CN117896468A - Deviation compensation echo cancellation method and system for telephone communication - Google Patents

Abstract

The application relates to the field of telephone communication, in particular to a deviation compensation echo cancellation method and a system for telephone communication, wherein the method comprises the following steps: respectively constructing an adaptive weight vector and a noise-containing input signal vector according to the adaptive weight and a noise-containing voice signal sampling value, and carrying out inner product on the adaptive weight vector and the noise-containing input signal vector to generate an output signal; generating an estimated error signal from the output signal and generating an error squared median therefrom; calculating an error signal variance estimation value and an adaptive weight vector power estimation value according to the error square median and the adaptive weight vector respectively; calculating an input noise variance estimation value according to the error signal variance estimation value and the self-adaptive weight vector power estimation value; calculating a deviation compensation term according to the augmentation weight vector, the self-adaptive weight vector and the input noise variance estimation value; and updating the self-adaptive weight vector according to the deviation compensation term, the noisy input signal vector and the estimated error signal. The adverse effect caused by input noise is effectively reduced.

Description

Deviation compensation echo cancellation method and system for telephone communication

Technical Field

The present invention relates to the field of telephone communication, and in particular to a deviation compensation echo elimination method and system for telephone communication.

Background technique

System identification is an important branch of adaptive signal processing. Many problems such as traditional adaptive channel equalization, adaptive noise cancellation, adaptive echo cancellation, active noise control, etc. can be attributed to system identification problems. The traditional least mean square (abbreviated as LMS) and normalized least mean square (abbreviated as NLMS) echo cancellation methods are easy to implement, but their performance will drop sharply in some special environments. For example, in some cases, the output signal of the unknown system may be contaminated by impulse noise. In order to enhance the anti-impulse interference ability of the echo cancellation method, a series of echo cancellation methods that are resistant to impulse noise have been proposed, such as the error-signed least mean square echo cancellation method, the mixed norm echo cancellation method, and the echo cancellation method based on maximum correlation entropy.

In addition, traditional echo cancellation methods are designed based on standard regression models, which assumes that the input signal obtained by the echo cancellation method is the same as the unknown system. However, due to sampling errors and other reasons in practical applications, the input signal is mixed with noise, which greatly affects the performance of traditional echo cancellation methods. To address this problem, deviation compensation methods and total least squares methods are proposed.

At present, scholars have proposed the bias-compensated maximum correlation entropy method (BC-MCC) and the bias-compensated minimum error entropy method (BC-MEE). Although these methods have good anti-impulse noise effects under models with error variables, they require the use of the instantaneous value of the noise when iteratively updating the weight vector. However, this data cannot be obtained in practical applications and therefore cannot be used for echo cancellation in telephone communications.

It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field.

Summary of the invention

To this end, the technical problem to be solved by the present invention is to overcome the problem that although the anti-pulse noise effect is good under the error variable model in the prior art, the instantaneous value of the noise needs to be used when the weight vector is iteratively updated. However, this data cannot be obtained in practical applications and therefore cannot be used for echo cancellation in telephone communications.

In order to solve the above technical problems, the first aspect of the present invention provides a deviation compensation echo cancellation method for telephone communication, the method comprising:

Obtaining adaptive weights of an echo canceller and sampling values of a noisy speech signal;

Constructing an adaptive weight vector and a noisy input signal vector according to the adaptive weight and the noisy speech signal sampling value respectively;

Performing an inner product of the adaptive weight vector and the noisy input signal vector to generate an output signal;

generating an estimated error signal according to the output signal, and generating a median square error according to the estimated error signal;

Calculating an error signal variance estimate and an adaptive weight vector variance power estimate according to the error square median and the adaptive weight vector respectively;

Calculate an input noise variance estimate based on the error signal variance estimate and the adaptive weight vector power estimate;

Constructing an augmented weight vector according to the adaptive weight vector;

Calculating a deviation compensation term based on the augmented weight vector, the adaptive weight vector and the input noise variance estimate;

The adaptive weight vector is updated according to the deviation compensation term, the noisy input signal vector and the estimated error signal.

