of each filter is equal to the near-end signal as a calculation condition, and the step size of each filter is calculated based on the sub-remote-end signal corresponding to each filter, wherein the product of the filter coefficient lengths of all filters is equal to the length of the far-end signal;

[0007] The next filter coefficient for each filter is calculated based on the step size of each filter.

[0008] The step of constructing a plurality of sub-remote-end signals based on the remote-end signal includes:

[0009] Calculation is performed based on the remote-end signal and current filter coefficients of other filters to obtain a sub-remote-end signal corresponding to each filter.

[0010] The step of constructing a plurality of sub-remote-end signals based on the remote-end signal includes:

[0011] constructing two sub-remote-end signals based on the remote-end signal;

[0012] There is a one-to-one correspondence between the two sub-remote-end signals and the two filters, and the product of the lengths of the filter coefficients of the two filters is equal to the length of the remote-end signal.

[0013] The step of calculating based on the far-end signal and the current filter coefficients of other filters includes:

[0014] Perform Kronecker multiplication on the current filter coefficients of other filters and the unit matrix to obtain the first matrix of other filters;

[0015] The first matrices of other filters are multiplied with the remote-end signal to obtain a sub-remote-end signal corresponding to each filter.

[0016] The step of calculating the step size of each filter based on the sub-remote-end signal corresponding to each filter, with the error signal obtained by the next filtering of each filter being equal to the near-end signal, comprises:

[0017] The near-end signal is substituted as the error signal of the next filtering into the first relationship formula of each filter. The first relationship formula is the relationship formula between the error signal of the next filtering, the error signal of the current filtering, the sub-remote-end signal and the step size of each filter to obtain the step size of each filter.

[0018] The first relationship formula of each filter is derived by using the next filter coefficient calculation formula of each filter, the next filtering error signal calculation formula and the current filtering error signal calculation formula.

[0019] Among them, the first relationship formula of the i-th filter is as follows:

[0020]

[0021] Where, v(n) is the near-end signal, e i (n) is the error signal currently filtered by the i-th filter, m i (n) is the remote sub-signal corresponding to the ith filter, is the transposed matrix of the remote sub-signal corresponding to the ith filter, μ i (n) is the step size of the i-th filter, and ξ is a preset constant.

[0022] The step of calculating the next filter coefficient of each filter based on the step size of each filter includes:

[0023] The filter coefficients of all filters are merged to obtain a total filter coefficient, so that the far-end signal can be processed by the total filter coefficient to obtain an analog echo signal of the far-end signal.

[0024] To solve the above problem, the present application provides an echo cancellation method, which includes:

[0025] Inputting the received far-end audio signal as a far-end signal into an adaptive filter to obtain a simulated echo signal, wherein the filter coefficient in the adaptive filter is determined according to the filter coefficient determination method described above;

[0026] A signal after echo cancellation processing is obtained based on the simulated echo signal and the near-end collected signal.

[0027] To solve the above problems, the present application provides an electronic device, which includes a processor; the processor is used to execute instructions to implement the steps of the above method.

[0028] In order to solve the above problems, the present application provides a computer storage medium on which instructions / program data are stored, and the steps of the above method are implemented when the instructions / program data are executed.

[0029] The method of the present application is: constructing multiple sub-remote-end signals based on a remote signal, wherein the multiple sub-remote-end signals correspond to the multiple filters one by one, so that the corresponding filters can be trained respectively using the multiple sub-remote-end signals, and since the product of the filter coefficient lengths of the multiple filters is equal to the length of the remote signal, the total number of filter coefficients updated in this implementation is less than the amount of filter coefficients in the existing "directly training a filter using the remote signal" scheme, and the multiple filters of this implementation can be updated simultaneously, thereby improving the speed of filter convergence. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the implementation modes of the present application, the drawings required for use in the description of the implementation modes will be briefly introduced below. Obviously, the drawings described below are only some implementation modes of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0031] Figure 1 It is a flowchart of an implementation method of a filter coefficient determination method 1 of the present application;

[0032] Figure 2 This is a flow chart of an implementation method of an echo cancellation method of the present application;

[0033] Figure 3 This is a schematic diagram of the structure of an embodiment of an electronic device of the present application;

[0034] Figure 4 It is a structural schematic diagram of an implementation method of a computer storage medium of the present application. DETAILED DESCRIPTION

[0035] In order to enable those skilled in the art to better understand the technical solution of the present application, the filter coefficient determination method, echo elimination method and device provided by the present application are further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0036] The terms "first", "second", and "third" in this application are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first", "second", and "third" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0037] Reference to "embodiment" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiment may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments in the absence of conflict.

