JP2002261660A - Multi-channel echo canceling method, its apparatus, its program and its recording medium - Google Patents

Abstract

(57) [Problem] To increase the convergence speed even when there is a difference in power between stereo signals and cross-correlation is strong. A received signal _{x 1 (k), x 2} 2 sum ‖X each predetermined time (k) _'1 (k) ‖ _^2, ‖x' ² (k) ‖ ²
(32 ₁ , 32 ₂ ), and, for example, ‖
For x ′ ₁ (k)}, μ ₁ = {x ′ ₁ (k)} /
√ (‖x ′ ₁ (k) ‖ ² + ‖x ′ ₂ (k) ‖ ² )
On the other hand, the reverberation path estimators 19 ₁ and 19 ₂ set the μ ₂ as the update step size and use the received signals x ₁ (k) and x ₂ (k) and the residual reverberation signal e (k) to generate pseudo echo paths 18 ₁ and 18. Compute the correction vector for ₂ .

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a multi-channel echo canceling method for erasing an indoor echo signal which causes howling and impairs hearing in, for example, a communication conference system having a multi-channel reproducing system, a multi-channel echo canceling method, an apparatus therefor, and a program therefor. And its recording medium.

[0002]

2. Description of the Related Art In order to provide a loudspeaker communication system which is excellent in simultaneous call performance and has little reverberation, there is an echo canceller.
First, an echo canceling method and device configuration of the echo canceling device for one channel will be described with reference to FIG. In a loudspeaker call, a reception signal obtained by the other party's utterance or the like at the reception terminal 11 is reproduced from the speaker 12 as it is, or before being transmitted to the speaker 12, it is automatically transmitted depending on the amplitude, power, etc. of the reception signal. In some cases, the received signal is reproduced from the speaker 12 after some processing has been performed on the received signal, such as by adjusting the gain. Therefore, in this specification, the reception signal x ₁ (k) is not limited to the reception signal itself from the other party, and refers to the reception signal after the reception signal is processed when the reception signal is processed. Shall be. k represents a discrete time. The echo canceller 14 cancels the echo signal y ₁ (k) obtained by collecting the received signal x ₁ (k) from the speaker 12 via the echo path 15 and the microphone 16. Here, the reverberation signal y ₁ (k) is represented by y ₁ (k) = Σh ₁₁ (k, n) × ₁ (k−n) where h ₁₁ (k, n) is the impulse response of the reverberation path 15 at time k. (1) Σ can be modeled as one obtained by a convolution operation such as n = 0 to L−1. L is the number of taps and is a constant that is set in advance so as to correspond to the reverberation time of the reverberation path 15. First, the reception signal accumulation / vector generation unit 17
, The received signal x ₁ (k) is stored up to the past L-1 times. The stored signal is a received signal vector x ₁ (k), that is, x ₁ (k) = [x ₁ (k), x ₁ (k−1),..., X ₁ (k−L + 1)] ^T (2 ) Is output as. Here, * ^T represents transposition of a vector. The pseudo echo signal generation unit 18 calculates the inner product y 積₁ (k) of the reception signal vector x ₁ (k) of Expression (2) and the pseudo echo vector h ＾ ₁₁ (k) obtained from the echo path estimation unit 19. = h ^ ₁₁ performs _{^{T (k) x 1 (k}} ) (3), as a result, it generates an estimated echo signal y ^ ₁ (k). This inner product operation is equivalent to a convolution operation such as Expression (1). The echo path estimating section 19 generates the pseudo echo path vector h ＾ used in the pseudo echo signal generation section 18.
₁₁ (k) is generated. The most common algorithm used for this echo path estimation is the NLMS algorithm (learning identification method). In the NLMS algorithm, the received signal vector x ₁ (k) at time k and the residual echo signal e
₁ (k), that is, a signal e ₁ (k) = y ₁ (k) −y ＾ ₁ (k) obtained by subtracting the pseudo echo signal y ＾ ₁ (k) from the output y ₁ (k) of the microphone 16 in the difference circuit 21. From (4), the pseudo echo path vector h ＾ ₁₁ (k + 1) to be used at time k + 1 is obtained as in the following equation. _{h ^ 11 (k + 1)} = h ^ 11 (k) + μe 1 (k) x 1 (k) / (x 1 T (k) x 1 (k)) (5) where, mu is a step size parameter Called 0 <μ <
It is used for adjusting the adaptive operation in the range of 2. By repeating the above processing, the echo path estimating unit 19 gradually converts the pseudo echo path vector ＾ ₁₁ (k) into the true echo path 15.
The echo path vector h ₁₁ (k) having a time series of the impulse response h ₁₁ (k, n) of each as an element, that is, h ₁₁ (k) = [h ₁₁ (k, 0), h ₁₁ (k, 1) ,..., H ₁₁ (k, L-1)] ^T (6), and as a result, the residual echo signal e ₁ (k) of the equation (4) can be reduced.

