JP2017098861A - Echo canceller and echo cancellation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To determine a delay time of an echo signal in a short time with high accuracy to effectively erase an echo even under an environment where the delay time of the echo signal is unknown or a change in the delay time easily occurs.SOLUTION: An echo canceller 10 comprises: a correlation processing part 103 for calculating cross correlation between an input signal (reception signal) from a far end and an echo signal (transmission signal) reflected by a hybrid circuit or the like; a statistical processing part 104 for determining a delay time based on a cross correlation value and an evaluation value of the delay time: a filter monitoring part 107 for monitoring a filter coefficient of an adaptive filter 106, and for reflecting a monitor result on the evaluation value; a delay processing part 108 for delaying the reception signal in accordance with the determined delay time, and for inputting the reception signal to echo removal means; and an echo erasure part 130 for estimating a pseudo echo from the reception signal, and for subtracting the pseudo echo from the transmission signal. Thus, it is possible to use even a change in the filter coefficient of the adaptive filter generated when the delay time of an echo changes for a determination factor of the delay time, and to determine the delay time in a shorter time.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an echo canceller, and more particularly to an echo canceller for canceling echo generated in a telephone network or the like.

  In a telephone network, a two-wire to four-wire conversion called a hybrid is performed between a telephone line and a subscriber line (two lines: two-way) and a trunk line (four lines: two one-way (two lines) independently). A call is made between two telephones connected by a circuit. However, in a system using such a hybrid, due to impedance mismatching, a signal received from the other party is reflected by the hybrid of the own telephone, and an echo that goes around to the other party is generated. Since the echo is heard by the other party with a delay and interferes with the call, it is necessary to cancel the echo.

JP 2004-297236 A

FIG. 16 shows the configuration of a conventional echo canceller. The echo canceller is installed in the vicinity of the hybrid circuit, and cancels an echo signal that returns to the far-end phone that is opposite to the near-end phone.
The echo canceller 1 normally includes an adaptive filter 2. The adaptive filter 2 performs a convolution operation on the received signal x (n) from the far-end speaker to generate a pseudo echo signal y ′ (n). The echo canceller 1 subtracts the pseudo echo signal y ′ (n) from the transmission signal y (n) to the far-end speaker including the echo to remove the echo and obtain an output transmission signal z (n). . Here, n (n = 1, 2, 3,...) Is a sampling point on the time axis, and indicates that the signal is a discrete signal.
The adaptive filter 2 is usually realized by an FIR (Finite Impulse Response Filter) filter. The output y ′ (n) of the FIR filter includes a plurality of filter coefficients h (i) (i = 1, 2, 3,...) Corresponding to a unit impulse response in each of the samples n on the time axis and an input x ( n−i + 1) and the sum of products, ie, convolutional addition. As an adaptive algorithm for estimating each filter coefficient, an algorithm such as a LMS (least mean square) or a learning identification method is used. The general techniques for adaptive signal processing and acoustic echo canceller are as follows: The Institute of Electronics, Information and Communication Engineers, 1 group, 9 editions, Signal / System, 3 chapters, Adaptive signal processing, 2 groups, 6 editions, Acoustic signal processing, Chapter 5, Acoustic echo canceller, etc. Is disclosed.

As an example, in an adaptive filter composed of a 128-tap FIR filter, the gain of each filter coefficient is adjusted to create a pseudo echo signal from the received signal x (n), and the echo is canceled in the echo canceller. Considering a received signal x sampled at 8 kHz, 128 × 125 μSec. = 16 mSec. It is possible to cancel echoes with delays up to.
On the other hand, 16 mSec. A sufficient amount of erasure cannot be expected for echoes that are delayed as described above. Such an echo can be eliminated by giving a fixed delay to the signal input to the echo canceller. In this case, the echo cannot be effectively eliminated unless the delay time is known in advance. Furthermore, there is a problem that when the delay time changes (for example, when the echo path changes), the echo cannot be effectively erased. In addition, as an example in which the communication time changes, a hybrid circuit is exchanged, software that performs processing such as voice conversion is changed or updated, and some functions such as voice conversion are applied / not applied. For example, switching.

