JP2006121589A - Echo cancellation method, apparatus for executing same, program, and recording medium thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To enable remarkably shortening an inevitably occurring processing delay in an algorithm while avoiding the deterioration in echo component ratio estimating performance in an echo cancellation method. <P>SOLUTION: The ratio of echo components included in a short-time spectrum of a residual signal is estimated by using not only the short-time spectrum of a multi-channel reproduction signal of a present frame but also the short-time spectrum of a reproduction signal of a past frame. More specifically, the present frame and past frame of the multi-channel reproduction signal are separated into a main component consisting of the short-time spectrum of a first channel reproduction signal of the present frame, and a sub-frame consisting of the short-time spectrum of the other reproduction signal from which the correlation with the main component is removed, the ratio of the echo of the main component occupied in the residual signal is obtained, the ratio of the echo of the sub-component is obtained in the voice picked-up signal from which the correlation with the main component is removed, and the ratio of echo components of the multichannel reproduction signal occupied in the residual voice signal is obtained, and the ratio of echo component of the multi-channel reproduction signal occupied in a picked-up voice signal is estimated from these two ratios. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a multi-channel echo cancellation method, an apparatus thereof, a program thereof, and a recording medium thereof, which are applied to, for example, a communication conference system having a multi-channel sound reproduction system and cancel acoustic echoes that cause acoustic feedback and cause hearing problems. Is.

  With the recent increase in capacity of digital networks, a multi-channel loudspeaker type teleconferencing system is being studied that allows multiple people to participate easily and provide a more natural calling environment. In this system, the received voice is reproduced from the speaker and picked up by the microphone to generate an acoustic echo. If the received voice is transmitted as it is, problems such as a telephone call failure and discomfort arise. Further, when the signal power of the acoustic echo signal is larger than the signal power of the speaker reproduction signal, the acoustic echo causes howling and makes a call impossible. In order to realize a comfortable telephone call environment, an acoustic echo canceller that erases a signal component corresponding to acoustic wraparound from the speaker to the microphone from the microphone sound pickup signal is required.

A communication conference system including an M (≧ 2) channel reproduction system and an N (≧ 1) channel sound collection system performs acoustic echo cancellation with the configuration shown in FIG. That is, the received signal from each receiving terminal 1 _m (m = 1,..., M) is reproduced as an acoustic signal by each speaker 2 _m (m = 1,..., M), and passes through each M acoustic echo paths. It goes around the microphone 3 _n (n = 1,..., N). M channel echoes between the receiving terminals 1 _m (m = 1,..., M) of all M channels on the receiving side and the transmitting terminals 5 _n (n = 1,..., N) on the N channel transmitting side. The cancel unit 6 _n (n = 1,..., N) is connected to cancel the acoustic echo. The configuration of FIG. 1 is a configuration in which N echo cancellation units of M input and 1 output in FIG. 2 are arranged in parallel.

Inside the M-channel echo canceling unit 6 _n, prediction echo signals from the multi-channel reception signal is generated, the difference between the collected signal and the predicted echo signal on the basis of the difference e (k) and the received signal with the microphone sound pickup signal The adaptive filter coefficient for predictive echo generation is sequentially updated so as to decrease. In a situation where a non-echo signal is included in the collected sound signal, the adaptive filter coefficient is erroneously updated by this amount, so there is a risk that the adaptive filter diverges. In order to avoid divergence, the update amount of the adaptive filter needs to be made sufficiently small. In practice, an update amount that is unnecessarily small is selected, or an update amount that is large enough not to diverge, but inaccurate updates due to non-echo signal interference are allowed with a certain probability, and the convergence speed is reduced. It will decrease.

Japanese Unexamined Patent Publication No. 2003-250193 (Patent Document 1) discloses a multi-channel echo cancellation method that operates robustly against this non-echo component. M channel when this method is combined with a multi-delay filter (hereinafter abbreviated as MDF) method having a short processing delay in adaptive filter processing disclosed in Japanese Patent Laid-Open No. 2002-223182 (Patent Document 2) A configuration example of the echo cancellation unit 6 is shown in FIG. In this method, the ratio of the echo component in the residual signal, that is, the difference signal between the collected sound signal and the predicted echo signal is estimated for each frequency component, and the update of the adaptive filter coefficient is controlled for each frequency component according to this ratio. . Further, the MDF method uses that the convolution process can be divided into overlap save processes between smaller blocks. The tap length of the adaptive filter L, the number of divisions D (where L is divisible by _D), when the L D = L / D, since the MDF method capable convolution processing for each _{L D} samples, each _{L D} Sample Adaptive signal processing can be applied. Therefore, the processing delay, which is always a delay in the algorithm regardless of the processing capability of hardware or the like, can be reduced to 1 / D.

