WO2017012350A1 - Filter state divergence judgement method and device - Google Patents

Abstract

A filter state divergence judgement method and device, which are used for solving the problems in the prior art that a filter state judgement method has a relatively high calculation complexity and has a possibility of misjudgement. The method comprises: acquiring a near-end signal and a far-end signal synchronized with the near-end signal (300), and respectively performing synchronous framing processing on the near-end signal and the far-end signal, so as to obtain N pairs of synchronous frames (310); filtering the N pairs of synchronous frames via a filter, so as to obtain a mean value of N filter coefficient vectors corresponding to the N pairs of synchronous frames (320); respectively calculating N filter coefficient state reference quantities corresponding to the N pairs of synchronous frames according to the mean value of the N filter coefficient vectors (330); and when at least M filter coefficient state reference quantities greater than a pre-set threshold value exist in the N filter coefficient state reference quantities, determining that the filter is in a divergent state (340). Therefore, whether a filter is in a divergent state can be effectively determined, and update of a filter coefficient vector can be controlled.

Description

Method and device for judging filter state divergence

The present application claims priority to Chinese Patent Application No. 201510433689.1, entitled "A Method and Apparatus for Determining Filter State Dispersion", filed on July 21, 2015, the entire contents of which are incorporated by reference. In this application.

Technical field

The present invention relates to the field of communications, and in particular, to a method and apparatus for determining a state of a filter divergence.

Background technique

In the communication system, the call quality is affected by the echo. If the echo cancellation is not performed during the call, especially under the hands-free condition, the normal call cannot be achieved. The so-called echo refers to the situation that the voice signal from the near-end speaker is directly or indirectly collected by the microphone and transmitted back to the far end, that is, the party in the call hears its own voice from the phone used by itself.

In the prior art, echo is typically cancelled using adaptive filter techniques. The adaptive filter uses the far-end signal to simulate the path generated by the echo and generates an estimated echo signal, which is then subtracted from the near-end signal and sent to the far end.

Specifically, FIG. 1 is a schematic diagram of the working principle of the echo canceller in the background art of the present invention. One straight line and two curves in the right solid line block diagram in FIG. 1 are used to simulate the sound of the far end from the speaker. And the process of propagation collected by the Mike. Among them, the curve indicates that the sound of the far end is received by the microphone through reflection. The dotted line frame in Fig. 1 is a typical acoustic echo canceller. Assuming that user A and user B are talking, the near-end signal d(n) collected by the microphone can be understood as the voice of user A's own voice and user B's. Echo, if user A does not speak, only the echo of user B, and the far-end signal x(n) transmitted by the remote end to synchronize with the near-end signal can be understood as the voice of the user B who is talking with user A, Input two signals into the echo canceller, according to the original adaptive filter The filter coefficient vector can be calculated, and the echo propagation path in the right solid line block diagram in FIG. 1 is simulated according to the filter coefficient vector, and the output signal e(n) processed by the echo canceller is calculated. Further, the output signal e(n) is transmitted back to the remote end, that is, the user B, and the entire echo cancellation process is completed to ensure the quality of the call between the users.

As can be seen from the above, referring to FIG. 2, FIG. 2 is a schematic diagram of the working principle of the adaptive filter in the background art of the present invention, wherein the input signal of the adaptive filter is the near-end signal d(n) and is synchronized with the near-end signal. The far-end signal x(n) is obtained by adaptive filter echo cancellation processing and the output signal is e(n), and the output signal is sent to the far end. Further, whether the filter state diverges or not determines the quality of the output signal e(n). If the filter state is diverged, it indicates that the output signal is distorted, and the echo canceller fails, which will seriously affect the communication quality.

Therefore, by detecting whether the filter state is divergent, it can be judged whether the output signal is distorted. If the filter state is diverged, the echo canceller will definitely be affected or even invalid, resulting in distortion of the output signal. If the filter state converges, the echo is large. Partially eliminated, the call quality is better. Therefore, the detection of the filter state is an important component of the echo cancellation system and even the communication system, and is an important criterion for judging the quality of the output signal and the quality of the communication.

In the prior art, a commonly used method for detecting whether a filter state is divergent is to calculate an Echo Return Loss Enhancement (ERLE), that is, a difference between an echo signal energy and a residual echo signal energy, and the method is used for evaluating an echo canceller. For the cancellation performance of the echo, the specific formula is as follows:

ERLE(n)=10*log ₁₀ E[d ² (n)]/E[e ² (n)]

In the above formula, E[d ² (n)] represents the near-end signal energy collected by the microphone, the near-end signal includes echo and other signals; E[e ² (n)] represents the output signal energy of the echo canceller, and the output signal Includes other signals and residual echo as described previously. The other signals mentioned here are the voice signals spoken by the user in the actual call, and the ambient noise at the near end. Therefore, the other signals are not affected by default during the processing.

When the filter state converges, ERLE is relatively stable. When the adaptive filter is disturbed and the filter state is diverged, ERLE will be significantly reduced. Specifically, the filter can be determined by setting the threshold. Whether the state of the device converges, when the ERLE is higher than the preset threshold, the filter state converges. When the ERLE is less than the preset threshold, the filter state is diverged.

The prior art is based on the energy of the signal to evaluate whether the filter state is divergent. Calculating the ERLE needs to calculate the energy difference between the near-end signal and the output signal, but since the speech signal itself fluctuates greatly, the signal energy fluctuation caused by the signal is also relatively large. In the calculation of ERLE, in order to smooth the volatility of the signal, additional operations are needed to smooth the signal, and in order to control the dynamic range of the ERLE, the calculation needs to be converted into a logarithmic domain. Therefore, the complexity of the whole calculation process is more complicated. high. In addition, if the environment of the near-end signal changes, such as noise changes, it will also affect the stability of ERLE, which may cause errors in the judgment of whether the filter state is divergent.

Summary of the invention

The embodiment of the invention provides a method and a device for determining the state of the filter to be diverged, which is used to solve the problem that the filter state judging method existing in the prior art has high computational complexity and may cause a judgment error.

The specific technical solutions provided by the embodiments of the present invention are as follows:

In a first aspect, an embodiment of the present invention provides a method for determining a state of a filter divergence, including:

Collecting a near-end signal and a far-end signal synchronized with the near-end signal, and

Performing synchronous framing processing on the near-end signal and the far-end signal respectively to obtain N-pair sync frames, each pair of sync frames including a near-end signal frame and a far-end signal synchronized with the near-end signal frame frame;

And filtering the N pairs of synchronization frames by using a filter to obtain N filter coefficient vector mean values corresponding to the N pairs of synchronization frames;

Calculating, according to the N filter coefficient vector mean values, N filter coefficient state reference quantities corresponding to the N pairs of synchronization frames;

Determining that the filter state is divergent when at least M filter coefficient state reference quantities are greater than a preset threshold value in the N filter coefficient state reference quantities;

Where M is a preset value, and both N and M are positive integers, and N≥M.

With reference to the first aspect, in a first possible implementation manner of the first aspect, the performing the synchronous framing processing on the near-end signal and the far-end signal respectively to obtain the N-pair synchronization frame includes:

According to the preset frame length, the near-end signals are sequenced according to the signal timing sequence to obtain N near-end signal frames;

At the same time, according to the preset frame length, the far-end signals synchronized with the near-end signals are sequentially processed according to the signal timing sequence to obtain N far-end signal frames synchronized with the near-end signal frame;

Any one of the obtained near-end signal frame and the far-end signal frame synchronized with the any one of the near-end signal frames constitutes a pair of synchronization frames, and a total of N pairs of synchronization frames are obtained.