In Patent Document 1, a delay time is estimated from a cross-correlation between a signal input from a far-end telephone and an echo signal, and the delay time having the maximum frequency is applied as a signal to be input to an echo canceller using a histogram. However, the technique of Patent Document 1 uses an integrated value of values based on cross-correlation to determine the delay time, and when the delay time changes, it may take time until the delay time is changed. Meanwhile, there was a problem that the echo cancellation performance deteriorated.
In view of the above, an object of the present invention is to provide an echo canceller that cancels echoes even when the delay time of an echo signal is unknown or changes occur in the delay time.

According to the first solution of the present invention,
An echo canceler for canceling an echo signal included in the transmission signal to the far end side using an adaptive filter;
Correlation for obtaining a plurality of delay time candidates for a received signal based on the cross-correlation by obtaining one of the received signal from the far end and the transmission signal to the far end while shifting one signal at a time. A processing unit;
A statistical processing unit for obtaining an evaluation value for a delay time candidate based on the obtained cross-correlation history;
A storage unit for storing the evaluation value obtained by the statistical processing unit in correspondence with the delay time;
A delay processing unit that selects one of delay time candidates based on the evaluation value stored in the storage unit, delays a received signal according to the selected delay time, and outputs the delayed signal to the adaptive filter;
There is provided an echo canceller comprising a filter monitoring unit that monitors a filter coefficient of the adaptive filter or an echo cancellation amount in the echo cancellation unit and changes an evaluation value stored in the storage unit according to a monitoring result. .

According to the second solution of the present invention,
An echo cancellation method for canceling an echo signal included in a transmission signal to a far end side using an adaptive filter,
One of the received signal from the far end side and the transmission signal to the far end side is shifted by a predetermined time to obtain the cross-correlation of both signals, and based on the cross-correlation, a plurality of delay time candidates for the received signal are obtained,
Based on the obtained cross-correlation history, obtain an evaluation value for the delay time candidate,
The obtained evaluation value is stored in correspondence with the delay time,
Select one of the delay time candidates based on the stored evaluation value, delay the received signal according to the selected delay time, and output to the adaptive filter,
An echo cancellation method is provided that monitors the filter coefficient or echo cancellation amount of an adaptive filter and reduces the stored evaluation value in accordance with the monitoring result.

  According to the present invention, it is possible to provide an echo canceller that cancels an echo even if the delay time of the echo signal is unknown or a change occurs in the delay time.

It is a block diagram which shows the structural example of the echo canceller in embodiment of this invention. It is a hardware block diagram of an echo canceller. It is an operation | movement flowchart which shows the operation | movement outline | summary of an echo canceller. It is explanatory drawing of a buffer. It is an operation | movement flowchart which shows the detailed operation example of a correlation process. It is a graph showing the relationship between the cross correlation acquired by a correlation process, and delay time. It is an example of a management table. It is an operation | movement flowchart which shows the detailed operation example of a statistical process. It is an operation | movement flowchart which shows the detailed operation example of a delay process. It is an operation | movement flowchart which shows the detailed operation example of a filter monitoring process. It is an operation | movement flowchart which shows the detailed operation example of evaluation value addition / subtraction processing. It is an example of the management table with which the evaluation value was reduced. FIG. 10 is an operation flowchart illustrating a detailed operation example of filter monitoring processing in the second embodiment. FIG. 10 is an operation flowchart illustrating a detailed operation example of filter monitoring processing in the third embodiment. It is a schematic block diagram of the telephone conference system in the modification of this invention. It is a block diagram which shows the structural example of the conventional echo canceller.