In general, N (>2) Channel playback system and M
(>1) Communication conference system consisting of a channel sound collection system
Elimination of the echo in the case of the system
Is performed. That is, all N channels on the playback side and the
Process N-input 1-output time-series signal between each channel
N-channel echo canceller 22₁, 22_Two,…, 22_MTo
Realized as a connected echo cancellation system 23
You. In this case, N × M reverberation paths 15 in the entire system are used.
_nm(1<n<N, 1<m<M) exists. This system
Each of the N channels on the playback side and the
N-channel echo canceller connected between one channel
22₁, 22_Two,…, 22_MAbout the echo shown in Figure 4
The configuration of the erasing device 14 is expanded as shown in FIG.
Is done. This is, for example, IEICE Transactions '86 /
10 Vol. J69-A No. 10 "Multi-channel suitable
Digital filters ". here
And the sound collecting side is the m-th sound collecting channel (1<m<M) connected
N-channel echo canceller 22_mthink of. M-th
Channel microphone 16_mEcho signal collected by the
Indicates that the reception signal of each reproduction channel is
_1m~ 15_NmThrough the summation on the sound collection side
To estimate the echo path for
Also, the same one residual echo signal e_m(K) only
It is necessary to devise a way to evaluate it. First, each playback
Received signal accumulation / vector for received signal of channel
Generation unit (17₁, 17_Two,…, 17_N), The received signal
Kutor x₁(k) = [x₁(k), x₁(k-1),…, x₁(k-L₁ +1)]^T (7) x_Two(k) = [x_Two(k), x_Two(k-1),…, x_Two(k-L_Two +1)]^T (8): x_N(k) = [x_N(k), x_N(k−1),…, x_N(k−L_N+1)]^T (9) is generated. Where L₁, L_Two,…, L_NIs the number of taps, each
Echoway 15_1m, 15_2m, ..., 15_NmCorresponding to the reverberation time
Is a constant to be set in advance. These vectors
X (k) = [x₁ ^T(k), x_Two ^T(k),…, x_N ^T(k)]^T Combine with (10). In addition, the echo path estimation unit 19_mIn each
N echo paths between the playback channel and the m-th sound pickup channel
To simulate _1m(k),
h ＾_2m(k),…, h ＾_Nm(k) and h ＾_m(k) = [h ＾_1m ^T(k), h ＾_2m ^T(k),…, h ＾_Nm ^T(k)]^T (11) Pseudo echo path coupling vector h ＾_m(k)
Newly, when the NLMS algorithm is used, h ＾_m(k + 1) = h ＾_m(k) + μe_m(k) x (k) / (x^T(k) x (k)) (12). Simulated echo signal generator 18_mThen
Dot product operation y ＾_m(K) = h ＾_m ^T(k) x (k) (13), the reverberation signal y collected in the m-th sound collection channel
_mPseudo echo signal y ＾ for (k)_mGenerate (k)
You. In this way, the vectors for each channel are combined to form 1
By treating them as two vectors, the basic processing flow
This is similar to the one-channel echo canceller shown in FIG.
You.