With reference to the first aspect, in a second possible implementation manner of the first aspect, the N pairs of synchronization frames are filtered by a filter to obtain N filter coefficient vector mean values corresponding to the N pairs of synchronization frames, include:

For each pair of sync frames, execute separately:

Determining, according to the preset frame length and the first preset algorithm, a filter coefficient vector update number T corresponding to the current pair of synchronization frames, and T update results corresponding to the T-th filter coefficient vector update, where T is a positive integer;

And obtaining, according to the filter coefficient vector update order T corresponding to the current pair of synchronization frames and the T update result of the T-th filter coefficient vector, obtaining a filter coefficient vector mean corresponding to the current pair of synchronization frames.

With reference to the first aspect, or any one of the possible implementation manners of the foregoing first aspect, in a third possible implementation manner, the calculating N times of the corresponding N pairs of synchronization frames according to the N filter coefficient vector mean values Filter coefficient status reference, including:

For each filter coefficient vector mean, perform:

Determining, according to a preset layout rule, a filter coefficient distribution map corresponding to the current filter coefficient vector mean according to a current filter coefficient vector mean value, and

According to the filter coefficient distribution map, the elements in the current filter coefficient vector mean are at least divided into a main reflection area element, a left side element of the main reflection area, and a right side element of the main reflection area, wherein the main reflection area The element refers to all the elements contained in the preset sliding window in the filter coefficient distribution map. The element included in the preset sliding window when the sum of the squares is the maximum value, and the element on the left side of the preset sliding window in the filter coefficient distribution map is the left element of the main reflection area, the pre- The element on the right side of the sliding window is an element on the right side of the main reflection area;

Calculating a sum of squares of elements on the right side of all the main reflection areas as a first energy value corresponding to the mean value of the current filter coefficient vector;

Calculating the sum of the squares of all the main reflection area elements, the elements on the left side of all the main reflection areas, and the elements on the right side of all the main reflection areas, as the second energy value corresponding to the mean value of the current filter coefficient vector, and

Calculating a smoothed value of the second energy value according to a second preset algorithm;

And determining, according to the first energy value and the smoothed value of the second energy value, a filter coefficient state reference quantity corresponding to the current filter coefficient vector mean value by using a third preset algorithm.

With reference to the first aspect, or any one of the possible implementation manners of the foregoing first aspect, in a fourth possible implementation, the calculating N corresponding to N pairs of synchronization frames according to N filter coefficient vector averages Filter coefficient status reference, including:

For each filter coefficient vector mean, perform:

Determining, according to a preset layout rule, a filter coefficient distribution map corresponding to the current filter coefficient vector mean according to a current filter coefficient vector mean value, and

According to the filter coefficient distribution map, the elements in the current filter coefficient vector mean are at least divided into a main reflection area element, a left side element of the main reflection area, and a right side element of the main reflection area, wherein the main reflection area An element is an element included in the preset sliding window when a sum of squares of all elements included in a preset sliding window in the filter coefficient distribution map is a maximum value, and the pre-predetermined in the filter coefficient distribution map The element on the left side of the sliding window is an element on the left side of the main reflection area, and the element on the right side of the preset sliding window is an element on the right side of the main reflection area;

Calculating the sum of the squares of all the main reflection area elements, the elements on the left side of all the main reflection areas, and the elements on the right side of all the main reflection areas, as the second energy value corresponding to the mean value of the current filter coefficient vector, and

Calculating a smoothed value of the second energy value according to the second preset algorithm;

Calculating the sum of the squares of the elements on the left side of all the main reflection areas and the elements on the right side of all the main reflection areas as the third energy value corresponding to the mean value of the current filter coefficient vector;

And determining, by the third preset algorithm, a filter coefficient state reference quantity corresponding to the current filter coefficient vector mean according to the third energy value and the smoothed value of the second energy value.

With reference to the first aspect, or any one of the possible implementation manners of the foregoing first aspect, in a fifth possible implementation, the element in the current filter coefficient vector mean is determined according to the filter coefficient profile At least the main reflection zone element, the left element of the main reflection zone, and the right element of the main reflection zone, including:

Determining, according to the filter coefficient distribution map, an element having the largest absolute value among the current filter coefficient vector mean values;

Constructing a sliding window, determining a window height of the sliding window according to an absolute value of the element having the largest absolute value, and using a preset value as a window length of the sliding window;

Sliding the sliding window from the first element of the current filter coefficient vector mean to the last element on the filter coefficient profile to determine that the sum of squares of all elements included in the sliding window is maximum When the value is, the element included in the sliding window is the main reflection area element, the element on the left side of the sliding window is an element on the left side of the main reflection area, and the element on the right side of the sliding window is the main reflection The right side of the area.

With reference to the first aspect, or any one of the foregoing possible implementation manners, in a sixth possible implementation, after determining that the filter state is diverged, the method further includes:

Stop updating the filter coefficient vector and clear the current filter coefficient vector.

In a second aspect, an embodiment of the present invention provides an apparatus for determining a state of a filter, including:

An acquisition module, configured to collect a near-end signal and a far-end signal synchronized with the near-end signal;

a framing module, configured to perform synchronous framing processing on the near-end signal and the far-end signal collected by the collecting module, respectively, to obtain N pairs of synchronization frames, each pair of synchronization frames including a near-end signal frame and a a far-end signal frame synchronized with the near-end signal frame;

a filtering module, configured to filter, by using a filter, the N pairs of synchronization frames obtained by the framing module to obtain N filter coefficient vector mean values corresponding to the N pairs of synchronization frames;

a determining module, configured to calculate, according to the N filter coefficient vector average values obtained by the filtering module, N filter coefficient state reference quantities corresponding to the N pairs of synchronization frames;

And determining, when the at least M filter coefficient state reference quantities are greater than a preset threshold value, in the N filter coefficient state reference quantities, determining the filter state divergence;

Where M is a preset value, and both N and M are positive integers, and N≥M.

With reference to the second aspect, in a first possible implementation manner of the second aspect, the framing module is specifically configured to:

According to the preset frame length, the near-end signals collected by the collection module are sequentially processed according to the signal timing sequence to obtain N near-end signal frames;

Simultaneously, according to the preset frame length, the far-end signals that are collected by the acquiring module and synchronized with the near-end signal are subjected to frame processing in sequence according to signal timing, to obtain N synchronization with the near-end signal frame. Remote signal frame;

Any one of the obtained near-end signal frame and the far-end signal frame synchronized with the any one of the near-end signal frames constitutes a pair of synchronization frames, and a total of N pairs of synchronization frames are obtained.

With reference to the second aspect, in a second possible implementation manner of the second aspect, the filtering module is specifically configured to:

For each pair of synchronization frames obtained by the framing module, respectively:

Determining, according to the preset frame length and the first preset algorithm, a filter coefficient vector update number T corresponding to the current pair of synchronization frames, and T update results corresponding to the T-th filter coefficient vector update, where T is a positive integer;

And obtaining, according to the filter coefficient vector update order T corresponding to the current pair of synchronization frames and the T update result of the T-th filter coefficient vector, obtaining a filter coefficient vector mean corresponding to the current pair of synchronization frames.

With reference to the second aspect, or any one of the possible implementation manners of the foregoing second aspect, in a third possible implementation manner, the determining module is specifically configured to:

For each filter coefficient vector mean obtained by the filtering module, respectively:

Determining, according to a preset layout rule, a filter coefficient distribution map corresponding to the current filter coefficient vector mean according to a current filter coefficient vector mean value, and

Entity of the current filter coefficient vector mean according to the filter coefficient profile At least the main reflection area element, the left side element of the main reflection area, and the right side element of the main reflection area, wherein the main reflection area element refers to all elements included in the preset sliding window in the filter coefficient distribution map. An element included in the preset sliding window when the sum of squares is a maximum value, and an element on the left side of the preset sliding window in the filter coefficient distribution map is an element on a left side of the main reflection area, the preset The element on the right side of the sliding window is the element on the right side of the main reflection area;

Calculating a sum of squares of elements on the right side of all the main reflection areas as a first energy value corresponding to the mean value of the current filter coefficient vector;

Calculating the sum of the squares of all the main reflection area elements, the elements on the left side of all the main reflection areas, and the elements on the right side of all the main reflection areas, as the second energy value corresponding to the mean value of the current filter coefficient vector, and

Calculating a smoothed value of the second energy value according to a second preset algorithm;

And determining, according to the first energy value and the smoothed value of the second energy value, a filter coefficient state reference quantity corresponding to the current filter coefficient vector mean value by using a third preset algorithm.