Example 1
FIG. 1 is a block diagram showing a configuration of an echo canceller 10 in the present embodiment. FIG. 1 shows a logical configuration of the echo canceller 10, which is realized by a heartware configuration such as a processor (CPU) and a memory which will be described later.
Similarly to the echo canceller 1 shown in FIG. 16, the echo canceller 10 receives a signal (Rin) from the far-end telephone and transmits a signal (Sout) to the far-end telephone. The echo canceller 10 receives a signal (Sin) from the near-end phone via, for example, a hybrid circuit, and outputs a signal (Rout) to the near-end phone, for example, to the hybrid circuit to transmit it to the near-end phone. . The echo canceller 10 can be provided in a communication device, for example.
The echo canceller 10 includes, for example, a buffer processing unit 100 that copies a reception signal x (n) from a far-end telephone and a transmission signal y (n) to the far-end telephone including a delayed echo, and a far-end telephone. Receive buffer 101 for accumulating received signals, transmit buffer 102 for accumulating transmission signals to the far-end telephone including delayed echo signals, and delay from cross-correlation of signals accumulated in receive buffer 101 and transmit buffer 102 Correlation processing unit 103 for selecting time candidates, statistical processing unit for performing statistical processing on cross-correlation and delay time obtained as an output result of correlation processing unit 103, and evaluating the validity of delay time and determining an evaluation value 104 and a management table (storage unit) that stores the cross-correlation, delay time, and evaluation value obtained as an output result of the statistical processing unit 104 05, an adaptive filter 106 used as echo cancellation means, a filter coefficient of the adaptive filter 106 and an echo cancellation amount, and a filter monitoring unit 107 that adjusts the evaluation value of the delay time stored in the management table 105 based on the monitoring result. A delay processing unit 108 that delays audio input from the data line 110 by a delay time having a high evaluation value out of the delay times stored in the management table 108, and outputs to the adaptive filter 106, and a far end including an echo A subtractor 109 that subtracts a pseudo echo signal from a transmission signal y (n) to the telephone to cancel the echo, a logical data line 110 from the far-end telephone to the near-end telephone, and a near-end telephone to the far-end telephone And a logical data line 120. The adaptive filter 106 and the subtractor 109 constitute an echo canceling unit 130.

Hereinafter, each block will be described.
The buffer processing unit 100 stores the reception signal received from the far-end telephone and the reception buffer 101. Further, the buffer processing unit 100 stores a transmission signal to the far-end telephone including the delayed echo signal in the transmission buffer 102.
The correlation processing unit 103 calculates the cross-correlation between the reception signal and the transmission signal accumulated in the reception buffer 101 and the transmission buffer 102. The correlation processing unit 103 determines a delay time at a position where the cross-correlation is large, and outputs the delay time to the statistical processing unit 104 described later as a delay time candidate.
The statistical processing unit 104 performs statistical processing on the cross-correlation and the delay time obtained as the output result of the correlation processing unit 103 and stores them in the management table 105. The reason for performing the statistical processing is to increase the resistance of the echo canceller to the instantaneous change by averaging the cross-correlation and delay time which change greatly in an instant.

Further, the statistical processing unit 104 calculates an evaluation value of the delay time and stores it in the management table 105. The evaluation value is, for example, a value that integrates the weight based on the cross correlation every time the management table 105 is updated. That is, the value reflects the history of cross-correlation from the past to the present. Specifically, the delay time is evaluated higher as the number of updates of the delay time held in the management table 105 is larger and the cross-correlation is larger.
The management table 105 is a storage area that stores delay times, cross-correlations, and evaluation values in association with each other. The management table 105 is updated by the statistical processing unit 104 and the filter monitoring unit 107.
The adaptive filter 106 is echo canceling means similar to the conventional one shown by the adaptive filter 2 described above.
The filter monitoring unit 107 monitors the filter coefficient of the adaptive filter 106. If the adaptive filter 106 does not have the assumed filter coefficient, the filter monitoring unit 107 reduces the evaluation value of the delay time stored in the management table 105. When the filter coefficient is assumed, the evaluation value is increased to prompt the change of the delay time used by the delay processing unit 108 described later.
The delay processing unit 108 selects a delay time having a high evaluation value from the plurality of delay times stored in the management table 105, delays the delay time by the delay time, and outputs the delay time to the adaptive filter 106.

