Multi-channel echo cancellation method, apparatus, device, and storage medium

Abstract

The present disclosure provides a multi-channel echo cancellation method, device, equipment and storage medium. The method comprises: determining a first echo vector by using a Kalman filter according to input signals of the previous N frames; determining a near-end speech correlation matrix according to the first echo vector; updating the Kalman filter according to the near-end speech correlation matrix; determining a second echo vector by using the Kalman filter according to an input signal of a current frame; and obtaining a near-end speech signal of the current frame according to the input signal of the current frame and the second echo vector, so as to complete the cancellation of the echo signal. According to the present disclosure, the echo vector can be determined by using the Kalman filter, the near-end speech correlation matrix can be corrected according to the echo vector, and the Kalman filter can be updated, so as to reduce the calculation complexity, make the algorithm more adaptable, and thus improve the echo cancellation effect.

Description

Technical Field

[0001] This disclosure relates to the field of echo cancellation technology, and in particular to a multi-channel echo cancellation method, apparatus, device and storage medium. Background Technology

[0002] In reverberant environments, echoes are generated by the acoustic connection between the speaker and the microphone. These echoes can severely impact the performance of telephone calls, video conferencing, and far-field speech recognition. With the widespread use of low-cost devices in mobile phones and video conferencing systems, the amplifiers or speakers in these devices introduce significant nonlinearity into the echo path. In such cases, the performance of linear echo cancellation (AEC) is limited, necessitating more sophisticated nonlinear multichannel echo cancellation methods. Summary of the Invention

[0003] This disclosure presents a multi-channel echo cancellation method, apparatus, device, and storage medium to address deficiencies in related technologies.

[0004] In a first aspect, this disclosure provides a multi-channel echo cancellation method, characterized by comprising:

[0005] Based on the input signals of the previous N frames, a Kalman filter is used to determine the first echo vector; wherein, the input signal is a speech signal acquired based on a microphone array, including near-end speech signal and echo signal, and N is a positive integer;

[0006] Based on the first echo vector, determine the near-end speech correlation matrix;

[0007] The Kalman filter is updated based on the near-end speech correlation matrix;

[0008] Based on the input signal of the current frame, the second echo vector is determined using the Kalman filter;

[0009] Based on the input signal of the current frame and the second echo vector, the near-end speech signal of the current frame is obtained to complete the echo cancellation.

[0010] In one possible implementation, determining the first echo vector using a Kalman filter based on the first input signal from the previous N frames includes:

[0011] Determine the acoustic transfer function vector based on the input signals of the previous N frames;

[0012] The first echo vector is determined based on the input signals of the previous N frames and the acoustic transfer function vector.

[0013] In one possible implementation, determining the near-end speech correlation matrix based on the first echo vector includes:

[0014] Based on the first echo vector, the near-end speech correlation matrix is ​​determined using a minimum variance distortionless response filter.

[0015] In one possible implementation, determining the near-end speech correlation matrix based on the first echo vector using a minimum variance distortionless response filter includes:

[0016] Determine the echo correlation matrix based on the first echo vector;

[0017] Determine the input correlation matrix based on the input signal;

[0018] The near-end speech correlation matrix is ​​determined based on the echo correlation matrix and the input correlation matrix.

[0019] In one possible implementation, updating the Kalman filter based on the near-end speech correlation matrix includes:

[0020] Based on the near-end speech correlation matrix, the update equation for the Kalman gain matrix is ​​determined to update the Kalman filter.

[0021] In one possible implementation, the echo signal is modeled using a convolutional transfer function and a short-time Fourier transform.

[0022] Secondly, this disclosure provides an echo cancellation device, characterized in that it includes:

[0023] The first echo determination module is used to determine the first echo vector based on the input signal of the previous N frames using a Kalman filter; wherein the input signal is a speech signal acquired based on a microphone array, including near-end speech signal and echo signal, and N is a positive integer;

[0024] The matrix determination module is used to determine the near-end speech correlation matrix based on the first echo vector;

[0025] An update module is used to update the Kalman filter based on the near-end speech correlation matrix;

[0026] The second echo determination module is used to determine the second echo vector based on the input signal of the current frame using the Kalman filter;

[0027] The cancellation module is used to obtain the near-end speech signal of the current frame based on the input signal of the current frame and the second echo vector, so as to complete the cancellation of the echo signal.

[0028] Thirdly, this disclosure provides an electronic device, including:

[0029] One or more processors;

[0030] A memory that is communicatively connected to the one or more processors;

[0031] One or more computer programs, wherein the one or more computer programs are stored in the memory, and when the one or more computer programs are executed by the electronic device, cause the electronic device to perform the method provided in the first aspect above.