The prediction echo generator 60, the reproduced signal _{x m (k) (m =} 1, ..., M) is stored after being converted into the frequency domain are framed into a signal vector of length 2L _D per L _D Sample A predicted echo signal is generated from the D frames via frequency domain processing. In the subtracting unit 61, the difference between the frame of the collected sound signal from the microphone 3 and the frame of the predicted echo signal is taken. The residual signal frame is input to the echo component ratio estimation unit 63 and the filter coefficient update unit 64 via the TF conversion unit 62. The echo component ratio estimation unit 63 estimates the ratio of echo components included in the residual signal for each frequency component. The filter coefficient updating unit 64 obtains the correction direction of the filter coefficient for each frequency component from the received signal and the residual signal, and controls the magnitude according to the ratio of the echo component.

エコー成分比率推定のフローを図５に示す。
特開２００３−２５０１９３号公報 特開２００２−２２３１８２号公報 The configuration of the echo component ratio estimation unit 63 is shown in FIG. In the correlation removal unit 631, multi-channel spectra X ₁ (j, f), X _{2 (1} ) that are not correlated with each other from the spectra X ₁ (j, f),..., X _M (j, f) of the multi-channel reproduction signal. ₎ (J, f),..., X _{M (M-1)} (j, f) is obtained. In the correlation removing unit 632, a spectrum E _(m−1) (j, f) (m = 2) obtained by removing the correlation component of the _{first to (m−1)} -th channel reproduction signals from the residual signal spectrum E (j, f). , ..., M). In the coherence calculation unit 633, the coherence calculation unit 633 ₁ converts the coherence γ _1y ² (j, f) of the first channel reproduction signal X ₁ (j, f) and the residual signal E (j, f) into the coherence calculation unit 633. _m for a short time spectrum X _{m (m−1)} (j, f) and E _(m−1) (j, f) (m = 2,..., M) coherence γ _{my (m−1)} ² (j , F). The echo component ratio calculation unit 634 obtains an echo component ratio γ ² (j, f) by the following equation.

The flow of echo component ratio estimation is shown in FIG.
JP 2003-250193 A JP 2002-223182 A

In the above conventional method, the tap length L of the adaptive filter is set to a length substantially corresponding to the reverberation time. The reverberation time is 300 ms or more in a normal room. If a frame length of 2L is taken in accordance with this time, a delay (processing delay) caused by signal buffering becomes L samples, which makes it very difficult to talk as a communication system. Therefore the above-described conventional method, set a short frame length 2L _D than tap length L of the adaptive filter in order to shorten this delay, estimating an echo component ratio, controls the update amount of the adaptive filter coefficients.
However, the echo component for delaying 2L _D samples above until picked up is reproduced from speaker to microphone, be treated as non-echo component becomes a problem. That is, the frame length 2L _D, when set much shorter than tap length L of the adaptive filter echo component considered non-echo component is increased. As a result, since the echo component ratio is set to be small or the estimated value of the echo component fluctuates, the estimation performance of the echo component ratio deteriorates and the echo cancellation performance deteriorates.

In the present invention, the ratio of echo components included in the short-time spectrum E (j, f) of the residual signal is calculated from the short-time spectrum X ₁ (j, f),. A method is proposed for estimation using not only _M (j, f) but also a short-time spectrum obtained from a past reproduction signal frame.
In the present invention, the current frame and the past frame of the multi-channel reproduction signal are further divided into the main component consisting of the first channel reproduction signal of the current frame and the sub-component consisting of other frames from which the correlation with the main component has been removed. The ratio of the principal component echo to the residual signal is obtained, and the ratio of the sub component echo to the residual signal from which the correlation with the principal component is removed is obtained. A method for estimating the echo component ratio of the channel reproduction signal is proposed.

  With this method, past signal frames can be incorporated into the estimation of the echo component ratio, and even when the frame length is set to be significantly shorter than the reverberation time, the estimation performance of the echo component ratio is avoided and the echo cancellation performance is improved. Deterioration can be prevented.