FIG. 2 is a hardware configuration diagram of the echo canceller 10.
The echo canceller 10 includes, for example, a processor 21, a memory 22, and an interface 23. The processor 21 is a processing unit that executes processing of the echo canceller 10. The memory 22 stores a program that is executed by the processor 21 and implements each unit shown in FIG. The memory 22 has storage areas corresponding to the reception buffer 101, the transmission buffer 102, and the management table 105 shown in FIG.
Hereinafter, the operation of the echo canceller 10 according to the present embodiment will be described in detail with reference to FIGS. Specifically, an example will be described in which processing is performed on a signal sampled at 8 kHz and echoes delayed up to a maximum of 128 ms are eliminated. The adaptive filter 106 used as the echo canceling means is assumed to be capable of canceling echoes of up to 16 ms delay with 128 taps.

FIG. 3 is a flowchart showing the operation of the echo canceller. The echo canceller 10 generally operates as follows.
In the buffer processing 210, the buffer processing unit 100 stores the reception signal and the transmission signal sequentially copied from the data lines 110 and 120 in the reception buffer 101 and the transmission buffer 102, respectively.
In the correlation processing 220, the correlation processing unit 103 calculates the cross-correlation between the two signals while shifting one of the signals accumulated in the reception buffer 101 and the transmission buffer 102 in the time direction (corresponding to the delay time), thereby obtaining a high cross-correlation. A plurality of delay times having “” are selected and used as an input to the statistical processing 230.
In the statistical processing 230, the statistical processing unit 104 updates the management table 105 based on the input from the correlation processing 220.
In the delay processing 240, the delay processing unit 108 acquires a delay time having a high evaluation value from the management table 105, delays the received signal by the delay time, and inputs it to the adaptive filter 106.
In the echo cancellation process 250, the adaptive filter 106 creates a pseudo echo for the received signal delayed by the delay process 240 by the same method as the conventional echo canceller 2, and subtracts the echo from the transmission signal to cancel the echo. .
In the filter monitoring process 260, the filter monitoring unit 107 monitors the position where the filter coefficient of the adaptive filter 106 is maximum, and uses the monitoring result as an input to the evaluation value addition / subtraction process 270.
In the evaluation value addition / subtraction processing 270, the filter monitoring unit 107 evaluates the delay time of the management table 105 when the maximum value of the filter coefficient is not located in the first half of the adaptive filter 106 based on the result of the filter monitoring processing 260. Reduce the value. On the other hand, when the maximum value of the filter coefficient is located in the first half of the adaptive filter 106, the filter monitoring unit 107 increases the evaluation value of the delay time that the management table 105 has.

FIG. 4 is an explanatory diagram of the buffer. The echo canceller 10 described in the present embodiment assumes that the echo delayed for a maximum of 128 ms is canceled, so that the size Rlen of the reception buffer 101 is an area corresponding to 128 ms. Since adaptive filter 106 assumes 128 taps, size Slen of transmission buffer 102 is set to an area corresponding to 16 ms. The size of the buffer can be set to an appropriate size according to the delay amount to be subjected to echo cancellation and the number of taps of the adaptive filter.
Each of the reception buffer 101 and the transmission buffer 102 is a FIFO (First In First Out) type buffer. When a new signal is input, the signal accumulated at the head of the buffer (the signal accumulated most recently) is discarded. And a new signal is stored at the end.