[0032] Fourthly, this disclosure provides a computer-readable storage medium storing computer instructions that, when executed on a computer, cause the computer to perform the method provided in the first aspect.

[0033] The technical solution provided in this disclosure includes at least the following beneficial effects:

[0034] The echo vector is determined by using a Kalman filter, and the near-end speech correlation matrix is ​​corrected based on the echo vector to update the Kalman filter, thereby reducing computational complexity, making the algorithm more adaptive, and thus improving the echo cancellation effect.

[0035] Other features and aspects of this disclosure will become clear from the following detailed description of exemplary embodiments with reference to the accompanying drawings. Attached Figure Description

[0036] Figure 1 This is a flowchart illustrating a multi-channel echo cancellation method according to an embodiment of the present disclosure;

[0037] Figure 2 A schematic diagram of echo cancellation principle according to an embodiment of the present disclosure;

[0038] Figure 3 Here is a block diagram of an echo cancellation algorithm according to an embodiment of the present disclosure;

[0039] Figure 4 This is a schematic diagram of the structure of an echo cancellation device according to an embodiment of the present disclosure;

[0040] Figure 5 This is a structural block diagram of an electronic device according to an embodiment of the present disclosure. Detailed Implementation

[0041] The present disclosure will now be described in further detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0042] Furthermore, to better illustrate this disclosure, numerous specific details are set forth in the following detailed description. Those skilled in the art will understand that this disclosure can be practiced without certain specific details. In some instances, methods, means, components, and circuits well known to those skilled in the art have not been described in detail in order to highlight the main points of this disclosure.

[0043] In related technologies, nonlinear multichannel echo cancellation methods require independently updating the acoustic impulse response (AIR) experienced by the linear and nonlinear components of the reference signal. By performing eigenvalue decomposition (EVD) on the near-end speech correlation matrix to achieve rank-1 approximation, the computational load is greatly increased. Furthermore, it requires the use of a parameterized multi-microphone Wiener filter (PMWF). The PMWF filter is overly dependent on parameter selection, which may lead to unstable algorithm performance.

[0044] To at least partially solve the above problems, this disclosure provides a multi-channel echo cancellation method that does not require independent updating of the AIR experienced by nonlinear components, does not require EVD to improve algorithm performance, and avoids the instability of algorithm performance caused by improper selection of PMWF parameters. This effectively reduces the algorithm complexity and improves the algorithm's adaptability.

[0045] Figure 1 This is a flowchart of a multi-channel echo cancellation method according to an embodiment of the present disclosure. The method includes:

[0046] S101. Based on the input signals of the previous N frames, a first echo vector is determined using a Kalman filter; wherein, the input signal is a speech signal acquired based on a microphone array, including near-end speech signal and echo signal, and N is a positive integer.

[0047] S102. Determine the near-end speech correlation matrix based on the first echo vector.

[0048] S103. Update the Kalman filter based on the near-end speech correlation matrix.

[0049] S104. Based on the input signal of the current frame, determine the second echo vector using the Kalman filter.

[0050] S105. Based on the input signal of the current frame and the second echo vector, obtain the near-end speech signal of the current frame to complete the echo signal cancellation.

[0051] In this embodiment of the disclosure, the schematic diagram of nonlinear echo cancellation is as follows: Figure 2 As shown, the signal received by the m-th microphone is:

[0052]

[0053] In the formula s m (t), d m (t), h m x(t) and x(t) represent the near-end speech, nonlinear echo, AIR, and far-end speech received by the m-th microphone, respectively. Here, m is greater than or equal to 2, meaning the microphone array consists of at least two microphones. The microphone array can receive near-end speech, nonlinear echo, AIR, and far-end speech through multiple channels.

[0054] The nonlinear mapping function is:

[0055]

[0056] Where P is the nonlinear extended order, a i For the corresponding coefficient, φ i (·)=(·) i ,Right now,

[0057]

[0058] It should be noted that when the room reverberation is high, the echo signal can be approximated using the Convolutive Transfer Function (CTF), and further expressed as follows using the Short-Time Fourier Transform (STFT):

[0059]

[0060] Where k is the frame index, n is the temporal band index, L is the number of short-time frames, and Y... m (k,n), S m (k,n), X (i) (k,n) are respectively y m (t), s m (t), φ i Let the time-frequency representation of (x(t)) be:

[0061]

[0062] Therefore, we can obtain:

[0063]

[0064] Assuming the linear component of the input signal experiences the same AIR as the nonlinear component, then:

[0065]

[0066] make:

[0067] h l (k,n)=[H 1,l (k,n),…,H M,l (k,n)] T (8)

[0068] y(k,n)=[Y1(k,n),…,Y M (k,n)] T (9)

[0069] s(k,n)=[S1(k,n),…,S M (k,n)] T (10)

[0070] In the formula, M represents the number of microphones. Then, the matrix form of equation (6) is:

[0071]

[0072] further:

[0073]

[0074] in:

[0075]

[0076] make:

[0077]

[0078] Wherein, M×M order matrix I M It is an M-order identity matrix.