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below with reference to the drawings. Corresponding portions in the respective drawings are given the same reference numerals, and redundant description is omitted.
[First Embodiment]
The present invention will be described with respect to the case where it is composed of an M (≧ 2) channel reproduction system and an N (≧ 1) channel sound pickup system as shown in FIG. Similar to the above-described conventional method, N echo cancellation units with M inputs and 1 outputs as shown in FIG. 2 are arranged in parallel to support an N-channel sound collection system. In the present invention, the M channel echo cancellation unit 6 shown in FIG. 1 whose internal configuration is shown in FIG. 3 is replaced with an M channel echo cancellation unit 7 whose internal structure is shown in FIG. FIG. 7 shows the configuration of the echo component ratio estimation unit 73 in FIG.

In the following, the tap length of the adaptive filter is L, the number of divisions is D (where L is divisible by D), L _D = L / D, the frame length is 2L _D samples, the shift length is L _D samples, and the frame time is j. To do. The signal frame at frame time j is composed of signal samples at sample times k = jL _D −2L _D +1 to jL _D. The relationship between the signal sampling time k and the frame time j at this time is as shown in FIG. An example in which the short-time spectrum X ₁ (j−1, f),..., X _M (j−1, f) one frame before is used as the spectrum obtained from the past reproduction signal frame will be described.

TF conversion unit ₇₀₁ m in the predicted echo generation unit 70 of FIG. 6 (m = 1, ..., M) in the playback signal _x m (k) in the time domain of each channel _{L D} sample length per 2L _D The signal vector is framed, converted to a short-time spectrum using FFT, and stored in the filter processing unit 702.

この各スピーカからの予測エコー信号Ｙ＾_ｍ（ｊ，ｆ）を足し合わせて周波数領域の予測エコー信号Ｙ＾（ｊ，ｆ）を次式のように得る。

Next, in the filter processing unit 702, the short-time spectrum X _m (j, f) of the m-th channel reproduction signal accumulated as shown in the following equation and the adaptive filter coefficient W _m (j, d, f) in the frequency domain are obtained. Multiplication processing and addition are performed to obtain a predicted echo signal Y ^ _m (j, f) in the frequency domain that wraps around the microphone from the m-th speaker.

Obtaining prediction echo signal Y ^ m _(j, f) from the respective speakers predicted echo signal Y ^ (j, f) in the frequency domain sum of the the following equation.

The FT transform unit 703 transforms the frequency domain prediction echo to the time domain by inverse FFT, and obtains the time domain prediction echo ＾ (k) as follows.

In the subtracting unit 71, the framed unit 711 frames the collected sound signal y (k) in the time domain, and the subtractor 712 subtracts the frame of the predicted echo signal y ^ (k) to obtain the residual signal e ( Find the frame of k).

The TF transform unit 72 fills the frame of the time domain residual signal e (k) with L _D zeros and transforms it into the frequency domain as shown in the following equation, thereby converting the frequency domain residual signal E (j, f )

In the echo component ratio estimation unit 73, the following steps F1 to F7 are used to perform a residual for each frequency component from the frequency domain multi-channel reproduction signal X _m (j, f) and the frequency domain residual signal E (j, f). A ratio S (j, f) of echo components included in the signal is obtained. FIG. 8 shows a flow for estimating the echo component ratio S (j, f).
Step F1
The short-time spectrum X ₁ (j, f),..., X _M (j, f) of the multi-channel reproduction signal obtained from the current frame is stored in the current accumulation unit 731a1 in the correlation removal unit 731 shown in FIG. .

ここで、ε［］は、平均をとることを意味し、平均処理の一例としては、

のように、１フレーム前の処理結果と０〜１の値をとる平滑化定数βを用いる方法がある。 Step F2
The decorrelation unit 731B1, for example, short-time spectrum _X 2 multichannel reproduction signal by the following equation method _{(j, f), ...,} X M (j, f) from _X 1 (j, f) the correlation components of the The spectrum X _{2 (1)} (j, f),..., X _{M (1)} (j, f) is obtained as a part of the sub-component of the multi-channel reproduction signal spectrum.

Here, ε [] means taking an average, and as an example of the averaging process,

As described above, there is a method using a processing result of one frame before and a smoothing constant β that takes a value of 0 to 1.

