JP3407392B2 - Stereo echo canceller - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a stereo echo canceller for canceling echo of stereo channels in a teleconference system or the like.

[0002]

2. Description of the Related Art In recent years, a teleconference system has been put to practical use, and attempts have been made to further improve the sense of presence by converting a voice into stereo and transmitting and spreading the voice. When performing stereo loudspeaking, a monaural acoustic echo canceller cannot sufficiently cancel echoes, so that a stereo echo canceller is desired to be put into practical use. To meet such a demand, a stereo echo canceller in which adaptive filters are linearly combined has been proposed. For example, B.Widr
ow, et al .: Adaptive noise cancelling: Principle an
d applications, Proc. IEEE, 63, 12, pp1692-1716,
An example was published in (1975).

The above-mentioned stereo echo canceller will be described with reference to the drawings. FIG. 4 is a block diagram showing the overall configuration of a conventional stereo echo canceller.
In the figure, input terminals 1 and 2 are input portions for audio signals output from first (left) and second (right) microphones installed in an acoustic space at a remote location. When the subscript j is a sample number or a sampling point, the first reception signal x1j transmitted to the input end 1 is power-amplified and given to the output end 3, and the second reception signal x2j transmitted to the input end 2. Is amplified in power and given to the output terminal 4. First and second speakers (not shown) are connected to the output terminals 3 and 4, respectively. The first and second speakers are the left and right speakers installed in the acoustic space at this point.

Microphones (not shown) are installed on the left and right of the acoustic space at this point. The respective audio signals input to these microphones are given to the input terminals 5 and 6 as the first transmission input y1j and the second transmission input y2j. Further, the output terminals 7 and 8 are output units for transmitting the local audio signal to a remote location as a first output signal e1j and a second transmission output e2j, respectively.

Now, the first adaptive filter 4 is provided at the input end 1.
The first and third adaptive filters 43 are connected, and the input terminal 2 is connected to the second adaptive filter 42 and the fourth adaptive filter 44. The adaptive filter 41 is a filter that adaptively estimates the acoustic impulse response from the speaker at the output end 3 to the microphone at the input end 5, and the adaptive filter 42.
Is a filter for adaptively estimating the acoustic impulse response from the speaker at the output end 4 to the microphone at the input end 5. Similarly, the adaptive filter 43 is a filter that adaptively estimates the acoustic impulse response from the speaker at the output end 3 to the microphone at the input end 6, and the adaptive filter 44 changes from the speaker at the output end 4 to the microphone at the input end 6. It is a filter that adaptively estimates the reaching acoustic impulse response.

The update gain calculating means 417 is for calculating the update gains of the adaptive filter 41 and the adaptive filter 42. In addition, the update gain calculation means 418 is configured by the adaptive filter 4
3 and the update gain of the adaptive filter 44 are calculated.

The first subtraction means 45 includes an adaptive filter 41.
In the circuit for subtracting the echo replica yh11j generated in (1) and the echo replica yh12j generated in the adaptive filter 42 from the transmission input y1j, the subtraction result is output from the output end 7 as the transmission output e1j. The second subtraction means 46 is
Echo replica yh21j generated by the adaptive filter 43
And the echo replica yh generated by the adaptive filter 44.
22j is a circuit for subtracting 22j from the transmission input y2j, and the subtraction result is output from the output terminal 8 as the transmission output e2j.

FIG. 5 is a block diagram showing a concrete configuration of the adaptive filters 41 and 42, the subtracting means 45, and the update gain calculating means 417. In FIG. 5, the update gain calculation means 417 of FIG. 4 corresponds to the update gain calculation means 510. In FIG. 5, the input section 1 of the first X register 51 corresponds to the input terminal 1 of FIG. X register 5
1 samples the received signal x1j at, for example, 8 kHz,
(All the registers below shall sample the input signal at 8 kHz) Received signal sequence X1j = {x1j, x1j-1, ... Of the past N samples at sample point j
., X1j-N + 1}.

The input section 2 of the second X register 52 corresponds to the input terminal 2 of FIG. The X register 52 receives the received signal x
2j is sampled at 8 kHz, and the received signal sequence X2j = {x2j, x2j-of the past N samples at the sampling point j
It is a register for storing 1, ..., X2j-N + 1}. The first H register 53 estimates the impulse response H11j = {h110j, h111j, ..., H11N-1 of the echo path from the speaker at the output end 3 to the microphone at the input end 5.
This is a register that stores j}. Second H register 54
Is the estimated value H12j = {h1 of the impulse response of the echo path from the speaker at the output end 4 to the microphone at the input end 5
20j, h121j, ..., H12N-1j}.