In one embodiment of the present invention, the adaptive weight vector is expressed as:

;

in, is M adaptive weights, the subscript n represents the time, the superscript /> represents the transpose operation;

The noisy input signal vector is expressed as:

;

in, are M sampling values of the noisy speech signal.

In one embodiment of the present invention, the output signal is expressed as:

;

in, is the noisy input signal vector, /> is the adaptive weight vector, superscript/> Represents a transpose operation.

In one embodiment of the present invention, the estimated error signal is expressed as:

;

in, represents the noisy echo signal at time n, that is, the noisy expected signal.

In one embodiment of the present invention, the median square error is expressed as:

;

in, An integer representing the length of the median filter.

In one embodiment of the present invention, the error signal variance estimate is expressed as:

;

in, is the smoothing factor, ranging from 0.99 to 0.999, used for smoothing estimation,/> is the estimated value of the error signal variance at time n-1;

The adaptive weight vector power estimate is expressed as:

;

in, is the smoothing factor, ranging from 0.99 to 0.999, used for smoothing estimation, and M is the length of the adaptive weight vector;

The input noise variance estimate is expressed as:

;

in, is the ratio of output noise to input noise variance,/> is the variance of the measurement noise without impulse noise,/> is the variance of the input noise, and M is the length of the adaptive weight vector.

In one embodiment of the present invention, the augmented weight vector is expressed as:

;

in, is the noise ratio, superscript/> Represents a transpose operation.

In one embodiment of the present invention, the deviation compensation term is expressed as:

;

in, is a positive kernel width parameter, /> To find the 2-norm.

In one embodiment of the present invention, the formula for updating the adaptive weight vector is:

;

in, is a positive step parameter.

A second aspect of the present invention provides a deviation-compensated echo cancellation system for telephone communication, the system comprising: a data acquisition module, a first construction module, a first calculation module, a second calculation module, and a second construction module;

The data acquisition module is configured to: acquire the adaptive weight of the echo canceller and the sampling value of the noisy speech signal;

The first construction module is configured to: construct an adaptive weight vector and a noisy input signal vector according to the adaptive weight and the noisy speech signal sampling value respectively;

The first calculation module is configured to: perform an inner product of the adaptive weight vector and the noisy input signal vector to generate an output signal; generate an estimated error signal according to the output signal, and generate a square median error according to the estimated error signal;

The second calculation module is configured to: calculate an error signal variance estimate and an adaptive weight vector power estimate according to the error square median and the adaptive weight vector respectively; calculate an input noise variance estimate according to the error signal variance estimate and the adaptive weight vector power estimate;

The second construction module is configured to: construct an augmented weight vector based on the adaptive weight vector; calculate a deviation compensation term based on the augmented weight vector, the adaptive weight vector and the input noise variance estimate; and update the adaptive weight vector based on the deviation compensation term, the noisy input signal vector and the estimated error signal.

The above technical solution of the present invention has the following advantages compared with the prior art:

The present invention discloses a method and system for deviation compensation echo cancellation for telephone communication, which calculates a deviation compensation term and then updates an adaptive weight vector by the deviation compensation term, so as to achieve stronger robustness without the need to output the instantaneous value of the noise, and effectively reduce the adverse effects of the input noise.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to make the contents of the present invention more clearly understood, the present invention is further described in detail below based on specific embodiments of the present invention in conjunction with the accompanying drawings.

FIG1 is a flow chart of a deviation compensation echo cancellation method for telephone communication provided by the present invention;

FIG2 is a waveform diagram of an echo channel in a deviation compensation echo cancellation system and a system for telephone communication provided by the present invention;

3 is a waveform diagram of a voice signal in a deviation compensation echo cancellation method and system for telephone communication provided by the present invention;

4 is a normalized mean square deviation curve diagram of a deviation compensation echo cancellation method and system for telephone communication provided by the present invention when the input noise signal-to-noise ratio is 10 dB;

FIG5 is an architecture diagram of a deviation compensation echo cancellation system for telephone communication provided by the present invention.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

In addition, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

1, the present invention provides a deviation compensation echo cancellation method for telephone communication, the method comprising:

S100, obtaining an adaptive weight value of an echo canceller and a sampling value of a noisy speech signal;

In step S100, the current Time echo canceller/> Adaptive weights/> , get the current /> Moment and its previous continuous /> The sampling value of the noisy speech signal at a certain moment/> .