[0038] like Figure 1 As shown, Figure 1 This is a flow chart of the first embodiment of the filter coefficient determination method of the present application. The filter coefficient determination method of the present application may include the following steps. Among them, the filter filtering algorithm of the present application may be a least mean square (Least Mean Square, LMS) algorithm or a normalized least mean square (Normalized Least Mean Squares, NLMS), etc., which are not limited here. For ease of description, the filter coefficient update process is described in the following content using the LMS algorithm as an example. It should be noted that the following step numbers are only used to simplify the description and are not intended to limit the execution order of the steps. The steps of this embodiment can be arbitrarily changed in execution order without violating the technical ideas of the present application.

[0039] S101: construct multiple sub-remote-end signals based on a remote-end signal.

[0040] A plurality of sub-remote-end signals can be constructed based on the remote-end signal, wherein the plurality of sub-remote-end signals correspond one-to-one to the plurality of filters. Thus, the plurality of sub-remote-end signals can be used to train the corresponding filters respectively. Moreover, since the product of the lengths of the filter coefficients of the plurality of filters is equal to the length of the remote-end signal, the total number of filter coefficients updated in this embodiment is less than the amount of filter coefficients in the existing scheme of "directly training a filter using the remote-end signal". Moreover, the plurality of filters in this embodiment can be updated simultaneously, thereby improving the speed at which the filters converge.

[0041] Optionally, calculations can be performed based on the remote signal and the current filter coefficients of other filters to obtain the sub-remote signal corresponding to each filter, so that the coefficient lengths of the sub-remote signal and the corresponding filter match, so that the filter can process its corresponding sub-remote signal, thereby being able to use the sub-remote signal to update the filter coefficients, and the filter coefficient update process of the two filters will use the data of the other filter, thereby ensuring that the filter can converge.

[0042] The "other filters" mentioned above may refer to at least one filter other than each filter itself among the multiple filters corresponding to the multiple sub-remote-end signals. For example, if the number of sub-remote-end signals is two and the number of filters is also two, the sub-remote-end signal corresponding to the first filter may be calculated using the current filter coefficient of the second filter, and the sub-remote-end signal corresponding to the second filter may be calculated using the current filter coefficient of the first filter.

[0043] Among them, the step of calculating based on the remote signal and the current filter coefficients of other filters may include: performing Kronecker multiplication on the current filter coefficients of other filters and the unit matrix to obtain the first matrix of other filters; multiplying the first matrix of other filters and the remote signal to obtain the sub-remote signal corresponding to each filter.

[0044] Further, the current filter coefficient of each filter can be divided into a plurality of coefficient data. Among them, the lengths of the plurality of coefficients of each filter can be equal. For example, the length of the current filter coefficient of a filter is 6 and is divided into two coefficient data, and the first three filter coefficients of the current filter coefficient of the filter can be used as the first coefficient data of the filter, and the last three filter coefficients of the current filter coefficient of the filter can be used as the second coefficient data of the filter.

[0045] The step of performing Kronecker multiplication operation on the current filter coefficients of other filters and the unit matrix can be specifically reflected as follows: performing Kronecker multiplication operation on each coefficient data of other filters and the unit matrix to obtain the first matrix of each coefficient data; accordingly, the step of multiplying the first matrix of other filters and the remote signal can be specifically reflected as follows: multiplying the first matrix of each coefficient data of other filters and the remote signal to obtain the signal processed by each coefficient data of other filters; then the signals processed by all coefficient data of other filters can be combined to obtain the sub-remote signal corresponding to each filter.