The output of the X register 51 and the output of the H register 53 are given to the first product-sum means 55. Accumulating means 55
Is N of X register 51 and H register 53 for each sample.
Echo replica yh11
This is a circuit that synthesizes j. The output of the X register 52 and the output of the H register 54 are given to the second product-sum means 56. The sum-of-products means 56 and the X register 52 and H for each sample.
This is a circuit for performing a convolution operation on N pieces of data in the register 54 to synthesize an echo replica yh12j.

The outputs of the reception power calculation means 58 and 59 are given to the update gain calculation means 510. Update gain calculation means 5
10 divides the constant α (0 <α ≦ 1) by the sum of the received power P1j and the received power P2j to obtain the first and second values.
Is a circuit for outputting the update gain σ1j to the updating means 511, 512 of FIG.

The first updating means 511 multiplies the output signal e1j of the subtracting means 57 by the updating gain σ1j, and the multiplication result σ
1j · e1j is converted to N data of the X register 51 {x1
j, x1j-1, ..., X1j-N + 1}. The calculation result here is added to the H register 53, and the estimated value H11j of the impulse response previously held in the H register 53 is updated.

The second updating means 512 multiplies the output signal e1j of the subtracting means 57 by the updating gain σ1j, and the multiplication result σ
1j · e1j is converted to N pieces of data in the X register 52 {x2
j, x2j-1, ..., X2j-N + 1}. The calculation result here is added to the H register 54, and the estimated value H12j of the impulse response previously held in the H register 54 is updated.

The configurations of the adaptive filters 43 and 44, the subtracting means 46, and the update gain calculating means 418 shown in FIG. 4 are the same as those in FIG. Hereinafter, the estimated value of the impulse response stored in the H register of the adaptive filter 43 is H21j, and the estimated value of the impulse response stored in the H register of the adaptive filter 44 is H22j. Further adaptive filter 4
The echo replicas synthesized in 3 and 44 are yh21j and yh2.
2j and the second transmission output is e2j.

The operation of the conventional stereo echo canceller configured as described above will be described using mathematical expressions. The updating means 511 in FIG. 5 inputs the received signal sequence X1j from the X register 51 and estimates a new impulse response by the normalized stereo echo canceller algorithm shown in the following (Equation 1).

[0016]

[Equation 1]

The impulse response estimated value H11j = {h110j, h111j, ..., H11N-1j} obtained in (Equation 1) is set to H.
Store in register 53.

Next, the sum of products means 55 calculates the echo replica yh11j using the following (Equation 2).

[0019]

[Equation 2]

On the other hand, the received power calculation means 58, 59 uses the following (Equation 3) and (Equation 4) to obtain the first received power P1j.
And the second received power P2j, respectively.

[0021]

[Equation 3]

[0022]

[Equation 4]

These values are input to the update gain calculation means 510, and the update gain σ1j is calculated using the following (Equation 5).

[0024]

[Equation 5]

In this way, the adaptive filter 41 shown in FIG.
The echo replica yh11j generated in (4) is output to the subtracting means 45.

Similarly, the adaptive filter 42 calculates each coefficient of the impulse response H12j of the echo path from the speaker at the output end 4 to the microphone at the input end 5 using the following (Equation 6).

[0027]

[Equation 6]

Then, the echo replica yh12j is generated using the following (Equation 7).

[0029]

[Equation 7]

Next, the subtracting means 45 includes an adaptive filter 41,
The echo replica generated at 42 is converted into the transmission output e1j using the following (Equation 8).

[0031]

[Equation 8]

When a sufficient amount of time has passed since the first received signal was given, the sum of the echo replicas synthesized by the adaptive filters 41 and 42 becomes substantially equal to the transmission input y1j. Therefore, the echo component of the transmission output e1j becomes almost zero, and only the voice signal of the speaker at this point is transmitted via the output end 7 to the listener at the remote point.

Similarly for the second transmission output e2j,
It is generated by the adaptive filters 43 and 44 and the subtraction means 46. That is, signal processing is performed using the following (Equation 9) to (Equation 14).