S200, constructing an adaptive weight vector and a noisy input signal vector according to the adaptive weight and the noisy speech signal sampling value respectively;

In step S200, the adaptive weight vector is expressed as:

(1);

Among them, Adaptive weights/> Composition, superscript /> represents the transpose operation;

The noisy input signal vector is expressed as:

(2);

Among them, the noisy speech signal sampling value constitute.

In actual application scenarios, the echo canceller at time n Adaptive weights/> Combine to form an adaptive weight vector, and combine the nth moment and its previous continuous / > The sampling value of the noisy speech signal at a certain moment/> Construct the noisy input signal vector.

S300, performing an inner product operation on the adaptive weight vector and the noisy input signal vector to generate an output signal;

In step S300, the output signal is represented as:

(3);

in, is the noisy input signal vector, /> is the adaptive weight vector.

In practical application scenarios, by converting the noisy input vector With adaptive weight vector/> Perform the inner product to calculate the output signal of the echo canceller at time n/> .

S400, generating an estimated error signal according to the output signal, and generating a square median error according to the estimated error signal;

In step S400, the estimated error signal is expressed as:

(4);

in, represents the noisy echo signal at time n, that is, the noisy expected signal. The median square error is expressed as:

(5);

in, An integer representing the length of the median filter.

In practical application scenarios, the estimated error signal at time n is calculated based on the output signal and formula (4), and then Whether the estimated error signal at the moment is interfered by impulse noise determines whether median filtering is performed. That is, if/> Then the median square error is calculated according to formula (5), otherwise the median square error is / > , where /> Yes/> Estimated value of variance of error signal at the moment;/> It is a threshold parameter used to determine whether it is affected by impulse noise, and can usually be between 9 and 25.

S500, calculating an error signal variance estimate and an adaptive weight vector power estimate according to the error square median and the adaptive weight vector respectively;

In step S500, the error signal variance estimate is expressed as:

(6);

in, is the smoothing factor, ranging from 0.99 to 0.999, used for smoothing estimation,/> is the estimated value of the error signal variance at time n-1;

The adaptive weight vector variance estimate is expressed as:

(7);

in, is a smoothing factor ranging from 0.99 to 0.999, used for smoothing estimation, and M is the length of the adaptive weight vector.

In practical application scenarios, the estimated variance of the error signal at time n is calculated according to formula (6): Among them, /> is the smoothing factor, which can usually be between 0.99 and 0.999, and is used for smoothing the estimation to ensure the accuracy of the estimation. According to formula (7), the adaptive weight vector variance estimate of the echo canceller at time n is calculated./> .

S600, calculating an input noise variance estimate according to the error signal variance estimate and the adaptive weight vector power estimate;

In step S600, the input noise variance estimate is expressed as:

(8);

in, is the ratio of output noise to input noise variance,/> is the variance of the measurement noise without impulse noise,/> is the variance of the input noise, and M is the length of the adaptive weight vector.

In practical application scenarios, the error signal variance estimate and the adaptive weight vector variance estimate are used to obtain the input noise variance of the echo canceller at time n: .

S700, constructing an augmented weight vector according to the adaptive weight vector;

In step S700, the augmented weight vector is expressed as:

(9);

in, is the noise ratio, superscript/> Represents a transpose operation.

In practical application scenarios, an augmented weight vector is constructed according to the noise ratio and the adaptive weight vector.

S800, calculating a deviation compensation term according to the augmented weight vector, the adaptive weight vector and the input noise variance estimate;

In step S800, the deviation compensation term is expressed as:

(10);

in, is a positive kernel width parameter, /> To find the 2-norm.

In practical application scenarios, the deviation compensation term uses the joint Gaussian distribution relationship between the estimation error and the noisy input, and is obtained by converting the expectation into an integral solution, that is, the initial deviation compensation term is , used to compensate for the estimated deviation of the echo channel introduced by the input noise, where, /> is the core width parameter, which is used to control the resistance to impulse noise interference;/> is the variance of the input noise; /> is the echo channel weight vector to be estimated; /> is the augmented echo channel weight vector,/> is the noise ratio, /> is the variance of the measurement noise in the desired signal without the interference of impulse noise. Bias compensation term/> The initial deviation compensation term is The input noise variance in and the echo channel weight vector Use the input noise variance at time n respectively/> and adaptive weight vector/> Obtained after replacement.

