JPH0774681A - Echo canceller - Google Patents

Abstract

PURPOSE:To improve the echo cancelling effect. CONSTITUTION:A decoded voice signal from a decoder 13 is fed to pseudo echo paths 24a, 24b and an impulse response estimate section 25a to control a filter coefficient of a band width magnification synthesis filter 59 in response to a decoded spectrum envelope parameter and a decoding exciting signal is fed to an impulse response estimate section 25b through the filter 59 as a signal clarified itself to be a signal for each coded block. Each output of the pseudo echo paths 24a, 24b is subtracted respectively from a signal of the echo path side by cancelling circuits 18a, 18b and outputs of the cancelling circuits 18a, 18b are fed respectively to impulse response estimate sections 25a, 25b, an evaluation selection section 56, and buffers 57a, 57b. Each estimate impulse response of the impulse response estimate sections 25a, 25b is set to the pseudo echo paths 24a, 24b. The evaluation selection section 56 selects an echo cancelling means corresponding to a minimum residual echo.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an echo canceller for canceling echo signals which are a cause of howling and an auditory obstacle in a loudspeaker telephone conference communication system, a two-wire to four-wire conversion system and the like.

[0002]

2. Description of the Related Art FIG. 2A shows a loudspeaker type communication terminal device equipped with a high-efficiency voice encoder / decoder. The signal from the transmission line received through the input terminal 11 is decoded by the transmission line decoder 12 into a baseband signal, and the baseband signal is converted into a coded voice signal by a voice decoder 13 into, for example, a telephone band voice signal. It is decoded and further converted into an analog signal by the D / A converter 14. This analog audio signal is sent to the speaker 1
5 and is emitted as an acoustic signal. On the other hand, the voice signal received by the microphone 16 is converted into a digital signal by the A / D converter 17, the echo signal is eliminated by the erasing circuit 18, and the voice signal is supplied to the voice coder 19 for high efficiency voice encoding. The encoded voice signal is encoded by the transmission line encoder 21 into a signal on the transmission line and transmitted from the output terminal 22 to the transmission line. In order to prevent the acoustic signal emitted from the speaker 15 from being captured by the microphone 16 and transmitted as an echo signal, the pseudo echo path 24 that imitates the echo path 23 connecting the speaker 15 and the microphone 16 is a speaker. The signal to the speaker 15 is branched and supplied to the pseudo echo path 24, the output passing through the pseudo echo path 24 is supplied to the cancellation circuit 18, and is subtracted from the signal from the microphone 16, that is, the echo signal is canceled. To be done. The input signal of the speaker 15 and the output signal of the erasing circuit 18 are input to the impulse response estimation unit 25, the impulse response of the echo path 23 is estimated, and the estimated impulse response characteristic is set in the pseudo echo path 24. Echo path 24
The impulse response is convolved with the signal input to the.

Similarly, in a 4-wire / 2-wire conversion system, FIG.
2A, the output side of the D / A converter 14 and the input side of the A / D converter 17 are connected to the 4-wire side terminal of the hybrid transformer 26, as indicated by the same reference numerals. The 2-wire type transmission line 27 is connected to the 2-wire side terminal of the hybrid transformer 26. There is an echo path 28 in which the output signal of the D / A converter 14 leaks from the hybrid transformer 26 and reaches the A / D converter 17 side, and the echo signal passing through this echo path 28 is canceled by the canceling circuit 18 in the case of FIG. 2A. Similarly, it is canceled.

Further, as shown in FIG. 3, in a mobile radio communication base station 29, a digital voice signal from an analog network 31 is encoded by a voice encoder 19 and further encoded by a transmission line encoder 21. In the mobile terminal device 32, the signal of the base station 29 is transmitted to the mobile terminal device 32 via a wireless line, and the signal from the base station 29 is converted into a baseband signal by the transmission line decoder 33, and further decoded into a voice signal by the voice decoder 34.
The audio signal is converted into an analog signal by the D / A converter 14 and supplied to the speaker 15. The audio signal from the microphone 16 is converted into a digital signal by the A / D converter 17,
The voice encoder 35 performs high-efficiency encoding, and the encoded output is converted into a code signal on the transmission line by the transmission line encoder 36 and transmitted to the base station 29 via a wireless line. In the base station 29, the received signal is decoded into a baseband signal in the transmission line decoder 12, and the baseband signal is decoded into a digital voice signal in the voice decoder 13 to obtain the analog network 31.
Sent to. Also in this case, the echo path 23 from the speaker 15 to the microphone 16 is formed, and the cancellation of the echo signal through the echo path 23 is performed by the voice encoder 19 of the base station 29.
Is performed by the pseudo echo path 24, the erasing circuit 18, and the impulse response estimation unit 25 provided between the input side of the ∘ and the output side of the speech decoder 13.