In this multi-channel echo cancellation method,
If the average power of the received signals of the channels is different, the convergence speed
Degrades significantly. From this point, each of the channel methods
By normalizing the received signal for each
It is proposed to be good. Define the following vector
You. Z (k) = [x₁ ^T(K) / P₁(K), ..., x_N ^T(K) / P_N(K)] ^T (14) P_n(K) = x_n ^T(K) x_n(K) / L_n (15) P_n(K) is for normalizing power for each channel.
It is a normalization coefficient. Pseudo echo path coupling vector h
_mThe updating equation (12) of (k) is obtained by calculating the vector Z₁(K)
Can be expressed by the following equation.

H ＾_m(K + 1) = h ＾_m(K) + μe_m(K) Z (k) / (Z^T(K) x (k)) (16) Z^TIn (k) x (k), the following properties hold. Z^T(K) x (k) = Σ_{n = 1} ^NL_n (17) In this way, the received signal of each channel is normalized by power.
By doing so, the convergence speed can be improved. In this case, each pseudo
Echo path h ＾_1m, ..., h ＾_NmUpdate each time independently
Think. For simplicity of explanation, the steps shown in FIG.
Leo's two-channel reception signal and one-channel transmission signal
think of. That is, the reception signal x₁(K) is a pseudo echo path 18₁
Supplied to the pseudo echo signal h 信号₁ ^T(K) x
₁(K) is generated and the received signal x_Two(K) is pseudo echo path 1
8_TwoSupplied to the pseudo echo signal h ＾_Two ^T(K) x
_Two(K) is generated, and these pseudo echo signals are added to the addition circuit 2
5 and the added signal is a pseudo echo signal y ＾ (k)
From the sound pickup signal y (k) of the microphone 16
Then, a residual echo signal e (k) is output. This residue
Echo signal e (k) and received signal x₁(K) and the reverberation path
Estimator 19₁Speaker 12₁Between the microphone 16
Echo path h₁Is estimated, and the pseudo echo path 18₁Pseudo-resonance of
Road 18₁Characteristic h ＾₁Is updated. As well as residual echo
Signal e (k) and received signal x_Two(K) Echo path estimation
Part 19_TwoSpeaker 12_TwoBetween microphone and microphone 16
Road _TwoIs estimated, and the pseudo echo path 18_TwoCharacteristic h ＾_Two
Is updated.

The reverberation path estimator 19₁, 19_TwoResonance path in
Fixed, that is, each pseudo echo path 18₁, 18_TwoUpdate the characteristics of
It is obtained from equation (16). The vector Z (k) in this case
Is Z (k) = [x₁ ^T(K) / P₁(K), x_Two ^T(K) / P_Two(K)]^T (18) H ＾ in equation (16)_m(K), x
(K) and Z (k) are as follows. h ＾_m(K) = h ＾ (k) = [h ＾₁ ^T(K),
h ＾_Two ^T(K)]^Tx (k) = [x₁ ^T(K), x_Two ^T(K)]^TZ (k) = [x₁ ^T(K) / (x₁ ^T(K) x₁
(K) / L₁), X _Two ^T(K) / (x_Two ^T(K) x
_Two(K) / L_Two)]^T= [L₁x₁ ^T(K) / (x
₁ ^T(K) x₁(K)), L_Twox_Two ^T(K) / (x_Two ^T
(K) x_Two(K))]^T Denominator Z in equation (16)^T(K) x (k) is Z^T (K) x (k) = [L₁x₁ ^T(K) / (x₁
^T(K) x₁(K)), L_Twox_Two ^T(K) / (x
_Two ^T(K) x_Two(K))] [x₁(K), x
_Two(K)]^T= L₁+ L_Two Substituting this into equation (16) gives h ＾ (k + 1) =
h ＾ (k) + μe (k) Z (k) / (L₁+ L_Two)
Is

[0007]

(Equation 1)