[0046] The specific calculation formula can be shown as follows:

[0047]

[0048]

[0049]

[0050]

[0051] in, is the d-th coefficient data of the ith filter, where d=[1,D], and D is the number of divisions of the current filter coefficients of each filter, which can be set according to actual conditions and is not limited here, for example, it can be 2; is the current filter coefficient of the i-th filter, is the unit matrix, L i is the filter coefficient length of the i-th filter, represents the Kronecker product, is the first matrix of the d-th coefficient data of the i-th filter, x(n) is the far-end signal, is the signal after the d-th coefficient data of the i-th filter is processed; x i (n) is the remote-end signal calculated using the current filter coefficient of the i-th filter. If i is in the range of 1 or 2, when i=1, x i (n) is the remote signal corresponding to the second filter. When i=2, x i (n) is the sub-remote-end signal corresponding to the first filter.

[0052] S102: Taking the error signal obtained by the next filtering of each filter as equal to the near-end signal as a calculation condition, the step length of each filter is calculated based on the sub-far-end signal corresponding to each filter.

[0053] After constructing multiple sub-remote-end signals based on step S101, the error signal obtained by the next filtering of each filter can be taken as a calculation condition to be equal to the near-end signal, and the step size of each filter can be calculated based on the sub-remote-end signal corresponding to each filter, so that the optimal step size can be calculated, so that the next filter coefficient calculated based on the calculated step size is better than the next filter coefficient calculated with a fixed step size, thereby accelerating the convergence speed of the filter.

[0054] The product of the filter coefficient lengths of all filters is equal to the length of the remote signal, and the filter coefficient lengths of all filters are greater than 1. In this way, the total number of filter coefficients updated by this implementation is smaller than the amount of filter coefficients in the existing "directly training a filter using a remote signal" scheme, thereby reducing the computational complexity and improving the convergence speed of the filter.

[0055] Optionally, the near-end signal can be substituted as the error signal of the next filtering into the first relationship formula of each filter to calculate the step size of each filter, wherein the first relationship formula is a relationship formula between the error signal of the next filtering, the error signal of the current filtering, the sub-remote-end signal and the step size of each filter.

[0056] Exemplarily, the first relationship formula of each filter may be respectively as follows:

[0057]

[0058] Where, v(n) is the near-end signal, e i (n) is the error signal currently filtered by the i-th filter, m i (n) is the remote sub-signal corresponding to the ith filter, is the transposed matrix of the remote sub-signal corresponding to the ith filter, μ i (n) is the step size of the ith filter, ξ is a preset constant that can be set according to actual conditions, for example, 0.0001 to ensure that the denominator does not appear to be 0. i (n) can be calculated according to step S101, which will not be described in detail here. For example, assuming that the number of sub-remote-end signals and filters is two, m1(n)=x2(n), m2(n)=x1(n).

[0059] The first relationship formula of each filter is derived by using the calculation formula of the next filter coefficient of each filter, the calculation formula of the error signal of the next filter, and the calculation formula of the error signal of the current filter. The next filter coefficient of each filter is the filter coefficient used in the next filtering of each filter, and the error signal of the next filtering of each filter is the error signal obtained in the next filtering of each filter.

[0060] The calculation formula of the error signal of the current filtering of each filter can be:

[0061] The calculation formula for the next filter coefficient of each filter can be:

[0062] The calculation formula of the next filtered error signal of each filter can be:

[0063] Among them, y(n) is the near-end acquisition signal, m i (n) is the remote sub-signal corresponding to the ith filter, is the transposed matrix of the current filter coefficients of the i-th filter, ei (n) is the current filtered error signal of the i-th filter, is the current filter coefficient of the i-th filter, is the next filter coefficient of the i-th filter, μ i (n) is the step size of the i-th filter, ∈ i (n) is the next filtered error signal of the i-th filter.

[0064] Optionally, the step of calculating the first relationship formula of each filter using the calculation formula of the next filter coefficient of each filter, the calculation formula of the error signal of the next filter, and the calculation formula of the error signal of the current filter may include: subtracting the calculation formula of the error signal of the next filter of each filter from the calculation formula of the error signal of the current filter of each filter to obtain the second relationship formula; substituting the calculation formula of the next filter coefficient of each filter into the second relationship formula of each filter to obtain the first relationship formula of each filter.