With reference to the second aspect, or any one of the possible implementation manners of the foregoing second aspect, in a fourth possible implementation, the determining module is further configured to:

For each filter coefficient vector mean obtained by the filtering module, respectively:

Determining, according to a preset layout rule, a filter coefficient distribution map corresponding to the current filter coefficient vector mean according to a current filter coefficient vector mean value, and

According to the filter coefficient distribution map, the elements in the current filter coefficient vector mean are at least divided into a main reflection area element, a left side element of the main reflection area, and a right side element of the main reflection area, wherein the main reflection area An element is an element included in the preset sliding window when a sum of squares of all elements included in a preset sliding window in the filter coefficient distribution map is a maximum value, and the pre-predetermined in the filter coefficient distribution map The element on the left side of the sliding window is an element on the left side of the main reflection area, and the element on the right side of the preset sliding window is an element on the right side of the main reflection area;

Calculating the sum of the squares of all the main reflection area elements, the elements on the left side of all the main reflection areas, and the elements on the right side of all the main reflection areas, as the second energy value corresponding to the mean value of the current filter coefficient vector, and

Calculating a smoothed value of the second energy value according to the second preset algorithm;

Calculate the sum of the squares of the elements on the left side of all the main reflection areas and the elements on the right side of all the main reflection areas. Determining a third energy value corresponding to a current filter coefficient vector mean;

And determining, by the third preset algorithm, a filter coefficient state reference quantity corresponding to the current filter coefficient vector mean according to the third energy value and the smoothed value of the second energy value.

With reference to the second aspect, or any one of the possible implementation manners of the foregoing second aspect, in a fifth possible implementation, the determining module is further configured to:

Determining, according to the filter coefficient distribution map, an element having the largest absolute value among the current filter coefficient vector mean values;

Constructing a sliding window, determining a window height of the sliding window according to an absolute value of the element having the largest absolute value, and using a preset value as a window length of the sliding window;

Sliding the sliding window from the first element of the current filter coefficient vector mean to the last element on the filter coefficient profile to determine that the sum of squares of all elements included in the sliding window is maximum When the value is, the element included in the sliding window is the main reflection area element, the element on the left side of the sliding window is an element on the left side of the main reflection area, and the element on the right side of the sliding window is the main reflection The right side of the area.

With reference to the second aspect, or any one of the possible implementation manners of the foregoing second aspect, in a sixth possible implementation, the device further includes:

And a resetting module, configured to stop updating the filter coefficient vector after the determining, according to the determining module, that the filter state is diverged, and clear the current filter coefficient vector.

The beneficial effects of the present invention are as follows:

In the embodiment of the present invention, a method for judging the state of the filter is proposed, which is: collecting a near-end signal and a far-end signal synchronized with the near-end signal, and performing synchronous framing on the near-end signal and the far-end signal, respectively. Processing, obtaining N pairs of synchronization frames, filtering the N pairs of synchronization frames through a filter, and obtaining N filter coefficient vector mean values corresponding to the N pairs of synchronization frames. Further, according to the N filter coefficient vector mean values, N filter coefficient state reference quantities corresponding to the N pairs of synchronization frames are respectively calculated, since the filter coefficient state reference quantity is insensitive to changes in signal energy, and is not affected by the near-end signal and The influence of the environment where the far-end signal is located, so the obtained result of the method is stable, and the calculation complexity of the filter coefficient state reference quantity is low, and no additional operation is needed to smooth the fluctuation of the signal. At last, When at least M filter coefficient state reference quantities are greater than a preset threshold value among the N filter coefficient state reference quantities, it is determined that the filter state diverges. Therefore, the embodiment of the present invention uses the filter coefficient state reference quantity to determine whether the filter state is divergent, so as to achieve the purpose of controlling the echo canceler filter coefficient vector update and guaranteeing the call quality, and determining whether the filter state is diverged by the method of the present invention. It can effectively judge whether there is double talk during the call, so as to achieve the purpose of controlling the filter coefficient vector update and obtaining high quality output signal.

DRAWINGS

1 is a schematic view showing the working principle of an echo canceller in the background art of the present invention;

2 is a schematic diagram showing the working principle of an adaptive filter in the background art of the present invention;

3 is a flowchart showing an overview of determining a state of a filter divergence according to an embodiment of the present invention;

4A and 4B are diagrams showing a filter coefficient distribution diagram according to an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of an apparatus for determining a state of a filter divergence according to an embodiment of the present invention; FIG.

FIG. 6 is a schematic structural diagram of an apparatus for determining a state of a filter divergence according to an embodiment of the present invention.

detailed description

The present invention will be further described in detail with reference to the accompanying drawings, in which FIG. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

The embodiment of the invention provides a method and a device for determining the state of the filter divergence, which are used to solve the problem that the filter state judging method existing in the prior art has high computational complexity and may cause a judgment error. The method and the device are based on the same inventive concept. Since the principles of the method and the device for solving the problem are similar, the implementation of the device and the method can be referred to each other, and the repeated description is not repeated.

In the communication system, the call quality is affected by the echo. If the echo cancellation is not performed during the call, especially under the hands-free condition, the normal call cannot be achieved. In the prior art, it is usually used The adaptive filter technique eliminates the echo, and the detection of the filter state is an important criterion for judging the echo cancellation effect. If the filter state is divergent, the echo cancellation effect is poor or even invalid, resulting in distortion of the output signal. If the filter state converges, the echo is mostly eliminated, and the call quality is good.

Preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

Referring to FIG. 3, FIG. 3 is a flowchart of an overview of determining a state of divergence of a filter according to an embodiment of the present invention. The specific process of determining a state of divergence of a filter in the present invention is as follows:

Step 300: Acquire a near-end signal and a far-end signal synchronized with the near-end signal.

The echo canceller combines the near-end signal collected by the microphone with the far-end signal synchronized with the near-end signal, which together serve as the acquired input signal. Among them, the near-end signal includes echo and other signals, and other signals mentioned here are voice signals spoken by the user during actual conversation, and environmental noise at the near end. The far-end signal is synchronized with the near-end signal, and the far-end signal is a voice signal from the near-end speaker. Specifically, synchronization refers to the fact that two signals maintain a certain relative relationship during the change process, and the phenomenon of consistency and unification occurs in time.

Step 310: Perform synchronous framing processing on the near-end signal and the far-end signal respectively to obtain N-pair synchronization frames, each pair of synchronization frames including a near-end signal frame and a far-end signal frame synchronized with the near-end signal frame.

Specifically, according to the preset frame length, the near-end signals are sequentially processed according to the signal timing sequence to obtain N near-end signal frames. At the same time, according to the same preset frame length, the far-end signals synchronized with the near-end signals are subjected to frame processing in sequence according to the signal timing, and N far-end signal frames synchronized with the near-end signal frames are obtained. Further, any one of the obtained near-end signal frames and the far-end signal frame synchronized with the any one of the near-end signal frames constitute a pair of synchronization frames, and a total of N pairs of synchronization frames are obtained.