なお、ｎフレーム前の短時間スペクトルＸ_１（ｊ−ｎ，ｆ），…，Ｘ_Ｍ（ｊ−ｎ，ｆ）をエコー成分比率推定に使用する場合にも、同様の計算により得られた結果を多チャネル再生信号スペクトルの副成分の一部とすればよい。 Step F3
From the spectrum X ₁ (j−1, f),..., X _M (j−1, f) of the multi-channel reproduction signal one frame before accumulated in the past accumulating unit 731a2 in the correlation removing unit 731b2, X ₁ A spectrum X _{1 (1)} (j−1, f),..., X _{M (1)} (j−1, f) obtained by removing the correlation with (j, f) as follows is obtained, and a multi-channel reproduction signal is obtained. Let it be part of a subcomponent of the spectrum.

Incidentally, n frames before short-time spectrum _{X 1 (j-n, f} ), ..., X M (j-n, f) in the case of using the echo component ratio estimation results obtained by the similar calculation May be a part of the sub-component of the multi-channel reproduction signal spectrum.

Step F4
The correlation removing unit 732 obtains a spectrum E ₍₁₎ (j, f) obtained by removing a correlation component with X ₁ (j, f) from the short-time spectrum E (j, f) of the residual signal of the current frame. .

Step F5
In the coherence calculation unit 733 ₁ , the short-time spectrum X ₁ (j, f) of the first channel reproduction signal of the current frame, which is the main component of the multi-channel reproduction signal spectrum, and the spectrum E (j, f) of the current residual signal. ) To find the next coherence.

のうちで、相関除去された残差信号スペクトルとの誤差
｜Ｅ_（１）（ｊ，ｆ）−Ｅ＾_（１）（ｊ，ｆ）｜^２
が最小となるスペクトルである。この誤差を最小にするスペクトルは、

とし、 Step F6
In subcomponent echo ratio calculation unit 733 _2, first decorrelation residual signal spectrum _{E (1) (j, f} ) echo component _{E ^ (1)} contained in the Request (j, f). The echo component E ^ ₍₁₎ (j, f) is a sub-component X _{2 (1)} (j, f),..., X _{M (1)} (j, f), X ₁ of the multi-channel reproduction signal short-time spectrum. ₍₁₎ (j−1, f),..., X _{M (1)} (j−1, f) linear sum

Of the residual signal spectrum from which correlation has been removed | E ₍₁₎ (j, f) −E ^ ₍₁₎ (j, f) | ²
Is the spectrum with the minimum. The spectrum that minimizes this error is

age,

により求められる。さらに、次式により副成分のエコー比率を周波数成分ごとに求める。

Is required. Further, the echo ratio of subcomponents is obtained for each frequency component by the following equation.

Step F7
In the echo component ratio calculation unit 734 in FIG. 7, the ratio of echo components in the residual signal spectrum E (j, f) is obtained from the ratios obtained in steps F5 and F6.

ただし、ｄ＝１，…，Ｄ、Ｘ_ｍ ^＊（ｊ，ｆ）はＸ_ｍ（ｊ，ｆ）の複素共役である。 In the filter update unit 74 of FIG. 6, first, the frequency domain adaptive filter coefficient correction amount dW _m (j, d) is calculated from the residual signal E (j, f) and the reproduced signal X _m (j, f) for each frequency component. , F). Next, the magnitude of the correction amount dW _m (j, d, f) of the adaptive filter coefficient is controlled using the ratio S (j, f) of the residual echo component included in each frequency component of the frequency domain residual signal. Meanwhile, the adaptive filter coefficient W _m (j, d, f) in the frequency domain is sequentially updated. Specifically, for example, Jia-sien Soo, Khee K. Pang, “Multidelay Block Frequency Domain Adaptive Filter,” IEEE Transactions on Acoustics, Speech, and Signal Processing, Vol.38, No.2, pp.373-376, In combination with the proposal of 1990., the correction amount dW _m (j, d, f) and the adaptive filter coefficient W _m (j, d, f) can be obtained as follows.

Here, d = 1,..., D, and X _m ^* (j, f) are complex conjugates of X _m (j, f).