[0034]

[Equation 9]

[0035]

[Equation 10]

[0036]

[Equation 11]

[0037]

[Equation 12]

[0038]

[Equation 13]

[0039]

[Equation 14]

However, the adaptive filters 41 and 42 respectively adjust the filter coefficients of the adaptive filters so that the root mean square of the transmission output e1j is minimized, and the adaptive filters 43 and 44 are the root mean square of the transmission output e2j. For updating, when there is a correlation between the received signal x1j and the received signal x2j, each adaptive filter may not converge to the impulse response to be estimated. For example, the impulse response sequence of the echo path from the speaker at the output end 3 to the microphone of the input end 5 is G11, and the impulse response sequence of the echo path from the speaker at the output end 4 to the microphone at the input end 5 is G12. Also, the impulse response sequence of the echo path from the speaker at the output end 3 to the microphone at the input end 6 is set to G21, and the speaker at the output end 4 is connected to the input end 6
Let G22 be the impulse response sequence of the echo path leading to the microphone.

Next, the adaptive filters 41, 42, 43, 4
The impulse response sequences estimated by 4 are H11, H12, H
If it is 21, H22, H11 = G11, H12 = G12, H21 = G21, H22 = G22 under the condition that the adaptation is sufficiently advanced.
The impulse response of the adaptive filter should converge so that However, if the received signal x1j and the received signal x2j are exactly the same signal, that is, a signal having a very strong correlation, H11
= H12 = (G11 + G12) / 2, H21 = H22 = (G21 + G
Even if it converges to 22) / 2, both the echo components of the transmission outputs e1j and e2j can be made zero.

In this way, when there is a correlation between the first received signal and the second received signal, each adaptive filter may erroneously converge on an impulse response different from the original one. When the degree of correlation between received signals changes midway due to a change or movement of the speaker in a state where the erroneous impulse response is converged, the filter coefficient of the adaptive filter is further updated from that point. Then, in the adaptive filter having the received signal with the smaller short-time power average of the received signal as the input, the filter coefficient of the adaptive filter having the impulse response already correctly estimated is disturbed, and as a result, the residual echo increases, In addition, there is a problem that the echo canceling speed becomes slow.

The present invention has been made in view of the above conventional problems, and it is a stereo echo which prevents a decrease in echo canceling speed caused by a change of speakers and suppresses an increase in residual echo caused thereby. The purpose is to realize a canceller.

[0044]

In order to achieve the above object, the stereo echo canceller of the first aspect of the present invention is a stereo echo canceller in which a microphone and a speaker are provided at a first position and a second position, respectively. A transmission signal and a reception are provided between the space and the other acoustic space which is located apart from the one acoustic space, and in which the microphone and the speaker are respectively provided at the third position and the fourth position. A stereo echo canceller used in a teleconference system for transmitting and receiving stereo audio information using a signal, comprising:
The first reception signal output from the microphone at the position is adaptively set to a tap coefficient equivalent to the acoustic impulse response when the first reception signal is input to the microphone at the first position via the speaker at the first position. Sound when the first adaptive filter and the second received signal output from the microphone at the fourth position are input to the microphone at the first position via the speaker at the second position A second adaptive filter adaptively having a tap coefficient equivalent to an impulse response, and a first reception signal output from the microphone at the third position are transmitted to the second speaker via the speaker at the first position. A third adaptive filter adaptively having a tap coefficient equivalent to an acoustic impulse response when being input to the microphone at the position 4; and the microphone at the position 4 A fourth adaptation adaptively having a tap coefficient equivalent to an acoustic impulse response when the second received signal output is input to the microphone at the second position via the speaker at the second position. A filter and a voice signal of the microphone at the first position, which subtracts the echo replicas generated by the first and second adaptive filters, respectively, and generates a first transmission output in which echo components are canceled. 1 and the echo signals generated by the third and fourth adaptive filters are subtracted from the voice signal of the microphone at the second position to obtain the second transmission output in which the echo component is canceled. Second subtracting means for generating, first updating gain calculating means for calculating updating gains of the first and second adaptive filters, and updating gains of the third and fourth adaptive filters Second update gain calculating means for calculating, and speaker change detecting means for inputting the first and second received signals and detecting the change of the speaker in the other acoustic space based on the input signals. First and second short-time power average calculating means for calculating short-time power averages of the first and second received signals, respectively, and magnitudes of output signals of the first and second short-time power average calculating means And the speaker change detecting means detects the change of the speaker in the other acoustic space, the small and large comparing means determines that the short-time power average of the first received signal is smaller . If it is determined, the update gain of the first adaptive filter is set to 0, and the short-time power average of the second received signal is smaller in the magnitude comparing means.
A first adaptive control means for zero updating gain of the second adaptive filter when it is determined that have, upon detecting a change of the talker at the other of the acoustic space in the speaker alternation detecting means, When the magnitude comparison means determines that the short-time power average of the first received signal is smaller , the update gain of the third adaptive filter is set to 0, and the magnitude comparison means performs the second reception. When it is determined that the short-time power average of the signal is smaller , the second adaptive control means for setting the update gain of the fourth adaptive filter to 0 is provided.