For example, the collected near-end signal is composed of 1000 sampling points, the preset frame length is 100 sampling points, and the near-end signals are subjected to frame processing in sequence according to the signal timing, thereby obtaining 10 near-end signal frames, and at the same time, The collected far-end signal synchronized with the near-end signal is composed of 1000 sampling points. According to the same preset frame length of 100 sampling points, the far-end signal synchronized with the near-end signal is subjected to frame processing according to the signal timing sequence. Get 10 far-end signal frames synchronized with the near-end signal frame. Further Alternatively, any one of the obtained near-end signal frames and the far-end signal frame synchronized with the any one of the near-end signal frames constitute a pair of synchronization frames, and a total of 10 pairs of synchronization frames are obtained.

Step 320: Filtering the N pairs of synchronization frames by a filter to obtain N filter coefficient vector mean values corresponding to the N pairs of synchronization frames.

Specifically, for each pair of synchronization frames, respectively:

Determining, according to the preset frame length and the first preset algorithm, a filter coefficient vector update number T corresponding to the current pair of synchronization frames, and T update results corresponding to the T-th filter coefficient vector update, where T is a positive integer;

Further, the filter coefficient vector mean value corresponding to the current pair of synchronization frames is obtained according to the filter coefficient vector update order T and the T-th filter coefficient vector corresponding to the current pair of synchronization frames to update the corresponding T update results.

Here, an adaptive filter technique based on an affine projection algorithm is taken as an example, but is not limited to the adaptive filtering technique.

The following uses an affine projection algorithm to explain how to update the filter coefficient vector and how to obtain the filter coefficient vector update result.

Firstly, according to the adaptive filtering technique of the affine projection algorithm, it is set to collect one sample point at each time, and then construct a signal vector from multiple samples, assuming that at the nth time, according to the number of elements in the filter coefficient vector L, constructing the far-end signal input vector from the far-end signal x(n):

x(nl)=[x(nl),...,x(nL-l+1)] ^T .

Further, the far-end signal input matrix is constructed by the far-end signal input vector according to the preset affine projection order p:

X(n)=[x(n),...,x(n-p+1)].

The resulting input matrix of the far-end signal can be expressed as:

It should be noted that in most cases, p is much smaller than L.

Then, according to the preset affine projection order p, the near-end signal input vector is constructed from the near-end signal d(n):

d(n)=[d(n),...,d(n-p+1)] ^T .

The filter coefficient vector is h(n)=[h _n (0),...,h _n (L-1)] ^T ,

The output vector corresponding to the output signal is denoted as e(n)=[e(n),...,e(n-p+1)] ^T ,

Where e(n)=d(n)-X ^T (n)h(n).

A sample point is acquired at each moment, and the filter coefficient vector is updated once at each sampling point to obtain the filter coefficient vector updated at the n+1th time:

h(n+1)=h(n)+μX(n)[X ^T (n)X(n)] ^-1 e(n).

Where μ is the update step size of the filter, which is the preset value, [*] ^-1 represents the matrix inversion operation, and [*] ^T represents the matrix transposition operation.

In general, filter coefficient vector updates are not performed for each sample point during signal processing. Instead, the acquired near-end signal and the far-end signal synchronized with the near-end signal are synchronously framed.

The above process clearly shows how to perform filter coefficient vector update and how to obtain the filter coefficient vector update result. Further, for the i-th sync frame, the filter coefficients of the i-th sync frame can be obtained by using, but not limited to, the following method. Vector mean:

Taking the preset frame length of 160 sample points as an example, synchronous frame processing is performed on the near-end signal and the far-end signal respectively, and N-pair sync frames are obtained, and the order of the affine projection algorithm is assumed to be p=4, therefore, i For the synchronization frame, the filter coefficient vector is updated every 4 sample points, the filter coefficient vector will be updated 40 times, and each filter coefficient vector h _k (i) is obtained as the update result of the filter coefficient vector. Where i is the frame number, then, for the ith pair of sync frames, the filter coefficient vector will be updated 40 times, resulting in 40 different h _k (i), ie k=0,1,2,...,39 , the filter coefficient vector update number is K, and the filter coefficient vector mean of the i-th sync frame is obtained:

Where k denotes an index of the number of times the filter coefficient vector h(i) is updated.

Step 330: Calculate N filter coefficient state reference quantities corresponding to the N pairs of synchronization frames according to the N filter coefficient vector mean values.

Here, N filter coefficient state reference quantities corresponding to N pairs of sync frames can be obtained by the following two methods, but not limited to the following two methods.

the first method:

For each filter coefficient vector mean, perform:

First, according to the current filter coefficient vector mean value, the filter coefficient distribution map corresponding to the current filter coefficient vector mean value is determined according to a preset layout rule.

According to the filter coefficient distribution map, the elements in the current filter coefficient vector mean are at least divided into the main reflection area element, the left side element of the main reflection area, and the right side element of the main reflection area, as shown in FIG. 4A.

4A is a diagram showing a distribution of filter coefficients in an embodiment of the present invention. Wherein, the number of elements in the filter coefficient vector is set by a specific adaptive filter algorithm, and the elements in the filter vector mean are sequentially extracted, and are sequentially arranged from left to right, and each two elements are in the horizontal direction. The spacing of the projections is equal, the window length of the sliding window can be determined by the empirical value, and the number of elements contained in the sliding window is a fixed value, and the point A is the element with the largest absolute value in the filter coefficient vector, and the absolute value is the largest element. The absolute value is represented by h, and the window height is at least twice the maximum absolute value of 2h.

The main reflection area element refers to an element included in the preset sliding window when the sum of the squares of all the elements included in the preset sliding window in the filter coefficient distribution map is the maximum value, and the preset sliding window in the filter coefficient distribution map The element on the left is the element on the left side of the main reflection area, and the element on the right side of the preset sliding window is the main reflection. The right side of the shot.

Then, the sum of the squares of the elements on the right side of all the main reflection areas is calculated as the first energy value corresponding to the current filter coefficient vector mean.

Calculate the sum of the squares of all the main reflection area elements, the elements on the left side of all the main reflection areas, and the elements on the right side of all the main reflection areas, as the second energy value corresponding to the current filter coefficient vector mean, and calculate the second according to the second preset algorithm The smoothed value of the energy value.

Finally, according to the first energy value and the smoothed value of the second energy value, the third preset algorithm is used to determine the filter coefficient state reference amount corresponding to the current filter coefficient vector mean.

The second method:

For each filter coefficient vector mean, perform:

First, according to the current filter coefficient vector mean value, the filter coefficient distribution map corresponding to the current filter coefficient vector mean value is determined according to a preset layout rule.

According to the filter coefficient distribution map, the elements in the current filter coefficient vector mean are at least divided into the main reflection area element, the left side element of the main reflection area, and the right side element of the main reflection area, as shown in FIG. 4A.

The main reflection area element refers to an element included in the preset sliding window when the sum of the squares of all the elements included in the preset sliding window in the filter coefficient distribution map is the maximum value, and the preset sliding window in the filter coefficient distribution map The element on the left is the element on the left side of the main reflection area, and the element on the right side of the preset sliding window is the element on the right side of the main reflection area.

Then, calculating the sum of the squares of all the main reflection area elements, the elements on the left side of all the main reflection areas, and the elements on the right side of all the main reflection areas, as the second energy value corresponding to the current filter coefficient vector mean, and calculating according to the second preset algorithm The smoothed value of the second energy value.

Calculate the sum of the squares of the elements on the left side of all the main reflection areas and the elements on the right side of all the main reflection areas as the third energy value corresponding to the current filter coefficient vector mean.

Finally, according to the third energy value and the smoothed value of the second energy value, the third preset algorithm is used to determine the filter coefficient state reference amount corresponding to the current filter coefficient vector mean.

Wherein, the above two methods need to divide at least the elements in the current filter coefficient vector mean into the main reflection area element, the left side element of the main reflection area, and the main reflection area right according to the filter coefficient distribution map. Side element.