ただし、ｍ＝１，…，Ｍ、ｄ＝１，…，Ｄ、ｆ＝１，…，２Ｌ_Ｄ、μは０〜１の値をとるステップサイズ、δは正則化定数である。また、ｐ（ｊ，ｆ）は次式により計算される。

Here, m = 1,..., M, d = 1,..., D, f = 1,..., 2L _D , μ is a step size having a value of 0 to 1, and δ is a regularization constant. P (j, f) is calculated by the following equation.

In this way, by updating the adaptive filter coefficient W _m (j, d, f) for each processing unit, the updated adaptive filter coefficient can be used in the next processing.
When the processing of the current frame is completed, the reproduction signal information stored last in the current storage unit 731a1 is transferred to and stored in the past storage unit 731a2.
Note that the current storage unit 731a1 and the past storage unit 731a2 are particularly distinguished in the storage unit 731a, and the reproduction signal information stored in the current storage unit 731a1 at the end of the series of processes as described above is stored in the past. There is also a method of processing the latest information as the current information in the information stored in one storage unit 731a instead of transferring to the unit 731a2. In addition, as shown in FIG. 9, there is a method of directly using the spectrum of the input reproduction signal instead of taking out the spectrum of the current reproduction signal used for processing from the storage unit.

[Second Embodiment]
The present invention is intended to improve the echo path estimation speed with respect to the first embodiment. FIG. 10 shows the overall configuration, and FIG. 11 shows the configuration of the M channel echo cancellation unit.
In the correlation fluctuation processing unit 9 _m (m = 1,..., M) in FIG. 10, for example, the received signal u _m (k) is amplitude-modulated with a random number and added to the original signal for each channel as shown in the following equation. A reproduction signal x _m (k) whose cross-correlation is constantly changing is generated and reproduced from each speaker.

Without this correlation variation processing, since the channel correlation of the M channel received signal is high, the true echo path characteristic of the mth channel is obtained from the input / output relationship between the mth channel reproduction signal and the first channel sound pickup signal. Is not always required. However, by applying this correlation variation process to the received signal, the echo path characteristics of the M channel can be reliably estimated.
In the M channel echo canceling unit 8 of FIG. 11, a signal vector having a smaller inter-channel correlation in which the additional signal component is emphasized than the received signal component is generated by the signal converting unit 85, and instead of a reproduced signal vector having a large inter-channel correlation. Use to update the adaptive filter. For this update, for example, the adaptive algorithm proposed in Patent Document 2 is used, and the step size is controlled by the ratio S (j, f) of the residual echo component included in the residual signal. By using a correction vector having a small correlation between channels, the convergence speed of the echo path by the adaptive filter is improved.

The figure which shows the general structure of the echo cancellation apparatus of M input N output. The figure which shows the general structure of the echo cancellation apparatus of M input 1 output. The figure which shows the structure of the conventional M channel echo cancellation part. The figure which shows the structure of the conventional echo component ratio estimation part. The figure which shows the flow of the conventional echo component ratio estimation. The figure which shows the structure of the M channel echo cancellation part of 1st Embodiment. The figure which shows the structure of the echo component ratio estimation part of 1st Embodiment. The figure which shows the flow of echo component ratio estimation of 1st Embodiment. The figure which shows the structure of the modification of the echo component ratio estimation part of 1st Embodiment. The figure which shows the general structure of the echo cancellation apparatus of M input 1 output channel of 2nd Embodiment. The figure which shows the structure of the M channel echo cancellation part of 2nd Embodiment. The figure which shows the relationship between the sample time k of a signal, and the frame time j.

Claims (16)

In a method of canceling an echo component by predicting an echo signal from a reproduced signal of a plurality of channels and subtracting the predicted echo signal from each collected signal of at least one channel of collected sound signals,
Estimating the ratio of echo components contained in the short-time spectrum of the residual signal for each frequency from the short-time spectrum of the reproduction signal of the current frame and the short-time spectrum of the reproduction signal of at least one past frame;
Correcting the adaptive filter coefficient from the ratio of echo components contained in the short-time spectrum of the residual signal,
An echo canceling method characterized by the above. The method of claim 1, wherein
Estimating the ratio of echo components contained in the short-time spectrum of the residual signal for each frame length shorter than the tap length of the adaptive filter;
An echo canceling method characterized by the above. The method according to claim 1 or 2, wherein
Modifying the adaptive filter coefficients for each frame;
An echo canceling method characterized by the above. In the method in any one of Claims 1-3,
The main component is the short-time spectrum of the first channel playback signal of the current frame,
The other short-time spectrum from which the correlation with the main component is removed is used as a subcomponent,
Find the ratio of the principal component echo to the short-time spectrum of the residual signal,
Find the fraction of the short-time spectrum of the residual signal that sub-component echoes have removed the correlation with the main component,
Estimating an echo component ratio in the short-time spectrum of the residual signal from the above two ratios;
An echo canceling method characterized by the above. The method of claim 4, wherein
Correlation between the echo component ratio S (f) in the short-time spectrum of the residual signal, the ratio γ ₁ ² (f) of the main component echo in the short-time spectrum of the residual signal, and the sub-component echo with the main component From the ratio γ ₂ ² (f) in the short-time spectrum of the residual signal from which