[0065] In order to avoid errors in step size calculation, which may cause the filter to fail to converge, the calculation formula of the next filter coefficient of each filter is substituted into the second relationship formula of each filter to obtain the third relationship formula of each filter, where the third relationship formula is an equation, and one end of the third relationship formula is the error signal of the next filter of each filter, and the expectation of the other end of the third relationship formula is calculated and set to be equal to the expectation of the near-end signal, thereby obtaining the first relationship formula of each filter.

[0066] Among them, the third relationship formula of each filter can be:

[0067]

[0068] The expectation of the other end of the third relationship formula and setting it equal to the expectation of the near-end signal can be specifically shown as follows:

[0069]

[0070] S103: Calculate the next filter coefficient of each filter based on the step size of each filter.

[0071] After the step length of each filter is calculated based on step S102 , the next filter coefficient of each filter can be calculated based on the step length of each filter.

[0072] Optionally, the calculation formula for the next filter coefficient of each filter (eg ) is used to calculate the next filter coefficient of each filter.

[0073] After calculating the next filter coefficient of each filter based on step S103, it can be determined whether each filter has converged; if the filter has converged, any filter can be used to perform echo elimination on the near-end acquisition signal, or the filter coefficients of all filters can be calculated to obtain the filter coefficient of a total filter, so that the total filter can be used to directly process the far-end signal corresponding to the near-end acquisition signal; if the filter has not converged, the next filter coefficient of each filter can be used as the current filter coefficient of each filter, and return to step S101 to update the filter coefficient again until the filter converges.

[0074] Assuming that the number of filters is two, the filter coefficients of all filters can be calculated using the following formula to obtain the filter coefficient of a total filter;

[0075]

[0076] is the d-th coefficient data of the ith filter, i=1 or 2, wherein d=[1,D], and D is the number of divisions of the current filter coefficients of each filter, which can be set according to actual conditions and is not limited here, for example, it can be 2; is the transposed matrix of the first matrix of the d-th coefficient data of the i-th filter, is the filter coefficient of the total filter.

[0077] In this embodiment, multiple sub-remote-end signals are constructed based on the remote signal, wherein the multiple sub-remote-end signals correspond to the multiple filters one-to-one, so that the corresponding filters can be trained respectively using the multiple sub-remote-end signals, and since the product of the filter coefficient lengths of the multiple filters is equal to the length of the remote signal, the total number of filter coefficients updated in this embodiment is less than the amount of filter coefficients in the existing "directly training a filter using the remote signal" scheme, and the multiple filters in this embodiment can be updated simultaneously, thereby improving the speed of filter convergence.

[0078] In order to better illustrate the filter coefficient determination method of the present application, the following specific implementation example of filter coefficient determination is provided for exemplary explanation.

[0079] Example

[0080] The essence of adaptive filtering is to find a finite impulse response so that the difference between the signal obtained by convolving the response with the far-end signal and the target echo signal is minimized, that is:

[0081]

[0082] h(n) is the received echo signal, d(n) is the target echo, x(n) is the far-end signal, is the quantity to be estimated;

[0083] Generally speaking, the near-end acquisition signal can be expressed as:

[0084] y(n)=ω T x(n)+v(n) (2)

[0085] y(n) represents the near-end acquisition signal, i.e., the input signal, v(n) represents the near-end signal, ω T x(n) represents the simulated echo signal;

[0086] Usually, the LMS algorithm is used to estimate ω, that is:

[0087]

[0088]

[0089] Where μ(n) is the step size and e(n) is the error signal;

[0090] This embodiment uses two filters to update filter coefficients, and can first construct two sub-remote-end signals using the current filter coefficients of the two filters and the remote-end signal;

[0091]

[0092]

[0093]

[0094]

[0095] Where i = 1, 2, where L1 × L2 = N; represents the Kronecker product, where D is the empirical coefficient, usually set to 2. is the unit matrix, L i is the filter coefficient length of the i-th filter;

[0096] The current filtered error signals of the two filters in this scheme can be expressed as the following two formulas respectively:

[0097]

[0098]

[0099] Among them, e1(n) and e2(n) are completely equal. The purpose of distinguishing e1(n) and e2(n) here is to keep them consistent with the two sub-remote-end signals split previously.