Further, a stereo-e canceller of the second aspect of the invention comprises the speaker change detecting means of the first aspect of the invention as the first and second aspects.
Using the output of the short-time power average calculation means of the division means for calculating the division value β, and the division value β of the division means
And two predetermined constants β1 and β2 (where β1> 1
> β2) and a size comparison means for comparing the size relationship with β2), and a state change detection means for detecting the output of the size comparison means and determining that there is a change in speaker if the output changes. is there.

Further, in addition to the configuration of the first invention, the stereo echo canceller of the third invention has the first transmission input and the first transmission input from the microphone at the first position.
First echo suppression amount calculation means for calculating the short-time power average ratio of the first transmission output output from the subtraction means, the second transmission input input from the microphone at the second position, and the second transmission input. A second echo suppression amount calculation unit for calculating a short-time power average ratio of the second transmission output output from the second subtraction unit, and the first adaptive control unit is configured to provide the first echo suppression amount. If the echo suppression amount calculated by the calculating means is larger than a preset value, the updating gains of the first and second adaptive filters are controlled so as not to be 0 even when the speaker is changed, and the second adaptive control means is , The second
If the echo suppression amount calculated by the echo suppression amount calculation means is larger than a preset value, the updating gains of the third and fourth adaptive filters are controlled so as not to be 0 even when the speaker changes. To do.

[0047]

According to the first and second aspects of the invention having such characteristics, the first adaptive filter estimates the impulse response of the echo path from the first reception output to the first transmission input and Synthesize a replica. The second adaptive filter estimates the impulse response of the echo path from the second reception output to the first transmission input, and synthesizes the echo replica. Then, the echo component contained in the first transmission input is canceled by subtracting the echo replica synthesized by the first and second adaptive filters from the first transmission input by the first subtraction means. Similarly, the second adaptive filter
The impulse response of the echo path from the reception output to the first transmission input is estimated, and the echo replica is synthesized.
The fourth adaptive filter estimates the impulse response of the echo path from the second reception output to the second transmission input and synthesizes the echo replica thereof. And the second subtraction means
The echo component included in the second transmission output is canceled by subtracting the echo replicas synthesized by the third and fourth adaptive filters from the transmission input of. Also, the first and second
By observing the received signal of, the speaker change detecting means detects the change of the speaker in the other acoustic space, and at that time, the magnitude comparing means determines that the short-time power average of the first received signal is larger. In this case, the update gain of the second adaptive filter is set to 0 to stop the adaptation, and when the magnitude comparing means determines that the short-time power average of the second received signal is larger, the first adaptation is performed. By stopping the adaptation by setting the update gain of the filter to 0, it is possible to prevent the coefficient of the adaptive filter from being erroneously updated at the time of speaker change, which causes an increase in residual echo and a decrease in the convergence speed of the filter coefficient. Can be prevented.

Further, according to the third invention, in addition to the actions of the first and second inventions, it is determined whether the filter coefficient of the adaptive filter converges to a true impulse response.
Judgment is made from the echo suppression amount calculated by the echo suppression amount calculation means, and if the impulse response is close to the true impulse response, the update gain of the adaptive filter is not set to 0 even when the speaker is changed, and the adaptation is continued. This makes it possible to prevent the convergence speed from decreasing.

[0049]

DESCRIPTION OF THE PREFERRED EMBODIMENTS A stereo echo canceller according to a first embodiment of the present invention will be described with reference to FIG.