  FIG. 5 is a detailed flowchart of the correlation process 220. In the correlation processing 220, the correlation processing unit 103 sequentially shifts the signal S stored in the transmission buffer 102 in the time direction with respect to the signal R stored in the reception buffer 101, while delaying the delay time Delay (i) of the transmission signal. Then, the cross-correlation CCor (i) between the received signal and the transmitted signal is calculated (processing 410 to 450). In the illustrated example, the signal S accumulated in the transmission buffer 102 is converted to 125 mSec. Although the values are shifted one by one, the value to be shifted may be an appropriate value. The correlation processing unit 103 outputs one or a plurality of sets of the cross-correlation having a high value among the obtained cross-correlation CCor (i) and the corresponding delay time to the statistical processing unit 230 (processing 460). For example, a predetermined number of pairs having higher cross-correlation may be output, a pair having a cross-correlation larger than a predetermined threshold may be output, or the cross-correlation may be performed according to another predetermined criterion. A relatively high set may be output.

  FIG. 6 is an image diagram of the cross-correlation CCor (i) and the delay time Delay (i) obtained by the correlation processing 220 of FIG. In this embodiment, as an example, a set (two sets) of cross-correlation and delay time corresponding to i = 256 (corresponding to a delay time of 32 mSec.) And i = 512 (corresponding to a delay time of 64 mSec.) In FIG. It is assumed that the data is output to the process 230.

  FIG. 7 is an explanatory diagram illustrating an example of the management table 105. The management table 105 holds the delay time D510, the cross-correlation C520, and the evaluation value E530 in association with each other. In this embodiment, a management table 500 having three arrays with column numbers m = 0 to 2 is considered.

FIG. 8 is a detailed flowchart of the statistical process 230. The statistical processing unit 104 updates the value of the cross-correlation C in the management table 500 corresponding to the delay time Delay (i) input from the correlation processing unit 103. At this time, the statistical processing unit 104 performs a statistical process such as taking a moving average of the previous value of the cross-correlation C to be updated and the value of the cross-correlation C input from the correlation processing unit 103, so that the instantaneous cross-correlation It is possible to provide resistance to change (operation 620). In addition to the moving average, an appropriate statistical process in which the previous value of the cross-correlation C to be updated affects the cross-correlation C after the update may be used.
Further, the statistical processing unit 104 calculates an evaluation value E. Here, the evaluation value E is a value obtained by integrating the weight w corresponding to the magnitude of the cross-correlation C every time the delay time D is updated. A higher evaluation value E can be obtained as the value of the cross-correlation C is larger and the number of updates of the delay time D is larger. In FIG. 8, as an example, 100 is used as a threshold for the cross-correlation C, the weight w = 0 when the value of the cross-correlation C is 100 or less, and the weight w = 2 when the cross-correlation C is greater than 100. The delay time having a high cross-correlation increases the evaluation value (processing 630 to 660).

  FIG. 9 is a detailed flowchart of the delay process 240. The delay processing unit 108 refers to the management table 105 (500), identifies the column with the maximum evaluation value E, and acquires the delay time D corresponding to the maximum evaluation value E (processing 710). The delay processing unit 108 delays the received signal input from the line 110 by an amount corresponding to the acquired delay time D and outputs the delayed signal to the adaptive filter 106 (processing 720). As a result, the received signal is adjusted to a position in the time direction so as to have a high correlation with the transmission signal, and is output to the adaptive filter 106. Therefore, the echo cancellation process 250 is an adaptive filter of 128 taps (16 ms). Also, echoes delayed up to 128 ms can be canceled.

  FIG. 10 is a detailed flowchart of the filter monitoring process 260. The filter monitoring unit 107 divides the 128-tap adaptive filter into N = 8 taps and M = 16 blocks to obtain have (n) that is an average value of the filter coefficients h for each block. A position (block identification information indicating the position of the block on the time axis) at which the average value have (n) of the coefficient is maximum is calculated (processing 820 to 840). Processes 820 to 840 are performed on all 16 blocks (processes 810 to 860), and a position (block identification information) where the calculated have (n) is maximized is provided to the evaluation value adjustment process 270 (process 870). Here, N is a division unit of the adaptive filter and indicates the number of taps included in one block. M is the number of blocks of the adaptive filter divided every N taps.