Where h ＾ ₁ and h ＾ ₂ are respectively given by the following equations. _{h ^ 1 (k + 1)} = h ^ 1 (k) + μ (e (k) / (L 1 + L 2)) (L 1 x 1 (k) / (x 1 T (k) x 1 (k)) ( _{19) h ^ 2 (k +} 1) = h ^ 2 (k) + μ (e (k) / (L 1 + L 2)) (L 2 x 2 (k) / x 2 T (k) x 2 (k)) (20) x _n ^T (k) x _n (k) = {x _n (k)} ² , ‖x _n
(K) || ^{_{2 = x n 2 (k)}} + x n 2 (k-1) + ... +
x _n ² (k−L _n +1) and the number of taps is equal and L ₁ = L ₂
If _{h ^ 1 (k + 1)} = h ^ 1 (k) + 0.5μe (k) x 1 (k) / ‖x 1 (k) || _{^{2 (21) h ^ 2 (}} k + 1) = h ^ 2 ( k) +0.5 μe (k) × ₂ (k) / {x ₂ (k)} ² (22) In the echo path estimating units 19 ₁ and 19 ₂ , the expression (2)
The reverberation paths h ₁ (k),
By estimating h ₂ (k), the pseudo echo paths 18 ₁ and 18 ₂ are updated respectively. The second term on each right side of Expressions (21) and (22) is a vector of the received signals x ₁ (k) and x ₂ (k) normalized by power.

[0009]

As described above, by normalizing the received signal of each channel by its power, it is possible to prevent the convergence speed from being deteriorated even if the average power of the received signal of each channel is different. However, if there is a strong cross-correlation between the received signals, such as a stereo signal, even if there is a difference in the average power between the received signals of the two channels, the pseudo echo path does not converge to the characteristics of the true echo path. Are changed, for example, before the change, the average power of x ₁ (k) is x
_Although it was larger than the average power of ₂ (k), if the average power of x ₁ (k) became smaller than the average power of x ₂ (k) due to speaker change, reverberation would occur and convergence of the pseudo-echo path would occur. Takes a long time, for example, if the speaker alternates in about 2 seconds, the pseudo echo path does not converge. This is received 2
The same can be said not only for the multi-channel echo cancellation method of one channel and sound collection, but also for the general multi-channel echo cancellation method of N-channel reception and M sound collection channels.

An object of the present invention is to form a pseudo echo path relatively quickly even if there is a difference in the average power between received signals, the cross-correlation is relatively strong, and the average power ratio of the received signal changes. It is an object of the present invention to provide a multi-channel echo canceling method, an apparatus thereof, a program thereof, and a recording medium capable of converging.

[0011]

According to the method of the present invention, a reception signal of N channels (N is an integer of 2 or more) is converted to N
N pseudo echo signals are generated through a pseudo echo path that simulates a number of echo paths, and these N pseudo echo signals are added to obtain a total pseudo echo signal. To obtain the residual echo signal, and in particular, to determine the size of the received signal of each channel, that is, the amplitude or power, and to determine the update step size of the received signal channel corresponding to the largest one of these sizes. Therefore, the update step size of a channel of a reception signal other than the reception signal channel is set to be small, and generally, the update step size of the former is set to 1 so that the convergence speed is the fastest, and that of the latter is set to be smaller than 1, and The received speech signal is obtained by using the updated step size of the channel and the residual echo signal to obtain a correction vector of a pseudo echo path of the channel, and the correction vector Le By modifying the properties of the estimated echo path in the channel.

[0012]

Embodiments of the present invention will be described below with reference to the drawings. For ease of understanding, a case will be described with reference to FIG. 1 as an example in which two reception channels are stereo channels and one transmission (sound collection) channel is one channel. 1A, parts corresponding to those in FIG. 7 are denoted by the same reference numerals. The received signals x ₁ (k) and x ₂ (k) are supplied to the pseudo echo paths 18 ₁ and 18 ₂ , respectively, to generate pseudo echo signals corresponding to the echo signals passing through the echo paths h ₁ and h ₂ , respectively. . That is, assuming that the number of taps of each of the pseudo echo paths 18 ₁ and 18 ₂ is L, the inner product h ＾ ₁ ^T (k) of the impulse response h ＾ ₁ ^T (k) of the pseudo echo path 18 ₁ and the reception signal vector x ₁ (k). ) X ₁ (k) and the impulse response h ＾ ₂ ^T (k) of the pseudo echo path 18 ₂ and the received signal vector x
₂ inner product of ^{_{(k) h ^ 2 T (}} k) x 2 (k) is generated.