FIG. 1 is a block diagram showing the overall configuration of the stereo echo canceller in the first embodiment. The same parts as those in the conventional example are designated by the same reference numerals, and detailed description thereof will be omitted. In the figure, input terminals 5 and 6 are input terminals for inputting audio signals of microphones respectively provided at a first position (left) and a second position (right) in one (medium) acoustic space. The output terminals 3 and 4 are the output terminals of the audio signal given to the speakers provided at the first and second positions of the one acoustic space. Similarly, the input terminals 1 and 2 are the other (remote point)
Of the microphones provided at the third position (left) and the fourth position (right) of the acoustic space. The output terminals 7 and 8 are the output terminals of the audio signal given to each speaker provided in the 3rd and 4th position of one acoustic space. And like the conventional example, the first to fourth
Adaptive filters 11 to 14, first and second subtraction means 1
5, 16 are provided respectively.

Unlike the conventional example, in this embodiment, the input terminals 1,
A speaker change detecting means 17 is connected between the two. The speaker change detecting means 17 is a circuit which receives the received signals x1j and x2j and detects the change of the speaker at the remote location from the change in the signal level. The first short-time power average calculating means 18 is provided at the input end 1, and the second short-time power average calculating means 1 is provided at the input end 2.
9 are connected to each other, and the result is given to the magnitude comparison means 110. The magnitude comparison means 110 compares the short-time power average values P1j and P2j of the first and second received signals calculated by the first and second short-time power average calculation means 18 and 19, and determines the magnitude thereof. . First adaptive control means 11
1, the speaker change detecting means 17 detects the change of the speaker at the remote location, and the magnitude comparing means 110 makes the first change.
The short-term power average of the received signal is smaller (P1j <
P2j), the first adaptive filter 1
The adaptation is stopped by setting the update gain of 1 to 0 for a certain period of time, and when the magnitude comparison means 110 determines that the short-time power average of the second received signal is smaller (P1j > P2j), the second The adaptation is stopped by setting the update gain of the adaptive filter 12 of 0 to 0 for a certain period.

For example, when the speaker changes from the left (first position side) to the right (second position side) at a remote point while each adaptive filter does not converge to the true impulse response, Since the correlation between the first and second received signals changes significantly when the person changes, an adaptive filter (the first adaptive filter in this example) that receives the received signal with the smaller short-time power average (X1j in this example) is input. 11 and the filter coefficients of the third adaptive filter 13) are erroneously updated. Then, the convergence speed of the filter coefficient decreases and the residual echo increases. Therefore, by changing the update gain of the adaptive filter that receives the received signal having the smaller short-time power average value as the input during the speaker change to a fixed period of 0 and stopping the adaptation, the filter coefficient is updated to an incorrect filter coefficient during the speaker change. It is possible to prevent an increase in residual echo and a decrease in the convergence speed of the filter coefficient, which are caused thereby.

As described above, according to the present embodiment, when the speaker is changed, the update gain of the adaptive filter which receives the received signal having the smaller short-time power average is set to 0 for a certain period of time to stop the adaptation. It is possible to prevent an incorrect filter coefficient from being updated at the time of a person change, and to prevent a decrease in the convergence speed of the filter coefficient and an increase in residual echo.

Next, a stereo echo canceller according to the second embodiment of the present invention will be described with reference to FIG. The present embodiment embodies the speaker change detecting means in the first embodiment, and other configurations are the same as those in FIG. FIG. 2 is a block diagram showing the overall configuration of the stereo echo canceller in the second embodiment. In FIG. 2, the division means 21, the magnitude comparison means 22, and the state change detection means 23 constitute a speaker change detection means 24. ing. The same components as those in the first embodiment are designated by the same reference numerals, and detailed description thereof will be omitted.

In FIG. 2, the division means 21 divides the output of the first short-time power average calculation means 18 by the output of the second short-time power average calculation means 19, and the divided value β is sent to the magnitude comparison means 22. Output. The magnitude comparison means 22 uses this divided value β and two predetermined constants β1 and β2 (β1> 1).
> Β2) to determine in which range the divided value β falls. This determination result is given to the state change detecting means 23. The state change detecting means 23 checks whether or not the range of the division value β has changed, and if the range has changed, it is determined that the speaker has changed in the other acoustic space, and a speaker change detection signal is output.

The operation of the stereo echo canceller having the speaker change detecting means 24 thus constructed will be described. For example, it is assumed that each speaker is located on one of the left side (third position side), the center, and the right side (fourth position side). Then, β1 and β2 are set in advance so that β>β1> 1 when the left speaker speaks and 1>β2> β when the right speaker speaks. Then, first, when the speaker on the left side speaks, the first short-time power average P1j becomes larger than the second short-time power average P2j, so that the output β of the dividing means 21 becomes larger than β1.