Specifically, according to the filter coefficient distribution map, an element having the largest absolute value among the current filter coefficient vector mean values is determined. Construct a sliding window to determine the window height of the sliding window based on the absolute value of the element with the largest absolute value, with the preset value being the window length of the sliding window.

Sliding the sliding window on the filter coefficient profile from the first element of the current filter coefficient vector mean to the last element, determining that the sum of the squares of all the elements contained in the sliding window is the maximum value, and the sliding window contains The element is the main reflection zone element, the element on the left side of the sliding window is the left element of the main reflection zone, and the element on the right side of the sliding window is the element on the right side of the main reflection zone.

The process of obtaining the filter coefficient state reference amount described above will be described below in a specific embodiment.

Taking the i-th filter coefficient vector mean value as an example, first, according to the filter coefficient vector mean value, the filter coefficient distribution map corresponding to the filter coefficient vector mean value is determined, as shown in FIG. 4A, the elements in the filter coefficient vector. The number is set by a specific adaptive filter algorithm, and the elements in the filter vector are sequentially taken out and arranged from left to right. The projection of each element in the horizontal direction is equal, and all the elements are connected. Get the filter coefficient distribution map.

Determining, according to the filter coefficient distribution map, an element having the largest absolute value among the i-th filter coefficient vector mean values, and determining a window height of the sliding window according to an absolute value of the element having the largest absolute value, for example, in FIG. 4A, the absolute value is the largest The element is A point, the absolute value corresponding to point A is h, and the window height is 2h. The preset window is the window length of the sliding window, and the window length is smaller than the filter in the filter coefficient distribution map. The length of the coefficient distribution can be arbitrarily set, but it is generally small, and can also be determined according to empirical values, such as one tenth of the length of the filter coefficient distribution, and the number of elements included in the sliding window is a fixed value. 4A is a simplified description of the filter coefficient distribution diagram of the present invention. The waveform drawn in FIG. 4A is relatively simple, and the number of elements included in the filter coefficient vector mean is small. In the actual application process, the filter coefficient distribution map As shown in Figure 4B.

Sliding the sliding window from the first element of the i-th filter coefficient vector mean to the last element on the filter coefficient profile, and determining that the sum of the squares of all the elements included in the sliding window is the maximum value, within the sliding window The included element is the main reflection area element, the element on the left side of the sliding window is the left element of the main reflection area, and the element on the right side of the sliding window is the right element of the main reflection area.

As shown in FIG. 4A, the elements contained in the sliding window are the main reflection area elements, and the elements on the left side of the sliding window are the left side elements of the main reflection area. The absolute value of the main reflection area element is large, the main reflection area occupies most of the energy of the filter coefficient vector, the element on the left side of the sliding window is the left element of the main reflection area, and the absolute value of the element on the left side of the main reflection area is small. The energy of the left side of the main reflection area is small, the element on the right side of the sliding window is the element on the right side of the main reflection area, the absolute value of the element on the right side of the main reflection area is larger than the absolute value of the element on the left side of the main reflection area, and the right side of the main reflection area Greater energy.

Further, the sum of the squares of the elements on the right side of all the main reflection areas is calculated as the first energy value corresponding to the mean value of the i-th filter coefficient vector, as follows:

Calculate the sum of the squares of all the main reflection area elements, the elements on the left side of all the main reflection areas, and the elements on the right side of all the main reflection areas, as the second energy value corresponding to the i-th filter coefficient vector mean, and calculate according to the following formula (2) The smoothed value of the second energy value, the smoothed values of the second energy value and the second energy value are as follows:

E _smooth (i)=α·E _smooth (i-1)+(1-α)·E _all (2)

Here, α is a preset value, for example, α=0.04.

Calculate the sum of the squares of the elements on the left side of all the main reflection areas and the elements on the right side of all the main reflection areas as the third energy value corresponding to the mean of the i-th filter coefficient vector, as follows:

The first method is used to calculate the filter coefficient state reference amount corresponding to the i-th filter coefficient vector mean according to the first energy value and the smoothed value of the second energy value:

In addition, for the first method, another method may be used to calculate a filter coefficient state reference quantity corresponding to the i-th filter coefficient vector mean value, specifically:

The second method is used to calculate the filter coefficient state reference amount corresponding to the i-th filter coefficient vector mean according to the first energy value and the smoothed value of the second energy value:

In addition, for the second method, another method may be used to calculate a filter coefficient state reference quantity corresponding to the i-th filter coefficient vector mean value, specifically:

Step 340: Determine that the filter state diverges when at least M filter coefficient state reference quantities are greater than a preset threshold value among the N filter coefficient state reference quantities.

Where M is a preset value, N and M are both positive integers, and N≥M.

For example, if N=M=1, the threshold value T _{R is set} , T _R =0.15. If R(1)>T _R , ST(1)=1 is determined to determine the filter state divergence; otherwise, ST(1) =0, determine the filter state convergence.

If it is determined that the filter state is divergent, the update of the filter coefficient vector is stopped, and the current filter coefficient vector is cleared.

However, it is sometimes determined that the filter state is diverged according to the comparison result between the filter coefficient state reference quantity corresponding to the pair of synchronization frames and the preset threshold value, which may cause a greater risk of resetting the filter coefficient vector, which is an embodiment of the present invention. The method also includes comparing the filter coefficient state reference quantity and the preset threshold value in a certain time window to determine whether the filter state is divergent.

For another example, the ratio of the filter coefficient state reference quantity corresponding to the preset threshold value corresponding to 10 pairs of synchronization frames is taken. More results:

ST(m)=[ST(m), ST(m-1), ST(m-2),...,ST(m-9)],

The ST(m) includes a comparison result between the filter coefficient state reference quantity corresponding to the 10 pairs of synchronization frames and the preset threshold value, and ST(i) represents the filter coefficient state reference quantity corresponding to the i-th synchronization frame and the preset. Comparison of threshold values.

By counting the number of elements st1 with a value of 1 in ST(m), a strategy Strategy(m) for determining the state of the filter divergence is set, that is, when there are 6 or more filter coefficient state reference quantities greater than a preset threshold value , to determine the filter state divergence, otherwise the filter state converges.

If Strategy(m)=1, there are 6 or more filter coefficient state reference quantities greater than a preset threshold value, determining filter state divergence, stopping filter coefficient vector update, and vice versa, Strategy(m)=0, Then, if the filter coefficient state reference quantity of 5 or less is greater than the preset threshold value, and the filter state is determined to be converged, the filter coefficient vector is normally updated.

Therefore, when it is determined that at least 6 filter coefficient state reference quantities are greater than a preset threshold in the 10 filter coefficient state reference quantities, the filter state is determined to be diverged, and the filter coefficient vector update is determined according to the determination.

The embodiment of the invention provides a method for determining the state of the filter divergence, which is: collecting a near-end signal and a far-end signal synchronized with the near-end signal, and performing synchronous framing processing on the near-end signal and the far-end signal, respectively. Obtaining N pairs of synchronization frames, filtering the N pairs of synchronization frames by a filter, and obtaining N filter coefficient vector mean values corresponding to the N pairs of synchronization frames. Further, according to the N filter coefficient vector mean values, N filter coefficient state reference quantities corresponding to the N pairs of synchronization frames are respectively calculated, since the filter coefficient state reference quantity is insensitive to changes in signal energy, and is not affected by the near-end signal and The influence of the environment where the far-end signal is located, so the obtained result of the method is stable, and the calculation complexity of the filter coefficient state reference quantity is low, and no additional operation is needed to smooth the fluctuation of the signal. Finally, when at least M filter coefficient state reference quantities are greater than a preset threshold value among the N filter coefficient state reference quantities, it is determined that the filter state diverges.