S900: Update the adaptive weight vector according to the deviation compensation term, the noisy input signal vector and the estimated error signal.

In step S900, the formula for updating the adaptive weight vector is:

(11);

in, is a positive step parameter.

In actual application scenarios, the weight vector of the echo canceller is updated according to the iterative formula, that is, formula (11), where: The step size is used to control the convergence speed. This embodiment uses a computer experiment method to verify the performance of the IBC-MCC echo cancellation method applied to the echo cancellation task in telephone communication and compares the experimental results with the existing methods: bias compensated least mean square (BC-LMS) echo cancellation method and maximum correlation entropy (MCC) echo cancellation method.

The experiment was conducted in the echo cancellation environment of telephone communication, using the echo channel shown in Figure 2 as the unknown system and the speech signal shown in Figure 3 as the input signal of the echo channel. The input noise is a Gaussian white signal with a signal-to-noise ratio of 10 dB. The measurement noise is generated by the mixed Gaussian model, that is, , where /> is a zero-mean Gaussian white noise satisfying a 30dB signal-to-noise ratio,/> is an impulse noise generated by a Bernoulli Gaussian process, i.e. /> ,/> is a Bernoulli process, and the probability of taking 0 is 0.95, and the probability of taking 1 is 0.05,/> is a zero-mean Gaussian white noise satisfying a -10dB signal-to-noise ratio. The normalized mean square deviation (NMSD) is used as the performance measure, i.e. , the unit is dB, where log means taking the logarithm, /> Represents the echo channel. The NMSD curves obtained by simulation in the figure are obtained by taking the average value of 100 independent iterations. As shown in FIG4 , the IBC-MCC echo cancellation method of the embodiment of the present application can effectively cancel the echo in the pulse environment containing input noise.

In a second aspect, as shown in FIG5 , the present application provides a deviation compensation echo cancellation system for telephone communication, the system comprising: a data acquisition module 100 , a first construction module 200 , a first calculation module 300 , a second calculation module 400 and a second construction module 500 ;

The data acquisition module 100 is configured to: acquire the adaptive weight of the echo canceller and the sampling value of the noisy speech signal;

The first construction module 200 is configured to: construct an adaptive weight vector and a noisy input signal vector according to the adaptive weight and the noisy speech signal sampling value respectively;

The first calculation module 300 is configured to: generate an output signal by performing an inner product of the adaptive weight vector and the noisy input signal vector; generate an estimated error signal according to the output signal, and generate a median square error according to the estimated error signal;

The second calculation module 400 is configured to: calculate an error signal variance estimate and an adaptive weight vector power estimate according to the error square median and the adaptive weight vector, respectively; calculate an input noise variance estimate according to the error signal variance estimate and the adaptive weight vector power estimate;

The second construction module 500 is configured to: construct an augmented weight vector based on the adaptive weight vector; calculate a deviation compensation term based on the augmented weight vector, the adaptive weight vector and the input noise variance estimate; and update the adaptive weight vector based on the deviation compensation term, the noisy input signal vector and the estimated error signal.

The effects of the above-mentioned system when applying the above-mentioned method can be found in the description of the above-mentioned method embodiment, which will not be repeated here.

Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

The present application is described with reference to the flowcharts and/or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and/or box in the flowchart and/or block diagram, as well as the combination of the processes and/or boxes in the flowchart and/or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate a device for implementing the functions specified in one process or multiple processes in the flowchart and/or one box or multiple boxes in the block diagram.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device that implements the functions specified in one or more processes in the flowchart and/or one or more boxes in the block diagram.

These computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more processes in the flowchart and/or one or more boxes in the block diagram.

Obviously, the above embodiments are merely examples for clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from these are still within the protection scope of the invention.