[0079] The ML×1 order acoustic transfer function (ATF) vector:

[0080]

[0081] Equation (3) can be further expressed as:

[0082]

[0083] at last:

[0084]

[0085] in, and These are the estimated ATF vector and echo vector, respectively, and e(k,n) is the error vector, which is the obtained near-end speech.

[0086] According to the scheme of this disclosure embodiment, the echo vector is determined by the Kalman filter, and the near-end speech correlation matrix is ​​corrected based on the echo vector to update the Kalman filter, thereby reducing the computational complexity, making the algorithm more adaptive, and thus improving the echo cancellation effect.

[0087] In one possible implementation, determining the first echo vector using a Kalman filter based on the first input signal from the previous N frames includes:

[0088] Determine the acoustic transfer function vector based on the input signals from the previous N frames.

[0089] The first echo vector is determined based on the input signals of the previous N frames and the acoustic transfer function vector.

[0090] In this embodiment of the disclosure, the algorithm block diagram for nonlinear echo cancellation is as follows: Figure 3 As shown, the Kalman estimation of the ATF vector is given by:

[0091]

[0092]

[0093] In the formula, A represents the inter-frame transmission coefficient. Let Φ be the ML×ML order error correlation matrix of h(k,n). W (k,n) is the noise correlation matrix, defined as:

[0094]

[0095] The update equation for the ML×M order Kalman gain matrix is:

[0096]

[0097] The above formula Φ e (k,n) is the near-end speech correlation matrix of the microphone array.

[0098]

[0099]

[0100] Where e + (k,n) represents the predicted near-end speech vector:

[0101]

[0102] After the update process is completed, the near-end audio is identical:

[0103]

[0104] It should be noted that N corresponds to L in formula (4). Taking N=3 as an example, the signal obtained after STFT transformation has one frequency point per row and one frame per column. Each row (one frequency point) is processed independently. The modeling method of the acoustic signal in this disclosure determines that only 3 consecutive frames of the row (frequency point) starting from the first column are processed each time to obtain the corresponding time-frequency point of the estimated echo signal. This signal is subtracted from the microphone input signal (STFT domain) to obtain the estimated near-end speech signal. After processing, the signal is shifted one position to the right and the next iteration continues.

[0105] In one possible implementation, determining the near-end speech correlation matrix based on the first echo vector includes:

[0106] Based on the first echo vector, the near-end speech correlation matrix is ​​determined using a minimum variance distortionless response (MVDR) filter.

[0107] In one possible implementation, determining the near-end speech correlation matrix based on the first echo vector using a minimum variance distortionless response filter includes:

[0108] The echo correlation matrix is ​​determined based on the first echo vector.

[0109] Determine the input correlation matrix based on the input signal;

[0110] The near-end speech correlation matrix is ​​determined based on the echo correlation matrix and the input correlation matrix.

[0111] It should be noted that MVDR can be replaced with a multichannel Wiener filter (MWF).

[0112] In one example, a minimum variance distortionless response filter is used to generate the multi-channel near-end speech correlation matrix. The gain function of the MVDR filter is:

[0113]

[0114] In the above formula, u m =[0,…,1,…] represents the microphone order. Φ y (k,n) represent the correlation matrices of the echo and the observed signal, respectively.

[0115] Φy (k,n)=α·Φ y (k,n-1)+(1-α)·y(k,n)y H (k,n) (27)

[0116]

[0117] make:

[0118]

[0119] final:

[0120]

[0121]

[0122] Formula (31) is the corrected near-end speech correlation matrix.

[0123] In one possible implementation, updating the Kalman filter based on the near-end speech correlation matrix includes:

[0124] Based on the near-end speech correlation matrix, the update equation for the Kalman gain matrix is ​​determined to update the Kalman filter.

[0125] In one possible implementation, the above method also includes modeling the echo signal using a convolutional transfer function and a short-time Fourier transform.