Therefore, the embodiment of the present invention uses the filter coefficient state reference quantity to determine whether the filter state is divergent, so as to achieve the purpose of controlling the echo canceler filter coefficient vector update and guaranteeing the call quality, and determining whether the filter state is diverged by the method of the present invention. It can effectively judge whether there is double talk during the call, so as to achieve the purpose of controlling the filter coefficient vector update and obtaining high quality output signal.

Referring to FIG. 5, FIG. 5 is a schematic structural diagram of an apparatus for determining a state of a filter to be diverged according to an embodiment of the present invention.

The acquiring module 501 is configured to collect a near-end signal and a far-end signal synchronized with the near-end signal;

The framing module 502 is configured to perform synchronous framing processing on the near-end signal and the far-end signal collected by the collecting module 501, respectively, to obtain an N-pair synchronization frame, where each pair of synchronization frames includes a near-end signal frame and a near-end signal. Frame-synchronized far-end signal frame;

The filtering module 503 is configured to filter the N pairs of synchronization frames obtained by the framing module 502 by using a filter to obtain N filter coefficient vector mean values corresponding to the N pairs of synchronization frames;

The determining module 504 is configured to calculate N filter coefficient state reference quantities corresponding to the N pairs of synchronization frames according to the N filter coefficient vector average values obtained by the filtering module 503;

And determining that the filter state is divergent when at least M filter coefficient state reference quantities are greater than a preset threshold value among the N filter coefficient state reference quantities;

Where M is a preset value, N and M are both positive integers, and N≥M.

Optionally, the framing module 502 is specifically configured to:

According to the preset frame length, the near-end signals collected by the acquisition module 501 are subjected to frame processing in sequence according to the signal sequence to obtain N near-end signal frames;

At the same time, according to the preset frame length, the far-end signals collected by the acquisition module 501 and synchronized with the near-end signals are subjected to frame processing in sequence according to the signal sequence, and N far-end signal frames synchronized with the near-end signal frame are obtained;

Any one of the obtained near-end signal frame and the far-end signal frame synchronized with any one of the near-end signal frames constitutes a pair of synchronization frames, and a total of N pairs of synchronization frames are obtained.

Optionally, the filtering module 503 is specifically configured to:

For each pair of synchronization frames obtained by the framing module 502, respectively:

Determining, according to the preset frame length and the first preset algorithm, a filter coefficient vector update number T corresponding to the current pair of synchronization frames, and T update results corresponding to the T-th filter coefficient vector update, where T is a positive integer;

The filter coefficient vector mean values corresponding to the current pair of synchronization frames are obtained according to the filter coefficient vector update order T and the T-th filter coefficient vector corresponding to the current pair of synchronization frames.

Optionally, the determining module 504 is specifically configured to:

For each filter coefficient vector mean obtained by the filtering module 503, respectively:

According to the current filter coefficient vector mean value, the filter coefficient distribution map corresponding to the current filter coefficient vector mean value is determined according to a preset layout rule, and

According to the filter coefficient distribution map, the elements in the current filter coefficient vector mean are at least divided into a main reflection area element, a left side element of the main reflection area, and a right side element of the main reflection area, wherein the main reflection area element refers to a filter coefficient The element contained in the preset sliding window when the sum of the squares of all the elements contained in the preset sliding window is the maximum value in the distribution map, and the element on the left side of the preset sliding window in the filter coefficient distribution map is the left element of the main reflection area. The element on the right side of the preset sliding window is the element on the right side of the main reflection area;

Calculating the sum of the squares of the elements on the right side of all the main reflection areas as the first energy value corresponding to the current filter coefficient vector mean;

Calculate the sum of the squares of all the main reflection area elements, the elements on the left side of all the main reflection areas, and the elements on the right side of all the main reflection areas, as the second energy value corresponding to the current filter coefficient vector mean, and

Calculating a smoothed value of the second energy value according to the second preset algorithm;

And determining, by the third preset algorithm, a filter coefficient state reference amount corresponding to the current filter coefficient vector mean according to the first energy value and the smoothed value of the second energy value.

Optionally, the determining module 504 is further configured to:

For each filter coefficient vector mean obtained by the filtering module 503, respectively:

According to the current filter coefficient vector mean value, the filter coefficient distribution map corresponding to the current filter coefficient vector mean value is determined according to a preset layout rule, and

According to the filter coefficient distribution map, the elements in the current filter coefficient vector mean are at least divided into The main reflection area element, the left side element of the main reflection area, and the right side element of the main reflection area, wherein the main reflection area element refers to the sum of the squares of all the elements included in the preset sliding window in the filter coefficient distribution map. Set the elements contained in the sliding window, and the element on the left side of the preset sliding window in the filter coefficient distribution map is the left element of the main reflection area, and the element on the right side of the preset sliding window is the right element of the main reflection area;

Calculate the sum of the squares of all the main reflection area elements, the elements on the left side of all the main reflection areas, and the elements on the right side of all the main reflection areas, as the second energy value corresponding to the current filter coefficient vector mean, and

Calculating a smoothed value of the second energy value according to the second preset algorithm;

Calculating the sum of the squares of the elements on the left side of all the main reflection areas and the elements on the right side of all the main reflection areas as the third energy value corresponding to the mean value of the current filter coefficient vector;

And determining, by the third preset algorithm, a filter coefficient state reference amount corresponding to the current filter coefficient vector mean according to the third energy value and the smoothed value of the second energy value.

Optionally, the determining module 504 is further configured to:

Determining an element having the largest absolute value of the current filter coefficient vector mean according to the filter coefficient distribution map;

Constructing a sliding window, determining the window height of the sliding window according to the absolute value of the element with the largest absolute value, and the preset value is the window length of the sliding window;

Sliding the sliding window on the filter coefficient profile from the first element of the current filter coefficient vector mean to the last element, determining that the sum of the squares of all the elements contained in the sliding window is the maximum value, and the sliding window contains The element is the main reflection zone element, the element on the left side of the sliding window is the left element of the main reflection zone, and the element on the right side of the sliding window is the element on the right side of the main reflection zone.

Here, the principle is the same as the method part, as shown in Fig. 4A.

Optionally, the device further includes:

The reset module 505 is configured to stop updating the filter coefficient vector after the determination module determines that the filter state is diverged, and clear the current filter coefficient vector.

It should be noted that, in the embodiment of the present invention, the division of the module is schematic, and only one logical function is divided. In actual implementation, there may be another division manner. In addition, each functional module in each embodiment of the present application may be used. Can be integrated in one processing module or individual modules There are two or more modules that can be integrated into one module. The above integrated modules can be implemented in the form of hardware or in the form of software functional modules.

The integrated modules, if implemented in the form of software functional modules and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application, in essence or the contribution to the prior art, or all or part of the technical solution may be embodied in the form of a software product stored in a storage medium. A number of instructions are included to cause a computer device (which may be a personal computer, server, or network device, etc.) or a processor to perform all or part of the steps of the methods described in various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like. .

The embodiment of the present invention further provides a device for determining the state of the filter to be diverged. As shown in FIG. 6 , FIG. 6 is a schematic structural diagram of a device for determining a state of a filter to be diverged according to an embodiment of the present invention, where the device includes a transceiver 601 and a processor. 602. Memory 603. The transceiver 601, the processor 602, and the memory 603 are connected to each other. The specific connecting medium between the above components is not limited in the embodiment of the present invention. In the embodiment of the present invention, the memory 603, the processor 602, and the transceiver 601 are connected by a bus 604 in FIG. 6. The bus is indicated by a thick line in FIG. 6, and the connection manner between other components is only schematically illustrated. , not limited to. The bus can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only one thick line is shown in Figure 6, but it does not mean that there is only one bus or one type of bus.

In the embodiment of the present invention, the transceiver 601 is configured to collect a near-end signal and a far-end signal synchronized with the near-end signal.