When the speaker in the center speaks, β =
1, and β1>β> β2 holds. Similarly, when the speaker on the right side speaks, β <β2. In this way, the relationship between β and β1 and β2 is determined by the position of the speaker, and the change of speaker can be detected by detecting the boundary of the change in the state of β by the state change detecting means 23.

As described above, according to the present embodiment, the change of the speaker is detected by observing the change in the division value β of the short-time power averages of the reception signals x1j and x2j. Then, by stopping the adaptation by setting the update gain of the adaptive filter that receives the received signal with the smaller short-time power average value to zero for a certain period of time, it is prevented that the wrong filter coefficient is updated when the speaker changes. It is possible to prevent the convergence speed of the filter coefficient from decreasing and the residual echo from increasing.

Next, a stereo echo canceller according to the third embodiment of the present invention will be described with reference to FIG. This embodiment has a first configuration (FIG. 1) of the first embodiment.
And second echo suppression amount calculation means are added, and the first and second adaptive control means are correspondingly changed. FIG. 3 is a block diagram showing the overall configuration of the stereo echo canceller of the third embodiment. The same components as those of the first embodiment are designated by the same reference numerals and detailed description thereof will be omitted.

In FIG. 3, the first echo suppression amount calculation means 31 has a first transmission input y1j and a first transmission output e1j.
Is a circuit for calculating the echo suppression amount ERLE1j in the first transmission output by calculating the short-time power average ratio with the second echo suppression amount calculation means 32.
Is a circuit for calculating the echo removal amount ERLE2j in the second transmission output by calculating the short-time power average ratio between the transmission input y2j and the second transmission output e2j. If each adaptive filter is close to the true impulse response, the convergence speed may decrease if the adaptation of the adaptive filter is stopped when the speaker changes. Therefore, the threshold γ is set in advance, and the ER
If LE1j is larger than γ, it is the first even when the speaker changes.
And the second adaptive filters 11 and 12 are not made to have an update gain of 0, and when the ERLE2j is larger than γ, the update gains of the third and fourth adaptive filters 13 and 14 are not made to be 0. It is possible to prevent the convergence speed from decreasing.

As described above, according to this embodiment, the threshold value γ is set in advance, and if the ERLE1j is larger than γ, the update gains of the first and second adaptive filters 11 and 13 are not set to 0, and ERLE2j is set. Is larger than γ, it is possible to prevent the convergence speed from decreasing by not setting the update gains of the third and fourth adaptive filters 13 and 14 to zero.

[0062]

As described above, according to the first and second aspects of the present invention, by providing the speaker change detecting means, the change of the speaker in the other acoustic space from the first and second received signals is performed. Can be detected. In this case, the adaptive filter that receives the received signal with the smaller short-time power average value as an input has an update gain of 0 for a certain period of time to stop adaptation, so that an incorrect filter coefficient is updated when the speaker changes. It is possible to prevent the convergence speed of the filter coefficient from decreasing and the residual echo from increasing. Therefore, when conducting a joint conference or the like at two different points using the stereo sound system, an increase in echo generated between the microphone and the speaker when the speaker changes or moves at another point can be quickly suppressed. The effect is demonstrated.

Further, according to the third aspect of the invention, in a situation where the coefficient of the adaptive filter is close to the true impulse response, it is possible to prevent the convergence speed from being lowered by stopping the adaptation. Therefore, when a joint conference or the like is held at two different points using the stereo sound system, the effect of promptly suppressing the echo generated between the microphone and the speaker is exerted.

[Brief description of drawings]

FIG. 1 is a block diagram showing an overall configuration of a stereo echo canceller according to a first embodiment of the present invention.

FIG. 2 is a block diagram showing the overall configuration of a stereo echo canceller according to a second embodiment of the present invention.

FIG. 3 is a block diagram showing the overall configuration of a stereo echo canceller according to a third embodiment of the present invention.

FIG. 4 is a block diagram showing a configuration example of a conventional stereo echo canceller.

FIG. 5 is a block diagram showing a specific configuration example of an adaptive filter, subtraction means, and update gain calculation means used in a conventional stereo echo canceller.