Claims (10)

1. A deviation compensation echo cancellation method for telephone communication, characterized in that the method comprises: Obtaining adaptive weights of an echo canceller and sampling values of a noisy speech signal; Constructing an adaptive weight vector and a noisy input signal vector according to the adaptive weight and the noisy speech signal sampling value respectively; Performing an inner product of the adaptive weight vector and the noisy input signal vector to generate an output signal; generating an estimated error signal according to the output signal, and generating a median square error according to the estimated error signal; Calculating an error signal variance estimate and an adaptive weight vector power estimate according to the error square median and the adaptive weight vector respectively; Calculate an input noise variance estimate based on the error signal variance estimate and the adaptive weight vector power estimate; Constructing an augmented weight vector according to the adaptive weight vector; Calculating a deviation compensation term based on the augmented weight vector, the adaptive weight vector and the input noise variance estimate; The adaptive weight vector is updated according to the deviation compensation term, the noisy input signal vector and the estimated error signal. 2. The method for eliminating the deviation-compensated echo for telephone communication according to claim 1, characterized in that: The adaptive weight vector is expressed as: ; in, is M adaptive weights, the subscript n represents the time, the superscript /> Represents the transpose operation; The noisy input signal vector is expressed as: ; in, are M sampling values of the noisy speech signal. 3. The method for eliminating the deviation-compensated echo in telephone communication according to claim 2, characterized in that: The output signal is expressed as: ; in, is the noisy input signal vector, /> is the adaptive weight vector, superscript/> Represents a transpose operation. 4. The method for eliminating the deviation-compensated echo for telephone communication according to claim 3, characterized in that: The estimated error signal is expressed as: ; in, represents the noisy echo signal at time n, that is, the noisy expected signal. 5. The method for eliminating the deviation-compensated echo for telephone communication according to claim 4, characterized in that: The median square error is expressed as: ; in, An integer representing the length of the median filter. 6. A deviation compensation echo cancellation method for telephone communication according to claim 5, characterized in that: The error signal variance estimate is expressed as: ; in, is the smoothing factor, ranging from 0.99 to 0.999, used for smoothing estimation,/> is the estimated value of the error signal variance at time n-1; The adaptive weight vector power estimate is expressed as: ; in, is the smoothing factor, ranging from 0.99 to 0.999, used for smoothing estimation, and M is the length of the adaptive weight vector; The input noise variance estimate is expressed as: ; in, is the ratio of output noise to input noise variance,/> is the variance of the measurement noise without impulse noise, is the variance of the input noise, and M is the length of the adaptive weight vector. 7. A deviation compensation echo cancellation method for telephone communication according to claim 6, characterized in that: The augmented weight vector is expressed as: ; in, is the noise ratio, superscript/> Represents a transpose operation. 8. The method for deviation compensation echo cancellation for telephone communication according to claim 7, characterized in that: The deviation compensation term is expressed as: ; in, is a positive kernel width parameter, /> To find the 2-norm. 9. A deviation-compensated echo cancellation method for telephone communication according to claim 8, characterized in that: The formula for updating the adaptive weight vector is: ; in, is a positive step parameter. 10. A deviation-compensated echo cancellation system for telephone communication, characterized in that the system comprises: a data acquisition module, a first construction module, a first calculation module, a second calculation module and a second construction module; The data acquisition module is configured to: acquire the adaptive weight of the echo canceller and the sampling value of the noisy speech signal; The first construction module is configured to: construct an adaptive weight vector and a noisy input signal vector according to the adaptive weight and the noisy speech signal sampling value respectively; The first calculation module is configured to: perform an inner product of the adaptive weight vector and the noisy input signal vector to generate an output signal; generate an estimated error signal according to the output signal, and generate a square median error according to the estimated error signal; The second calculation module is configured to: calculate an error signal variance estimate and an adaptive weight vector power estimate according to the error square median and the adaptive weight vector respectively; calculate an input noise variance estimate according to the error signal variance estimate and the adaptive weight vector power estimate; The second construction module is configured to: construct an augmented weight vector based on the adaptive weight vector; calculate a deviation compensation term based on the augmented weight vector, the adaptive weight vector and the input noise variance estimate; and update the adaptive weight vector based on the deviation compensation term, the noisy input signal vector and the estimated error signal.