The processor 602 is configured to perform synchronous framing processing on the near-end signal and the far-end signal collected by the transceiver 601 to obtain N pairs of synchronization frames, where each pair of synchronization frames includes a near-end signal frame and a near-end signal frame. Synchronizing the far-end signal frame, and further filtering the obtained N pairs of synchronization frames through a filter to obtain N filter coefficient vector mean values corresponding to N pairs of synchronization frames, and finally, according to the obtained N filter coefficient vector mean values , respectively calculating N filter coefficient states corresponding to N pairs of sync frames a reference quantity, and determining that the filter state diverges when at least M filter coefficient state reference quantities are greater than a preset threshold value among the N filter coefficient state reference quantities, wherein M is a preset value, N and M Both are positive integers and N ≥ M.

In the embodiment of the present invention, the memory 603 is used to store the program code executed by the processor 602, and may be a volatile memory, such as a random access memory (English: random-access memory, abbreviation: RAM); The memory 603 may also be a non-volatile memory (English: non-volatile memory), such as read-only memory (English: read-only memory, abbreviation: ROM), flash memory (English: flash memory), hard disk (English: hard Disk drive, abbreviated as: HDD) or solid state drive (English: solid-state drive, SSD), or memory 603 can be used to carry or store desired program code in the form of an instruction or data structure and can be accessed by a computer. Any other medium, but not limited to this. The memory 603 may be a combination of the above memories.

The processor 602 in the embodiment of the present invention may be a central processing unit (English: central processing unit, CPU for short).

In summary, the embodiment of the present invention uses the filter coefficient state reference quantity to determine whether the filter state is diverged, so as to achieve the purpose of controlling the echo canceler filter coefficient vector update and guaranteeing the call quality, because the filter coefficient state reference quantity pair The change of signal energy is not sensitive and is not affected by the environment of the near-end signal and the far-end signal. Therefore, the obtained result of the method has good stability, and the calculation complexity of the filter coefficient state reference quantity is low, therefore, Whether the filter state is diverged by the method of the present invention can effectively determine whether there is double talk during the call, thereby achieving the purpose of controlling the filter coefficient vector update and obtaining a high quality output signal.

Those skilled in the art will appreciate that embodiments of the present invention can be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or a combination of software and hardware. Moreover, the invention can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) including computer usable program code.

The present invention is directed to a method, apparatus (system), and computer program according to an embodiment of the present invention. The flow chart and/or block diagram of the product is described. It will be understood that each flow and/or block of the flowchart illustrations and/or FIG. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing device to produce a machine for the execution of instructions for execution by a processor of a computer or other programmable data processing device. Means for implementing the functions specified in one or more of the flow or in a block or blocks of the flow chart.

The computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture comprising the instruction device. The apparatus implements the functions specified in one or more blocks of a flow or a flow and/or block diagram of the flowchart.

These computer program instructions can also be loaded onto a computer or other programmable data processing device such that a series of operational steps are performed on a computer or other programmable device to produce computer-implemented processing for execution on a computer or other programmable device. The instructions provide steps for implementing the functions specified in one or more of the flow or in a block or blocks of a flow diagram.

While the preferred embodiment of the invention has been described, it will be understood that Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and the modifications and

It is apparent that those skilled in the art can make various modifications and variations to the embodiments of the invention without departing from the spirit and scope of the embodiments of the invention. Thus, it is intended that the present invention cover the modifications and modifications of the embodiments of the invention.

Claims (14)

1. A method for determining a state of a filter divergence, comprising:

   Collecting a near-end signal and a far-end signal synchronized with the near-end signal, and

   Performing synchronous framing processing on the near-end signal and the far-end signal respectively to obtain N-pair sync frames, each pair of sync frames including a near-end signal frame and a far-end signal synchronized with the near-end signal frame frame;

   And filtering the N pairs of synchronization frames by using a filter to obtain N filter coefficient vector mean values corresponding to the N pairs of synchronization frames;

   Calculating, according to the N filter coefficient vector mean values, N filter coefficient state reference quantities corresponding to the N pairs of synchronization frames;

   Determining that the filter state is divergent when at least M filter coefficient state reference quantities are greater than a preset threshold value in the N filter coefficient state reference quantities;

   Where M is a preset value, and both N and M are positive integers, and N≥M.
2. The method according to claim 1, wherein said performing synchronous framing processing on said near-end signal and said far-end signal respectively to obtain N-pair synchronization frames comprises:

   According to the preset frame length, the near-end signals are sequenced according to the signal timing sequence to obtain N near-end signal frames;

   At the same time, according to the preset frame length, the far-end signals synchronized with the near-end signals are sequentially processed according to the signal timing sequence to obtain N far-end signal frames synchronized with the near-end signal frame;

   Any one of the obtained near-end signal frame and the far-end signal frame synchronized with the any one of the near-end signal frames constitutes a pair of synchronization frames, and a total of N pairs of synchronization frames are obtained.
3. The method according to claim 1, wherein said filtering the N pairs of synchronization frames by a filter to obtain N filter coefficient vector mean values corresponding to N pairs of synchronization frames, comprising:

   For each pair of sync frames, execute separately:

   Determining a filter coefficient vector corresponding to the current pair of synchronization frames according to a preset frame length and a first preset algorithm The quantity update number T, and the T update results corresponding to the T-th filter coefficient vector update, where T is a positive integer;

   And obtaining, according to the filter coefficient vector update order T corresponding to the current pair of synchronization frames and the T update result of the T-th filter coefficient vector, obtaining a filter coefficient vector mean corresponding to the current pair of synchronization frames.
4. The method according to claim 1 or 2, wherein the N filter coefficient state reference quantities corresponding to the N pairs of synchronization frames are respectively calculated according to the N filter coefficient vector mean values, including:

   For each filter coefficient vector mean, perform:

   Determining, according to a preset layout rule, a filter coefficient distribution map corresponding to the current filter coefficient vector mean according to a current filter coefficient vector mean value, and

   According to the filter coefficient distribution map, the elements in the current filter coefficient vector mean are at least divided into a main reflection area element, a left side element of the main reflection area, and a right side element of the main reflection area, wherein the main reflection area An element is an element included in the preset sliding window when a sum of squares of all elements included in a preset sliding window in the filter coefficient distribution map is a maximum value, and the pre-predetermined in the filter coefficient distribution map The element on the left side of the sliding window is an element on the left side of the main reflection area, and the element on the right side of the preset sliding window is an element on the right side of the main reflection area;

   Calculating a sum of squares of elements on the right side of all the main reflection areas as a first energy value corresponding to the mean value of the current filter coefficient vector;

   Calculating the sum of the squares of all the main reflection area elements, the elements on the left side of all the main reflection areas, and the elements on the right side of all the main reflection areas, as the second energy value corresponding to the mean value of the current filter coefficient vector, and

   Calculating a smoothed value of the second energy value according to a second preset algorithm;

   And determining, according to the first energy value and the smoothed value of the second energy value, a filter coefficient state reference quantity corresponding to the current filter coefficient vector mean value by using a third preset algorithm.
5. The method according to claim 1 or 2, wherein the N filter coefficient state reference quantities corresponding to the N pairs of synchronization frames are respectively calculated according to the N filter coefficient vector mean values, including:

   For each filter coefficient vector mean, perform:

   Determining the current filter according to a preset layout rule according to a current filter coefficient vector mean a coefficient coefficient map corresponding to the coefficient vector mean, and

   According to the filter coefficient distribution map, the elements in the current filter coefficient vector mean are at least divided into a main reflection area element, a left side element of the main reflection area, and a right side element of the main reflection area, wherein the main reflection area An element is an element included in the preset sliding window when a sum of squares of all elements included in a preset sliding window in the filter coefficient distribution map is a maximum value, and the pre-predetermined in the filter coefficient distribution map The element on the left side of the sliding window is an element on the left side of the main reflection area, and the element on the right side of the preset sliding window is an element on the right side of the main reflection area;