Here, h ＾ ₁ (k) = [h ＾ ₁ (k, 0), h ＾ ₁ (k, 1),..., H ＾ ₁ (k, L−1)] ^T h ＾ ₂ (k _{) = [h ^ 2 (k} , 0), h ^ 2 (k, 2), ..., h ^ 2 (k, L -1)] T x 1 (k) = [x 1 (k), x 1 .., X ₁ (k−L + 1)] ^T x ₂ (k) = [x ₂ (k), x ₂ (k−1),..., X ₂ (k−L + 1)] ^T ( 23) These h ＾ ₁ ^T (k) x ₁ (k) and h ＾ ₂ ^T
(K) x ₂ (k) is synthesized, and the total pseudo echo signal y ＾
(K), and the difference e (k) = y from the picked-up signal y (k)
(K) −y ＾ (k) is obtained as a residual echo signal.

In this embodiment, the reception signal x of each channel
₁ (k) and x ₂ (k) are input to the step size determination unit 31, and the update step sizes μ ₁ ,
μ ₂ is calculated. In this case, the update step size is such that the convergence is fast for a channel in which the size of the received signal is large, and the update step is a small update step for a channel in which the size of the received signal is small. For example, as shown in FIG. 1B, power is used as the size of a received signal, and the sum of squares of each received signal x ₁ (k) and x ₂ (k) within a predetermined time {x ′ ₁ (k)} ^2. , ‖X '
₂ (k) ‖ ^2, i.e. power square sum calculation unit 32 ₁ in the predetermined time, 32 ₂ are calculated.

‖X '₁(K) ‖^Two= X₁ ^Two(K) + x
₁ ^Two(K-1) + ... + x₁ ^Two(K−L ′ + 1) ‖x ′_Two(K) ‖^Two= X_Two ^Two(K) + x_Two ^Two(K-1) + ...
+ X_Two ^Two(K−L ′ + 1) The calculation section L ′ of the sum of squares is, for example, about 16 samples.
Short sections are preferred in that they operate correctly. These sum of squares
Is input to the minimum value detection unit 33 and the sum of squares ‖x ′
₁(K) ‖^Two, ‖X '_Two(K) ‖^TwoIs found,
In this example, which is smaller is detected. Its square
The minimum value of the sum (‖x ′_min(K) ‖^Two) But Kaikai removal
The data is supplied to the calculation unit 34. On the other hand, the sum of squares of each channel ‖
x '₁(K) ‖^Two , ‖X '_Two(K) ‖^TwoIs the addition unit 35
The sum is added to the square root division unit 34.
The square root division unit 34 performs the following calculation.

Μ _min = {x ′ _min (k)} ² / {({x ′ ₁ (k)} ² + {x ′ ₂ (k)} ² ) (24) The output unit 36 calculates the sum of squares. The update step size for the channel corresponding to the minimum value {x ′ _min (k)} ² is μ _min
And the update step size for the other channels is output as μ = 1. Since this example is for two channels, {x ′ ₁ (k)} ² <{x ′
If ₂ (k) ‖ ^{2, μ} ₁ = ^μ _min, and mu ₂ = 1, ‖
If x _'1 (k) || ^2>‖x' a ₂ (k) || ^{2, μ} ₁
= 1, μ ₂ = μ _min are output.