[0100] According to formula (4), the iteration of the adaptive filter can be converted into:

[0101]

[0102]

[0103] And equations (11) and (12) can be used to calculate the next error signal in equations (9) and (10);

[0104] Equations (9) and (10) are called the prior errors of adaptive filtering, and the a posteriori errors are calculated as follows:

[0105]

[0106]

[0107] Subtracting equations (13) and (14) from equations (9) and (10) respectively, and using the conversion formulas of equations (11) and (12), we can obtain:

[0108]

[0109]

[0110] Calculate the expectation of formula (15) and formula (16) and set it equal to the expectation of the near-end signal, that is:

[0111]

[0112]

[0113] Then we can get:

[0114]

[0115]

[0116] ξ is a constant, usually set to 0.0001, to ensure that the denominator does not appear to be 0;

[0117] After calculating the step sizes μ1(n) and μ2(n) of the two filters based on formula (19) and formula (20), the calculated step sizes μ1(n) and μ2(n) can be substituted into formula (11) and (12) to calculate the next filter coefficients of the two filters.

[0118] If the calculated next filter coefficients of the two filters converge, the next filter coefficients of the two filters can be calculated using the following formula to obtain the filter coefficient of the total filter:

[0119]

[0120] See also Figure 2 , Figure 2 FIG. 1 is a flow chart of an embodiment of the echo cancellation method of the present application. It should be noted that if there is substantially the same result, this embodiment is not used. Figure 2 The process sequence shown is limited. In this implementation, the echo cancellation method includes the following steps:

[0121] S301: Inputting the received far-end audio signal as a far-end signal into an adaptive filter to obtain an analog echo signal.

[0122] The filter coefficients in the adaptive filter are determined according to the above filter coefficient determination method.

[0123] S302: Obtain a signal after echo cancellation processing based on the simulated echo signal and the near-end collected signal.

[0124] See also Figure 3 , Figure 3 1 is a schematic diagram of the structure of an electronic device of the present application. The electronic device 10 includes a processor 12, and the processor 12 is used to execute instructions to implement the filter coefficient determination method and echo cancellation method. Please refer to the description of the above embodiment for the specific implementation process, which will not be repeated here.

[0125] The processor 12 may also be referred to as a CPU (Central Processing Unit). The processor 12 may be an integrated circuit chip having signal processing capabilities. The processor 12 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. A general-purpose processor may be a microprocessor or the processor 12 may also be any conventional processor, etc.

[0126] The electronic device 10 may further include a memory 11 for storing instructions and data required for the processor 12 to run.

[0127] The processor 12 is used to execute instructions to implement the method provided by any embodiment of the filter coefficient determination method and echo cancellation method of the present application and any non-conflicting combination.

[0128] See also Figure 4 , Figure 4The schematic diagram of the structure of the computer-readable storage medium in the embodiment of the present application. The computer-readable storage medium 30 of the embodiment of the present application stores instruction / program data 31, and when the instruction / program data 31 is executed, the method provided by any embodiment of the filter coefficient determination method and the echo elimination method of the present application and any non-conflicting combination is realized. Among them, the instruction / program data 31 can form a program file and be stored in the above-mentioned storage medium 30 in the form of a software product, so that a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (processor) executes all or part of the steps of each embodiment method of the present application. The aforementioned storage medium 30 includes: various media that can store program codes, such as a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, RandomAccess Memory), a disk or an optical disk, or a terminal device such as a computer, a server, a mobile phone, and a tablet.