[Explanation of symbols]

1, 2, 5, 6 input end 3, 4, 7, 8 output end 11-14, 51-54, 71-74 Adaptive filter 15, 16, 45, 46, 57 Subtracting means 17 Speaker change detection means 18, 19 Short-time power average calculation means 21 Division means 22 Large and small comparison means 23 State change detection means 31, 32 Echo suppression amount calculation means 51,52 X register 53, 54 H register 55,56 Sum of products means 57 Subtraction means 58,59 Received power calculation means 110 size comparison means 111 Adaptive control means 112, 113, 33, 34 Adaptive control means 417, 418, 510 Update gain calculation means 511, 512 update means

─────────────────────────────────────────────────── ─── Continuation of front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) H04B 3/00-3/44

Claims (3)

(57) [Claims] 1. An acoustic space in which a microphone and a speaker are provided at a first position and a second position, respectively, and an acoustic space located apart from the acoustic space and a third position. A stereo echo canceller used in a teleconference system that transmits and receives stereo audio information using a transmission signal and a reception signal between the other acoustic space in which a microphone and a speaker are respectively provided at a fourth position. And a tap equivalent to an acoustic impulse response when the first reception signal output from the microphone at the third position is input to the microphone at the first position via the speaker at the first position. A first adaptive filter that adaptively has a coefficient and a second received signal output from the microphone at the fourth position are passed through a speaker at the second position. Second adaptive filter adaptively having a tap coefficient equivalent to the acoustic impulse response when input to the microphone at the first position, and a first received signal output from the microphone at the third position Is a third adaptive filter that adaptively has a tap coefficient equivalent to an acoustic impulse response when input to the microphone at the second position via the speaker at the first position, and the fourth position. A second reception signal output from the microphone of ## EQU1 ## having a tap coefficient equivalent to an acoustic impulse response when the second reception signal is input to the microphone of the second position via the speaker of the second position. 4 and the echo signals generated by the first and second adaptive filters are respectively subtracted from the voice signal of the microphone at the first position. , A first subtraction unit for generating a first transmission output in which echo components are canceled, and an echo replica generated by the third and fourth adaptive filters from the voice signal of the microphone at the second position. Second subtraction means for respectively subtracting to generate a second transmission output in which echo components are canceled, first update gain calculation means for calculating update gains of the first and second adaptive filters, Second update gain calculation means for calculating the update gains of the third and fourth adaptive filters, and the speaker in the other acoustic space, which is inputted with the first and second received signals, Speaker change detecting means for detecting the change of the signal, first and second short time power average calculating means for calculating the short time power averages of the first and second received signals, respectively, and the first and second Short-time power average calculation The magnitude comparison means for comparing the magnitudes of the output signals of the stages, and the speaker change detection means detects the change of the speaker in the other acoustic space, the short time power of the first received signal in the magnitude comparison means. When it is determined that the average is smaller , the update gain of the first adaptive filter is set to 0, and when the magnitude comparison means determines that the short-time power average of the second received signal is smaller. First adaptive control means for setting the update gain of the second adaptive filter to 0, and when the speaker change detection means detects a change of the speaker in the other acoustic space, the magnitude comparison means performs the change. When it is determined that the short-time power average of the first received signal is smaller , the update gain of the third adaptive filter is set to 0, and the magnitude comparison means sets the second received signal to the second received signal. A stereo echo canceller, comprising: second adaptive control means for setting the update gain of the fourth adaptive filter to 0 when it is determined that the short-time power average is smaller . 2. The speaker change detecting means, using the outputs of the first and second short time power average calculating means, a dividing means for calculating a divided value β thereof, and a divided value β of the dividing means, Two types of constant β
1 and β2 (however, β1>1> β2) and a magnitude comparison means for comparing the magnitude relationship with each other, and a state change detection means for detecting the output of the magnitude comparison means and judging that there is a speaker change if the output changes. The stereo echo canceller according to claim 1, further comprising: 3. A first echo suppression for calculating a short-time power average ratio of a first transmission input input from a microphone at a first position and a first transmission output output from a first subtraction means. A second echo for calculating a short-time power average ratio of the amount calculation means, the second transmission input input from the microphone at the second position, and the second transmission output output from the second subtraction means. Suppression amount calculation means is further provided, and if the echo suppression amount calculated by the first echo suppression amount calculation means is larger than a preset value, the first adaptive control means can perform the first and the second operations. The update gain of the second adaptive filter is controlled so as not to be 0, and the second adaptive control means talks if the echo suppression amount calculated by the second echo suppression amount calculation means is larger than a preset value. Social interaction Stereo echo canceller of claim 1, wherein characterized in that it also controls the third and fourth update gain of the adaptive filter so as not to 0 at the time.