The update step size μ₁, Μ_TwoHaso
Each echo path estimation unit 19₁, 19_TwoSupplied to the reflection path
Estimator 19₁, 19_TwoHas the received signal x
₁(K), x _Two(K) and the residual echo signal e (k) are input.
Reverberation path estimation unit 19₁, 19_TwoThen, the formula (2)
The following equation is calculated corresponding to (1) and (22). h ＾₁(K + 1) = h ＾₁(K) + 0.5μ₁e (k) x₁(K) / ‖x₁ (K) ‖^Two (25) h ＾_Two(K + 1) = h ＾_Two(K) + 0.5μ_Twoe (k) x_Two(K) / ‖x_Two (K) ‖^Two (26) These h ＾₁(K + 1), h ＾_Two(K + 1) is pseudo anti
Hibiki 18₁, 18_TwoEach as an impulse response.
Is determined. That is, the previous impulse response h ＾
₁(K), h ＾_Two(K) can be expressed by equations (25) and (2), respectively.
6) Correction by the correction vector of the second item on the right side
And The calculation of the correction vector is performed by, for example, Expression (25)
, The received signal x as shown in FIG.₁(K)
Is input to the vector generation unit 41 as shown in Expression (23).
Received signal vector x₁(K) is generated and the sum of squares
The vector x in the circuit 42₁Sum of squares of each element of (k) ‖
x₁(K) ‖ ^TwoIs calculated and these x₁(K), ‖
x₁(K) ‖^TwoAnd the update step size μ ₁And residual echo
The signal e (k) is input to the correction vector operation unit 43 and
The calculation of the second term on the right side of Expression (25) is performed.

As described above, the update step sizes μ ₁ and μ ₂
By using the value calculated by equation (24) for the smaller power and setting the other to 1, the received signal x
The average power ratio of ₁ (k) and x ₂ (k) is different, and the average power ratio changes due to the change of speakers, etc., and the pseudo echo path approaches the true echo path at a high speed. FIG. 3 shows the result of the simulation. Received signals x ₁ (k) and x
₂ The power ratio with (k) is 6 dB, and the pseudo echo path 18 ₁ , 1
8 number ₂ taps 1024,2 square sum circuit 32 _1, 32 ₂
Is the case where the number of added samples is 16.

The solid line is the case of this embodiment, and the broken line is
This is the case of the prior art using the learning identification method. From this figure
According to the present invention, the convergence speed is high, so that the speaker can compare
Even after a frequent change, echo cancellation is performed following this
You. This also applies to the case where there are three or more receiving channels in the above.
In this case, the received signal x of each channel can be applied.
₁(K), ..., x_N(K) Sum of squares of predetermined time ‖x '₁
(K) ‖^Two, ..., ‖x '_N(K) ‖^TwoThe smallest of (‖
x '_min(K) ‖^TwoAnd find the minimum value
Μ for_min= √‖x '_min(K) ‖^Two/ √ (‖x '₁(K)
‖^Two+ ... + ‖x '_N (K) ‖^Two), And use this as the update step size.
All channel step sizes may be one. Above
In the step size determination unit 31,
And determine the update step size for that channel
Then, the step sizes of the other channels were set to 1. This
Originally, the size of the received signal was large.
For example, by setting the update step size to 1,
A channel that converges quickly but has a small received signal size
If there is, the estimation of the corresponding echo path is performed correctly.
This makes it difficult to
The estimation of the echo path corresponding to the large channel is correct
Will not be performed properly, and the
Update so that it is less susceptible to the
This is to reduce the step size. So, for example,
When there are three talk channels, the received signal of one channel
Is relatively large, and the other two channels receive
If the size of the signal is small, updating the larger channel
The new step size is 1 and the other two are smaller
Set the update step size of the received signal to a value smaller than 1, for example
For example, it may be obtained by an equation corresponding to equation (24). In short, each
The size of a predetermined time (for example, the square of the reception signal of the channel)
Sum), and from these sizes, the larger one, that is,
Channel that can perform echo path estimation without being affected by noise
The update step size is set to 1
And it is small, that is, influenced by noise,
Large for channels where echo path estimation is not successful
Is smaller than the update step size for large channels
Update step size. The size is electric
Not only force but also amplitude may be used.

When there are a plurality of sound pickup (transmission) channels,
Each of the N-channel echo cancellers 22 ₁ ,..., 2 shown in FIG.
As for 2 _M , a pseudo echo path is passed for each received signal, and the update step size in each echo path estimation may be selected as described above. When the correction vector is obtained, the received signal is not limited to the one obtained by normalizing the received signal for each channel with the power, but the received signal normalized by the amplitude may be used. Further, the present invention can be applied not only to the learning identification method but also to a case where another algorithm such as a projection method is used.