FIG. 11 is a detailed flowchart of the evaluation value addition / subtraction process 270.
In general, the impulse response of an echo generating element (for example, a hybrid circuit or a wall) existing in the echo path first suddenly rises first and then shows a damped oscillation. Therefore, the filter coefficient that takes the maximum value also in the adaptive filter is a filter coefficient corresponding to the first half of the time axis, and the filter coefficient becomes smaller as the position of the filter coefficient on the time axis becomes farther from the filter coefficient that takes the maximum value. It can be considered that the state of taking the value approximates the echo generating element with high accuracy.
Based on this assumption, the filter monitoring unit 107 determines that the position of the block where the average value of the filter coefficients of the adaptive filter is the maximum obtained in the filter monitoring process is the first half part on the time axis among all the filter coefficients of the adaptive filter. If it is not located at (step 910: Yes), the maximum evaluation value E530 in the management table 105 (500) is reduced (step 920). On the other hand, when the position of the block having the maximum filter coefficient average value is located in the first half portion on the time axis among all the filter coefficients of the adaptive filter (processing 910: Yes), the filter monitoring unit 107 in the management table 500 The maximum evaluation value E530 is increased (step 930).
In this embodiment, since an adaptive filter with 128 taps is assumed, when the position of the block having the maximum filter coefficient average value input from the statistical processing unit 104 is positioned after 16/2 = 8, Then, when it is not located in the first half part on the time axis, the evaluation value E having the maximum value in the management table 500 is reduced to ½. On the other hand, when the position of the block where the average value of the filter coefficients is maximum is located before 8, in other words, when it is located in the first half part on the time axis, the evaluation value E having the maximum value in the management table 500 Is doubled (processes 910 to 930).

When it is determined that the filter coefficient of the adaptive filter 105 does not become the expected filter coefficient when the delay time D derived from the cross correlation C and having the highest evaluation value E is applied, the applied delay The evaluation value E corresponding to the time can be forcibly reduced to encourage the application of the delay time D having the next highest evaluation value E. On the other hand, when the expected filter coefficient is obtained in the delay time D, the evaluation of the delay time D can be enhanced by updating to a higher evaluation value E.
Note that the constants used in FIG. 11 are merely examples, and when it is desired to improve the accuracy of the impulse response of the adaptive filter, the position at which the filter coefficient is maximized is set to the first half of the adaptive filter (for example, the forward 1/3 or 1). / 4) and the like. Further, regarding the adjustment of the evaluation value, it is possible to adjust the degree of adjustment of the evaluation value by appropriately setting a constant for multiplying the evaluation value. Further, the evaluation value may be adjusted by an appropriate method other than multiplication by a constant.

FIG. 12 is another example of the management table. The evaluation value E (1) is reduced by half from the state of FIG. 7 by the above-described filter monitoring process 260 and evaluation value adjustment process 270.
As a result of performing the delay process 240 and the echo cancellation process 250 using the delay time D (1) for which the high evaluation value E (1) is obtained by the correlation process 220 and the statistical process 230, the filter monitoring process 260 and the evaluation value adjustment process 270 are performed. Thus, it is determined that the position of the block where the average value of the filter coefficients is the maximum is different from the assumption, and the value of the evaluation value E (1) is reduced to 50 (= 1/2 × 100). As a result, the echo canceller 10 operates using the delay time D (0) having the evaluation value E (0). From the above operation, the delay time to be used is switched from D (1) to D (0) according to the monitoring result of the filter, so that it is possible to quickly follow the change in the delay time.
In the above example, the average value of the filter coefficient is obtained in units of blocks, and the determination is made based on the block position where the average value of the filter coefficient is maximum. An appropriate value based on the above may be used as a reference, or individual filter coefficients may be used as a reference. If the average value of the filter coefficients is obtained in units of blocks as in the above example, the instantaneous behavior can be absorbed, and the operation is likely to be stable.