[0126] According to the embodiments of this disclosure, a nonlinear echo cancellation algorithm with minimum variance distortionless response (MVDR) filter correction is provided. By using this method, the computational complexity is reduced, the near-end speech correlation matrix is ​​corrected, the algorithm becomes more adaptive, and thus the echo cancellation effect is improved.

[0127] To address the issues of excessive computational cost and poor adaptability of GMA-NAEC and PMWF, this embodiment first recursively estimates the acoustic transfer function using a Kalman filter, thereby obtaining the echo vector. Then, it estimates the near-end speech correlation matrix using a minimum variance distortionless filter. This correlation matrix is ​​used as a parameter for the next recursive update of the Kalman filter. Correcting the near-end speech correlation matrix using an MVDR filter enhances the algorithm's robustness; and updating both the linear and nonlinear components (AIR) as a whole reduces the algorithm's complexity, resulting in high robustness and relatively low computational complexity.

[0128] Experimental results show that the four-channel MVDR nonlinear echo cancellation algorithm has better performance than GMA-NAEC under high signal-to-echo ratio, and the overall robustness of the algorithm is better than GMA-NAEC, while the algorithm complexity is also effectively improved.

[0129] Figure 4 This is a schematic diagram of an echo cancellation device according to an embodiment of the present disclosure. The device 400 includes:

[0130] The first echo determination module 401 is used to determine the first echo vector based on the input signal of the previous N frames using a Kalman filter. The input signal is a speech signal acquired based on a microphone array, including near-end speech signal and echo signal, where N is a positive integer.

[0131] The matrix determination module 402 is used to determine the near-end speech correlation matrix based on the first echo vector.

[0132] The update module 403 is used to update the Kalman filter based on the near-end speech correlation matrix.

[0133] The second echo determination module 404 is used to determine the second echo vector based on the input signal of the current frame using the Kalman filter.

[0134] The cancellation module 405 is used to obtain the near-end speech signal of the current frame based on the input signal of the current frame and the second echo vector, so as to complete the cancellation of the echo signal.

[0135] In one possible implementation, the first echo determination module 401 is used for:

[0136] Determine the acoustic transfer function vector based on the input signals from the previous N frames.

[0137] The first echo vector is determined based on the input signals of the previous N frames and the acoustic transfer function vector.

[0138] In one possible implementation, the matrix determination module 402 is used for:

[0139] Based on the first echo vector, the near-end speech correlation matrix is ​​determined using a minimum variance distortionless response filter.

[0140] In one possible implementation, the matrix determination module 402 includes:

[0141] The echo vector determination submodule is used to determine the echo correlation matrix based on the first echo vector.

[0142] The input signal determination submodule is used to determine the input correlation matrix based on the input signal.

[0143] The matrix determination submodule is used to determine the near-end speech correlation matrix based on the echo correlation matrix and the input correlation matrix.

[0144] In one possible implementation, update module 403 is used for:

[0145] Based on the near-end speech correlation matrix, the update equation for the Kalman gain matrix is ​​determined to update the Kalman filter.

[0146] In one possible implementation, the device also includes a modeling module for modeling the echo signal using a convolutional transfer function and a short-time Fourier transform.

[0147] The functions of each module and / or unit in the device embodiments of this disclosure can be found in the relevant descriptions in the above method embodiments of this disclosure, and will not be repeated here.

[0148] It should be noted that the division of functional units in this embodiment is illustrative and represents only one logical functional division. In actual implementation, other division methods may be used. The functional units in this embodiment can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0149] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods provided in the various embodiments of this disclosure. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory, random access memory, magnetic disks, or optical disks.

[0150] Figure 5 This is a structural block diagram of an electronic device according to an embodiment of the present disclosure. Figure 5As shown, the electronic device includes a memory 510 and a processor 520. The memory 510 stores a computer program that can run on the processor 520. The number of memories 510 and processors 520 can be one or more. The memory 510 can store one or more computer programs, which, when executed by the electronic device, cause the electronic device to perform the methods provided in the above-described method embodiments. The electronic device may also include a communication interface 530 for communicating with external devices and performing data exchange and transmission.

[0151] If the memory 510, processor 520, and communication interface 530 are implemented independently, they can be interconnected via a bus to communicate with each other. This bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0152] Optionally, in a specific implementation, if the memory 510, processor 520, and communication interface 530 are integrated on a single chip, then the memory 510, processor 520, and communication interface 530 can communicate with each other through an internal interface.

[0153] This disclosure also provides a computer-readable storage medium storing computer instructions that, when executed on a computer, cause the computer to perform the methods provided in the above-described method embodiments.

[0154] This disclosure also provides a computer program product for storing a computer program. When the computer program is executed by a computer, the computer can implement the methods provided in the above-described method embodiments.