The above-described multi-channel echo canceller according to the present invention can be operated by executing a program by a computer. In this case, the multi-channel echo canceling program is stored in a CD-ROM, a floppy (registered trademark).
The multi-channel echo canceling program stored in the program memory is installed by being installed from a disk, a magnetic disk, or the like into the program memory of the computer or downloaded to the program memory via a communication line.

[0022]

As described above, according to the present invention, when there is a power ratio of a multi-channel received signal, even if the cross-correlation is strong, the convergence speed can be improved and the amount of calculation does not increase.

[Brief description of the drawings]

FIG. 1A is a diagram illustrating a functional configuration example of an embodiment of the present invention, and FIG. 1B is a diagram illustrating a specific functional configuration example of a step size determination unit 31;

FIG. 2 shows a portion of a specific example functional configuration of the echo path estimation unit 19 in ₁ in Figure 1A.

FIG. 3 is a diagram showing an example of a computer simulation of the convergence of a pseudo echo path in the embodiment of the present invention and the prior art.

FIG. 4 is a diagram showing a functional configuration of a conventional one-channel echo canceller.

FIG. 5 is a diagram showing a functional configuration of a conventional multi-channel echo canceller.

FIG. 6 is a diagram showing a specific functional configuration of the N-channel echo canceller in FIG. 5;

FIG. 7 is a diagram showing a functional configuration of a conventionally improved two-channel echo canceller.

Claims (6)

[Claims] 1. A reception signal of N channels (N is an integer of 2 or more) is passed through a pseudo echo path simulating N echo paths.
The pseudo echo signals are generated, and the N pseudo echo signals are added to obtain a total pseudo echo signal. The total pseudo echo signal is subtracted from the collected signal to obtain a residual echo signal. In the received signal, the update step size of the received signal channels other than the received signal channel is smaller than the updated step size of the received signal channel having the highest strength, and the received signal is normalized by the received signal for each channel. And a correction vector of the pseudo echo path of the channel based on the update step size of the channel and the residual echo signal, and correcting the characteristics of the pseudo echo path of the channel by the correction vector. Channel echo cancellation method. 2. The update step size for each channel is obtained by calculating the sum of squares of the reception signal of each channel for a predetermined time, obtaining the minimum of these square sums, and defining the update step size of the channel as 2. The multi-channel echo cancellation method according to claim 1, wherein a value corresponding to a value obtained by normalizing the sum of squares by the sum of the sum of squares of all channels is used, and the update step size of other channels is set to 1. . 3. An N-channel (N is an integer of 2 or more) receiving signal is input, and N pseudo echo paths for generating and outputting pseudo echo signals are added to the N pseudo echo signals. An adder for generating a total pseudo echo signal by subtracting the total pseudo echo signal from the picked-up signal to obtain a residual echo signal; A step size determining unit that determines and updates the update step size of the reception signal channel other than the reception signal channel to a smaller value than the update step size, and receives the reception signal of each channel, the channel, and the residual echo signal. , Normalizing the received signal with the received signal to calculate a correction vector, and using the corrected vector to update the characteristics of the pseudo echo path of the corresponding channel. Multi-channel echo canceller comprising a number of echo path estimation unit. 4. The method according to claim 1, wherein the step size determination unit is configured to:
Sum 乗 x 'of the received signal of the₁‖ 、… 、
‖X '_N、, and the minimum value of these sums of squares ‖x ′
_min‖^TwoAbout √‖x '_min‖^Two/ √ (‖x '₁
‖ + ... + ‖x '_N‖), And as a result μ_minTo
‖X '_min‖^TwoChannel update step
And update step size μ for other channels.
The echo path estimating unit has a function of outputting the received signal of the channel for a predetermined time.
Vector x with each value of_n(K), and
Vector x_nSum of squares of each element of (k) ‖x _n(K)
‖^Two, And μ_ne (k) x_n(K) / ‖x_n(K) ‖^Two With the function of calculating as a correction vector
The multi-channel echo canceller according to claim 3. 5. A multi-channel echo canceling program for causing a computer to execute the method according to claim 1. 6. A computer-readable recording medium on which the multi-channel echo canceling program according to claim 5 is recorded.