[0129] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0130] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0131] The above are only implementation methods of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for determining filter coefficients, characterized in that: The method comprises: Constructing a plurality of remote-end sub-signals based on the remote-end signal, wherein the plurality of remote-end sub-signals correspond to a plurality of filters in a one-to-one manner; The step length of each filter is calculated based on the sub-remote-end signal corresponding to each filter, with the error signal obtained by the next filtering of each filter being equal to the near-end signal as a calculation condition, wherein the product of the filter coefficient lengths of all filters is equal to the length of the far-end signal; Calculating the next filter coefficient of each filter based on the step size of each filter; Wherein, constructing a plurality of sub-remote-end signals based on the remote-end signal comprises: constructing two sub-remote-end signals based on the remote-end signal; The step length of each filter is calculated based on the sub-remote-end signal corresponding to each filter, with the error signal obtained by the next filtering of each filter being equal to the near-end signal as a calculation condition, including: the step length of each filter is calculated based on the error signal of the current filtering of each filter and the sub-remote-end signal corresponding to each filter, with the error signal obtained by the next filtering of each filter being equal to the near-end signal as a calculation condition, and the sub-remote-end signal corresponding to each filter being the sub-remote-end signal calculated using the current filtering coefficient of another filter.

2. The filter coefficient determination method according to claim 1, characterized in that: The step of constructing a plurality of sub-remote-end signals based on the remote-end signal comprises: The remote-end signal and current filter coefficients of other filters are calculated to obtain a remote-end sub-signal corresponding to each filter, wherein the other filter refers to a filter among the multiple filters except each filter itself.

3. The filter coefficient determination method according to claim 2, characterized in that: The step of constructing a plurality of sub-remote-end signals based on the remote-end signal comprises: constructing two sub-remote-end signals based on the remote-end signal; The two sub-remote-end signals correspond to the two filters one by one, and the product of the lengths of the filter coefficients of the two filters is equal to the length of the remote-end signal.

4. The filter coefficient determination method according to claim 3, characterized in that: The step of calculating based on the far-end signal and the current filter coefficients of other filters comprises: Performing a Kronecker multiplication operation on the current filter coefficients of the other filter and the unit matrix to obtain a first matrix of the other filter; The first matrix of the other filters and the remote-end signal are multiplied to obtain a sub-remote-end signal corresponding to each of the filters.

5. The filter coefficient determination method according to claim 1, characterized in that: The step of calculating the step size of each filter based on the sub-remote-end signal corresponding to each filter, taking the error signal obtained by the next filtering of each filter as a calculation condition equal to the near-end signal, comprises: The near-end signal is substituted as the error signal of the next filtering into the first relationship formula of each filter, wherein the first relationship formula is the relationship formula between the error signal of the next filtering, the error signal of the current filtering, the sub-remote-end signal and the step size of each filter, so as to obtain the step size of each filter.

6. The filter coefficient determination method according to claim 5, characterized in that: The first relationship formula of each filter is derived by using the next filter coefficient calculation formula of each filter, the next filtering error signal calculation formula and the current filtering error signal calculation formula.

7. The filter coefficient determination method according to claim 5, characterized in that: The first relational formula for the ith filter is as follows: Among them, v(n) is the near-end signal, e1(n) is the error signal currently filtered by the i-th filter, and m i (n) is the remote sub-signal corresponding to the ith filter, is the transposed matrix of the remote sub-signal corresponding to the ith filter, μ i (n) is the step size of the i-th filter, and ξ is a preset constant.

8. The filter coefficient determination method according to claim 1, characterized in that: The step of calculating the next filter coefficient of each filter based on the step size of each filter comprises: The filter coefficients of all filters are merged to obtain a total filter coefficient, so as to process the far-end signal by the total filter coefficient to obtain a simulated echo signal of the far-end signal.

9. An echo cancellation method, characterized in that: The method comprises: Inputting the received far-end audio signal as a far-end signal into an adaptive filter to obtain a simulated echo signal, wherein a filter coefficient in the adaptive filter is determined according to the filter coefficient determination method according to any one of claims 1 to 7; A signal after echo cancellation processing is obtained based on the simulated echo signal and the near-end collected signal.

10. An electronic device, characterized in that: The electronic device comprises a processor; the processor is used to execute instructions to implement the steps of the method according to any one of claims 1-9.

11. A computer-readable storage medium having a program and / or an instruction stored thereon, characterized in that: When the program and / or instruction is executed, the steps of the method according to any one of claims 1 to 9 are implemented.

Images