   Calculating the sum of the squares of all the main reflection area elements, the elements on the left side of all the main reflection areas, and the elements on the right side of all the main reflection areas, as the second energy value corresponding to the mean value of the current filter coefficient vector, and

   Calculating a smoothed value of the second energy value according to the second preset algorithm;

   Calculating the sum of the squares of the elements on the left side of all the main reflection areas and the elements on the right side of all the main reflection areas as the third energy value corresponding to the mean value of the current filter coefficient vector;

   And determining, by the third preset algorithm, a filter coefficient state reference quantity corresponding to the current filter coefficient vector mean according to the third energy value and the smoothed value of the second energy value.
6. The method according to claim 4 or 5, wherein the elements in the current filter coefficient vector mean are at least divided into a main reflection area element and a left reflection element of the main reflection area according to the filter coefficient distribution map. And the elements to the right of the main reflection area, including:

   Determining, according to the filter coefficient distribution map, an element having the largest absolute value among the current filter coefficient vector mean values;

   Constructing a sliding window, determining a window height of the sliding window according to an absolute value of the element having the largest absolute value, and using a preset value as a window length of the sliding window;

   Sliding the sliding window from the first element of the current filter coefficient vector mean to the last element on the filter coefficient profile to determine that the sum of squares of all elements included in the sliding window is maximum When the value is, the element included in the sliding window is the main reflection area element, the element on the left side of the sliding window is an element on the left side of the main reflection area, and the element on the right side of the sliding window is the main reflection The right side of the area.
7. The method of claim 3 wherein said determining said filter state is divergent Thereafter, the method further includes:

   Stop updating the filter coefficient vector and clear the current filter coefficient vector.
8. A device for determining a state of divergence of a filter, comprising:

   An acquisition module, configured to collect a near-end signal and a far-end signal synchronized with the near-end signal;

   a framing module, configured to perform synchronous framing processing on the near-end signal and the far-end signal collected by the collecting module, respectively, to obtain N pairs of synchronization frames, each pair of synchronization frames including a near-end signal frame and a a far-end signal frame synchronized with the near-end signal frame;

   a filtering module, configured to filter, by using a filter, the N pairs of synchronization frames obtained by the framing module to obtain N filter coefficient vector mean values corresponding to the N pairs of synchronization frames;

   a determining module, configured to calculate, according to the N filter coefficient vector average values obtained by the filtering module, N filter coefficient state reference quantities corresponding to the N pairs of synchronization frames;

   And determining, when the at least M filter coefficient state reference quantities are greater than a preset threshold value, in the N filter coefficient state reference quantities, determining the filter state divergence;

   Where M is a preset value, and both N and M are positive integers, and N≥M.
9. The apparatus according to claim 8, wherein the framing module is specifically configured to:

   According to the preset frame length, the near-end signals collected by the collection module are sequentially processed according to the signal timing sequence to obtain N near-end signal frames;

   Simultaneously, according to the preset frame length, the far-end signals that are collected by the acquiring module and synchronized with the near-end signal are subjected to frame processing in sequence according to signal timing, to obtain N synchronization with the near-end signal frame. Remote signal frame;

   Any one of the obtained near-end signal frame and the far-end signal frame synchronized with the any one of the near-end signal frames constitutes a pair of synchronization frames, and a total of N pairs of synchronization frames are obtained.
10. The device according to claim 8, wherein the filtering module is specifically configured to:

    For each pair of synchronization frames obtained by the framing module, respectively:

    Determining, according to the preset frame length and the first preset algorithm, a filter coefficient vector update number T corresponding to the current pair of synchronization frames, and T update results corresponding to the T-th filter coefficient vector update, where T is a positive integer;

    And obtaining, according to the filter coefficient vector update order T corresponding to the current pair of synchronization frames and the T update result of the T-th filter coefficient vector, obtaining a filter coefficient vector mean corresponding to the current pair of synchronization frames.
11. The device according to claim 8 or 9, wherein the determining module is specifically configured to:

    For each filter coefficient vector mean obtained by the filtering module, respectively:

    Determining, according to a preset layout rule, a filter coefficient distribution map corresponding to the current filter coefficient vector mean according to a current filter coefficient vector mean value, and

    According to the filter coefficient distribution map, the elements in the current filter coefficient vector mean are at least divided into a main reflection area element, a left side element of the main reflection area, and a right side element of the main reflection area, wherein the main reflection area An element is an element included in the preset sliding window when a sum of squares of all elements included in a preset sliding window in the filter coefficient distribution map is a maximum value, and the pre-predetermined in the filter coefficient distribution map The element on the left side of the sliding window is an element on the left side of the main reflection area, and the element on the right side of the preset sliding window is an element on the right side of the main reflection area;

    Calculating a sum of squares of elements on the right side of all the main reflection areas as a first energy value corresponding to the mean value of the current filter coefficient vector;

    Calculating the sum of the squares of all the main reflection area elements, the elements on the left side of all the main reflection areas, and the elements on the right side of all the main reflection areas, as the second energy value corresponding to the mean value of the current filter coefficient vector, and

    Calculating a smoothed value of the second energy value according to a second preset algorithm;

    And determining, according to the first energy value and the smoothed value of the second energy value, a filter coefficient state reference quantity corresponding to the current filter coefficient vector mean value by using a third preset algorithm.
12. The device according to claim 8 or 9, wherein the determining module is further configured to:

    For each filter coefficient vector mean obtained by the filtering module, respectively:

    Determining, according to a preset layout rule, a filter coefficient distribution map corresponding to the current filter coefficient vector mean according to a current filter coefficient vector mean value, and

    According to the filter coefficient distribution map, the elements in the current filter coefficient vector mean are at least divided into a main reflection area element, a left side element of the main reflection area, and a right side element of the main reflection area, wherein The main reflection area element refers to an element included in the preset sliding window when the sum of squares of all elements included in the preset sliding window in the filter coefficient distribution map is a maximum value, and the filter coefficient distribution map The element on the left side of the preset sliding window is an element on the left side of the main reflection area, and the element on the right side of the preset sliding window is an element on the right side of the main reflection area;

    Calculating the sum of the squares of all the main reflection area elements, the elements on the left side of all the main reflection areas, and the elements on the right side of all the main reflection areas, as the second energy value corresponding to the mean value of the current filter coefficient vector, and

    Calculating a smoothed value of the second energy value according to the second preset algorithm;

    Calculating the sum of the squares of the elements on the left side of all the main reflection areas and the elements on the right side of all the main reflection areas as the third energy value corresponding to the mean value of the current filter coefficient vector;

    And determining, by the third preset algorithm, a filter coefficient state reference quantity corresponding to the current filter coefficient vector mean according to the third energy value and the smoothed value of the second energy value.
13. The device according to claim 11 or 12, wherein the determining module is further configured to:

    Determining, according to the filter coefficient distribution map, an element having the largest absolute value among the current filter coefficient vector mean values;

    Constructing a sliding window, determining a window height of the sliding window according to an absolute value of the element having the largest absolute value, and using a preset value as a window length of the sliding window;

    Sliding the sliding window from the first element of the current filter coefficient vector mean to the last element on the filter coefficient profile to determine that the sum of squares of all elements included in the sliding window is maximum When the value is, the element included in the sliding window is the main reflection area element, the element on the left side of the sliding window is an element on the left side of the main reflection area, and the element on the right side of the sliding window is the main reflection The right side of the area.
14. The device of claim 10, wherein the device further comprises:

    And a resetting module, configured to stop updating the filter coefficient vector after the determining, according to the determining module, that the filter state is diverged, and clear the current filter coefficient vector.

Images