(Example 2)
FIG. 13 is a flowchart of the filter monitoring process in the second embodiment. In the filter monitoring process 260 (1000) in the second embodiment, the filter position is monitored by calculating the maximum value of the filter coefficient instead of obtaining the block position where the average value of the filter coefficient is the maximum as in the first embodiment. (Processing 1010-1040). Also in this embodiment, the filter coefficient is divided into blocks, the average value have of filter coefficients in the block is obtained, and the maximum value of the block average value of the filter coefficients is obtained. In this embodiment, the maximum value of the average value of the filter coefficients is provided to the evaluation value adjustment process 270 (process 1070). The evaluation value addition / subtraction process is the same as the process of the first embodiment shown in FIG. 11, but the process 910 is different. In this embodiment, the filter monitoring unit 107 determines whether or not the maximum value of the average value of the filter coefficients is equal to or smaller than a predetermined threshold value. If the threshold value is equal to or smaller than the threshold value, the process proceeds to process 920. Move on. When the maximum average value of the filter coefficients is equal to or less than a predetermined threshold value, it can be determined that there is no clear peak in the adaptive filter and the impulse response is not assumed. For example, when the actual delay time is larger than the applied delay time, it may be considered that there is no peak of the impulse response during the period (for example, 16 ms) in which the adaptive filter effectively operates.
The apparatus configuration and other processes are the same as those in the first embodiment.

(Example 3)
FIG. 14 is a flowchart of the filter monitoring process in the third embodiment. In the filter monitoring process 260 (1100) in the third embodiment, the filter is monitored based on the echo cancellation amount. In this embodiment, an index called ERLE (Echo Return Loss Enhancement) is used as the echo cancellation amount. Based on the echo signal y (n) and the pseudo echo signal y ′ (n), the filter monitoring unit 107 calculates an average value ERLEave within a specific time of the ERLE of the adaptive filter, for example, according to the formula shown in the process 1100 of FIG. (Process 1110). ERLEave is provided to the evaluation value adjustment process 270 (process 1120).
The evaluation value addition / subtraction process is the same as the process of the first embodiment shown in FIG. 11, but the process 910 is different. In this embodiment, the filter monitoring unit 107 determines whether or not ERLEave is equal to or less than a predetermined threshold value. If the threshold value is equal to or less than the threshold value, the filter monitoring unit 107 proceeds to process 920. When ERLEave is equal to or less than a predetermined threshold, it is determined that the amount of echo cancellation is not sufficient, and the evaluation value E of the delay time can be reduced.
According to the echo canceller and the echo cancellation method according to each of the embodiments described above, the delay time of the echo signal is accurately determined in a short time even in an environment where the delay time of the echo signal is unknown or the delay time is likely to change. It is possible to achieve stable echo cancellation

(Modification)
FIG. 15 is a diagram showing a modification of the present embodiment.
In addition to being connected to a telephone, the echo canceller 10 may be used in, for example, a telephone conference system (speaker phone) including a microphone 31 and a speaker 32. In this case, the transmission signal Sin from the near-end telephone in each of the above-described embodiments corresponds to the voice signal of the speaker input via the microphone 31 in this modification. In addition, the reception signal Rout to the near-end telephone in each of the above-described embodiments corresponds to an audio signal output from the speaker 32 in this modification.
In addition, this invention is not limited to an above-described Example, Various modifications are included. For example, the above-described embodiments have been described in detail for easy understanding of the present invention, and are not necessarily limited to those having all the configurations described. Further, a part of the configuration of one embodiment can be replaced with the configuration of another embodiment, and the configuration of another embodiment can be added to the configuration of one embodiment. Further, it is possible to add, delete, and replace other configurations for a part of the configuration of each embodiment.