[0155] This disclosure also provides a chip coupled to a memory, the chip being used to implement the method provided in the above method embodiments.

[0156] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. General-purpose processors can be microprocessors or any conventional processor. It is worth noting that the processor can be a processor supporting Advanced Reduced Instruction Set Machines (ARM) architecture.

[0157] Further, optionally, the aforementioned memory may include read-only memory and random access memory, and may also include non-volatile random access memory. The memory may be volatile or non-volatile, or may include both. Non-volatile memory may include read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may include random access memory (RAM), which serves as an external cache. Many forms of RAM are available by way of example, but not limitation. Examples include Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate Synchronous DRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct RAMBUS RAM (DR RAM).

[0158] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this disclosure are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line, DSL) or wireless (e.g., infrared, Bluetooth, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer, or a data storage device such as a server or data center that integrates one or more available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., Digital Versatile Discs (DVDs)), or semiconductor media (e.g., Solid State Disks (SSDs)). It is worth noting that the computer-readable storage media mentioned in this disclosure can be non-volatile storage media; in other words, it can be non-transient storage media.

[0159] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0160] In the description of the embodiments of this disclosure, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0161] In the description of the embodiments disclosed herein, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. "And / or" in this document is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0162] In the description of embodiments of this disclosure, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of embodiments of this disclosure, unless otherwise stated, "a plurality of" means two or more.

[0163] The above description is merely an exemplary embodiment of this disclosure and is not intended to limit this disclosure. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the protection scope of this disclosure.

Claims

1. A multi-channel echo cancellation method, characterized in that, include: Based on the input signals of the previous N frames, a Kalman filter is used to determine the first echo vector; wherein, the input signal is a speech signal acquired based on a microphone array, including near-end speech signal and echo signal, and N is a positive integer; Based on the first echo vector, the near-end speech correlation matrix is ​​determined using a minimum variance distortionless response filter, including: Determine the echo correlation matrix based on the first echo vector; Determine the input correlation matrix based on the input signal; The near-end speech correlation matrix is ​​determined based on the echo correlation matrix and the input correlation matrix; The Kalman filter is updated based on the near-end speech correlation matrix; Based on the input signal of the current frame, the second echo vector is determined using the Kalman filter; Based on the input signal of the current frame and the second echo vector, the near-end speech signal of the current frame is obtained to complete the echo cancellation.

2. The method according to claim 1, characterized in that, The step of determining the first echo vector using a Kalman filter based on the first input signal of the previous N frames includes: Determine the acoustic transfer function vector based on the input signals of the previous N frames; The first echo vector is determined based on the input signals of the previous N frames and the acoustic transfer function vector.

3. The method according to claim 1, characterized in that, The step of updating the Kalman filter based on the near-end speech correlation matrix includes: Based on the near-end speech correlation matrix, the update equation for the Kalman gain matrix is ​​determined to update the Kalman filter.

4. The method according to claim 1, further comprising: The echo signal is modeled using the convolutional transfer function and the short-time Fourier transform.

5. An echo cancellation device, characterized in that, include: The first echo determination module is used to determine the first echo vector based on the input signal of the previous N frames using a Kalman filter; wherein the input signal is a speech signal acquired based on a microphone array, including near-end speech signal and echo signal, and N is a positive integer; The matrix determination module is used to determine the near-end speech correlation matrix based on the first echo vector using a minimum variance distortionless response filter, including: Determine the echo correlation matrix based on the first echo vector; Determine the input correlation matrix based on the input signal; The near-end speech correlation matrix is ​​determined based on the echo correlation matrix and the input correlation matrix; An update module is used to update the Kalman filter based on the near-end speech correlation matrix; The second echo determination module is used to determine the second echo vector based on the input signal of the current frame using the Kalman filter; The cancellation module is used to obtain the near-end speech signal of the current frame based on the input signal of the current frame and the second echo vector, so as to complete the cancellation of the echo signal.

6. The apparatus according to claim 5, characterized in that, The matrix determination module is used to determine the near-end speech correlation matrix based on the first echo vector using a minimum variance distortionless response filter.

7. An electronic device, characterized in that, include: One or more processors; A memory that is communicatively connected to the one or more processors; One or more computer programs, wherein the one or more computer programs are stored in the memory, and when the one or more computer programs are executed by the electronic device, cause the electronic device to perform the method as described in any one of claims 1 to 4.

8. A computer-readable storage medium, characterized in that, It stores computer instructions that, when executed on a computer, cause the computer to perform the method as described in any one of claims 1 to 4.

Images