Asking for,
An echo canceling method characterized by the above. The method of claim 5, wherein
The echo component E ^ ₍₁₎ (f) included in the short-time spectrum E ₍₁₎ (f) of the residual signal from which the correlation with the principal component is removed is represented by | E ₍₁₎ (f) -E ^ _{( 1)} (f) | Obtain as a linear sum that minimizes ² ;
The ratio γ ₂ ² (f) in the short-time spectrum of the residual signal from which the echo of the subcomponent is removed from the correlation with the main component is expressed as follows:

Asking for,
An echo canceling method characterized by the above. In the method in any one of Claims 1-6,
As a reproduction signal, using a signal obtained by adding a reception signal subjected to correlation variation processing to the reception signal,
Use a signal in which the additional signal component is emphasized over the received signal component in the adaptive filter update;
An echo canceling method characterized by the above. Means for receiving a reproduced signal of a plurality of channels and a collected sound signal of at least one channel;
An adaptive filter that predicts an echo signal;
Means for subtracting the predicted echo signal from the signal from each sound collection unit;
Means for estimating the ratio of echo components included in the short-time spectrum of the residual signal for each frequency from the short-time spectrum of the reproduced signal of the current frame and the short-time spectrum of the reproduced signal of at least one past frame;
Means for correcting an adaptive filter coefficient from a ratio of echo components included in the short-time spectrum of the residual signal;
An echo canceller comprising: The apparatus of claim 9.
Means for estimating the ratio of echo components contained in the short-time spectrum of the residual signal for each frame length shorter than the tap length of the adaptive filter;
An echo canceller comprising: The device according to claim 8 or 9,
Means for correcting the adaptive filter coefficients for each frame;
An echo canceller comprising: In the apparatus in any one of Claims 8-10,
Means for determining the proportion of the short-term spectrum (principal component) of the first channel reproduction signal of the current frame in the short-time spectrum of the residual signal;
Means for determining the proportion of echoes of other short-time spectra (subcomponents) that have been de-correlated with the principal component in the short-time spectrum of the residual signal that has been de-correlated with the main component;
Means for estimating an echo component ratio in the short-time spectrum of the residual signal from the output of the means for obtaining the two ratios;
An echo canceller comprising: The apparatus of claim 11.
As a means for estimating the echo component ratio S (f) in the short-time spectrum of the residual signal, the ratio γ ₁ ² (f) in which the main component echo occupies the short-time spectrum of the residual signal and the sub-component echo are mainly used. From the ratio γ ₂ ² (f) in the short-time spectrum of the residual signal from which the correlation with the component is removed,

Means to obtain,
An echo canceller comprising: The apparatus of claim 12.
The short-time spectrum of the residual signal from which the correlation with the main component has been removed is used as a means for obtaining the ratio γ ₂ ² (f) of the residual signal from which the echo of the subcomponent has been removed from the short-term spectrum from which the correlation with the main component has been removed. E ₍₁₎ Echo component E ^ ₍₁₎ (f) included in (f)
| E ₍₁₎ (f) −E ^ ₍₁₎ (f) | ²
And the ratio γ ₂ ² (f) in the short-time spectrum of the residual signal from which the echo of the subcomponent is removed from the correlation with the main component,

Means to obtain,
An echo canceller comprising: The apparatus according to any one of claims 8 to 13,
A correlation fluctuation processing unit for fluctuating the correlation of the received signal;
A signal converter that generates a signal in which the additional signal component is emphasized over the received signal component;
An echo canceller comprising:   An echo cancellation program for executing the echo cancellation method according to claim 1 by a computer. The computer-readable recording medium which recorded the echo cancellation program of Claim 15.

Images