Each of the above-described configurations, functions, processing units, processing means, and the like may be realized by hardware by designing a part or all of them with, for example, an integrated circuit. Each of the above-described configurations, functions, and the like may be realized by software by interpreting and executing a program that realizes each function by the processor. Information such as programs, tables, and files for realizing each function can be stored in a memory, a hard disk, a recording device such as an SSD (Solid State Drive), or a recording medium such as an IC card, an SD card, or a DVD.
Further, the control lines and information lines indicate what is considered necessary for the explanation, and not all the control lines and information lines on the product are necessarily shown. Actually, it may be considered that almost all the components are connected to each other.

  The present invention is applicable to, for example, a communication system, a conference system via a communication network, and the like.

DESCRIPTION OF SYMBOLS 10: Echo canceller 100: Buffer processing part 101: Reception buffer 102: Transmission buffer 103: Correlation processing part 104: Statistical processing part 105: Management table 106: Adaptive filter 107: Filter monitoring part 108: Delay processing part 110: Data line 120 : Data line 130: Echo canceler

Claims (7)

An echo canceler for canceling an echo signal included in the transmission signal to the far end side using an adaptive filter;
Correlation for obtaining a plurality of delay time candidates for a received signal based on the cross-correlation by obtaining one of the received signal from the far end and the transmission signal to the far end while shifting one signal at a time. A processing unit;
A statistical processing unit for obtaining an evaluation value for a delay time candidate based on the obtained cross-correlation history;
A storage unit for storing the evaluation value obtained by the statistical processing unit in correspondence with the delay time;
A delay processing unit that selects one of delay time candidates based on the evaluation value stored in the storage unit, delays a received signal according to the selected delay time, and outputs the delayed signal to the adaptive filter;
An echo canceller comprising: a filter monitoring unit that monitors a filter coefficient of the adaptive filter or an echo cancellation amount in the echo cancellation unit, and changes an evaluation value stored in the storage unit according to a monitoring result.   The filter monitoring unit determines whether or not the result of the convolution addition by the adaptive filter has a desired impulse response shape based on the filter coefficient of the adaptive filter. If the result is not the desired shape, the filter monitoring unit stores the result in the storage unit. The echo canceller according to claim 1, which reduces an evaluation value.   The filter monitoring unit determines that the peak area of the filter coefficient of the adaptive filter is not in the shape of a desired impulse response unless the peak area of the filter coefficient is in the first half of all filter coefficients, and reduces the evaluation value stored in the storage unit. Item 3. The echo canceller according to Item 2.   The filter monitoring unit determines that the peak value of the filter coefficient of the adaptive filter is not a desired impulse response shape if the peak value of the filter coefficient of the adaptive filter is equal to or less than a predetermined threshold, and reduces the evaluation value stored in the storage unit. Item 3. The echo canceller according to Item 2.   The filter monitoring unit obtains an index indicating an echo cancellation amount in the echo cancellation unit, and if the index or a statistical value based on the index is equal to or less than a predetermined threshold, the evaluation value stored in the storage unit is obtained. The echo canceller according to claim 1 to be reduced.   The echo canceller according to claim 1, wherein the statistical processing unit obtains an evaluation value for a delay time candidate by integrating weights according to the obtained cross-correlation. An echo cancellation method for canceling an echo signal included in a transmission signal to a far end side using an adaptive filter,
One of the received signal from the far end side and the transmission signal to the far end side is shifted by a predetermined time to obtain the cross-correlation of both signals, and based on the cross-correlation, a plurality of delay time candidates for the received signal are obtained,
Based on the obtained cross-correlation history, obtain an evaluation value for the delay time candidate,
The obtained evaluation value is stored in correspondence with the delay time,
Select one of the delay time candidates based on the stored evaluation value, delay the received signal according to the selected delay time, and output to the adaptive filter,
An echo cancellation method for monitoring a filter coefficient or an echo cancellation amount of an adaptive filter and reducing an evaluation value stored according to a monitoring result.

Images