The speech coder and speech decoder shown in FIGS. 2A, 2B, and 3 are those for encoding and decoding speech signals with high efficiency using linear prediction, for example, CELP (Code).
Excited Linear Predictio
n: code-excited linear prediction) coding method is used. Briefly speaking, the input voice signal is L as shown in FIG. 4A.
The PC analysis unit 41 performs LPC analysis to obtain a spectrum envelope parameter for each block, and this parameter is set as a filter coefficient in the linear prediction synthesis filter 42. The excitation signal selected from the excitation source 43 is given a gain in the gain section 44 and is supplied to the linear prediction synthesis filter 42 as an excitation signal. The distortion evaluation unit 45 performs selection of the excitation signal of the excitation source 43 and gain control given to the gain unit 44 so that the distortion of the synthesized signal synthesized by the synthesis filter 42 with respect to the input speech signal is minimized. A code indicating an excitation signal (vector) in which an audio signal is selected in block units, a code indicating a set gain, and a spectrum envelope parameter are output as a coded signal.

As shown in FIG. 4B, the decoder for decoding the coded signal extracts the spectrum envelope parameter in the spectrum envelope decoder 47, sets it as the filter coefficient in the linear prediction synthesis filter 48, and also sets the excitation source. The excitation signal is selectively decoded by the decoder 49, and the excitation signal is provided with the gain decoded by the gain unit 51 and is input to the linear prediction synthesis filter 48 as the excitation signal.
The audio signal is restored and output from 8.

[0007]

Although the conditions required for echo signal cancellation differ between acoustic echo and line echo, both may be used together with a high-efficiency speech codec (CODEC). Since they are common, only the acoustic echo canceller will be described below.

As a method for estimating the impulse response of the echo path, there is a method in which noise in a wide band is emitted from the speaker 15 before voice communication is started and a signal input from the microphone 16 is used. In this method, since the band of the signal from the speaker 15 is wide, an accurate impulse response can be estimated in a short time, but there is a drawback that it cannot follow the fluctuation of the impulse response of the echo path 23.

Apart from this method, there is a method of sequentially correcting the estimation of the impulse response while using the voice signal. The problems of this method are the speed that follows the fluctuation of the impulse response, the estimation accuracy, and the estimation calculation amount. It is difficult to achieve both speed and accuracy that follow. If a simple learning and identification method is used as the estimation method, the speed of following a signal that is biased to a low frequency region such as voice as an input signal is extremely reduced. Various methods have been tried to solve these difficulties, but practical problems such as an increase in the amount of calculation have not been sufficiently solved.

[0010]

According to the present invention, a plurality of echo canceling means including a pseudo echo path, a canceling circuit, and an impulse response estimating means are provided, and the plurality of echo canceling means estimate the impulse response. The means are different from each other. From these erasing means, the one having the smallest residual echo amount is selected by the selecting means.

In this way, in order to evaluate the echo canceller and select the best one, in order to estimate the impulse response,
Although there is a delay in the update, there is a delay in encoding / decoding, if an encoder / decoder for performing encoding / decoding for transmitting / receiving voice signals to the echo path is generated. During, the impulse response is estimated and the one with the smallest residual reverberation is selected, without increasing the delay as a system.

[0012]

FIG. 1 shows an embodiment of the present invention, in which parts corresponding to those in FIG. 3 are designated by the same reference numerals. In this example, an erasing circuit 18a, a pseudo echo path 24a, an echo canceling means 55a including an impulse response estimating section 25a, and an erasing circuit 18
b, the pseudo echo path 24b, and the impulse response estimation unit 25.
and an echo canceller 55b including b. That is, in this example, the echo canceling means 55a and 55b are provided, and the impulse response estimating means 25a and 25b of the echo canceling means 55a and 55b have different estimation methods. The outputs of the elimination circuits 18a and 18b of the echo elimination means 55a and 55b are supplied to the evaluation selection unit 56, and the one having the smallest residual echo amount is selected. In order to estimate the impulse response and compensate for the delay due to the selection of the one having the smallest residual echo amount, the outputs of the elimination circuits 18a and 18b are supplied to the buffers 57a and 57b, and the output of the evaluation selection unit 56 causes the switch 58 to operate. Under control, the buffers 57 a and 57 b which have the smallest residual echo amount are selected and supplied to the speech coder 19.

In the conventional echo canceller, one sample of a voice signal is processed as a processing unit, but the present invention involves a delay. Here, the learning identification method will be taken as an example of the sequential processing, and an example of the present invention will be described in which the echo canceling effect is enhanced by expanding when the delay is allowed. In the learning identification method, an impulse response sequence (L-dimensional vertical vector) at time _n is h _n , and an input sequence vector up to time _n , which has been traced up to the time L, is x _n , (x _n , x _n-1 , ...).
x _{n-L + 1} ) ^T , the echo sequence is y _n , (y _n , y _n-1 ,
, Y _{n-L + 1} ) ^T , where the residual echo value at time _n is e _n and α is a constant, a new impulse response sequence h at time n
Update _{n + 1} with the following formula.

[0014] _{h n + 1 = h + α} (e n / | x n | 2) x n (1) The echo signal is eliminated as ^{_{e n = y n -h n T}} x n (2). In the prior art, this is done, but for the sake of simplicity, consider a case where a delay of one sample is allowed in this embodiment. That is, it is assumed that x _{n + 1} and e _{n + 1} are known for estimating the impulse response at the time point _n . At this time, h _n and h _{n + 1} are estimated using the equation (1) for a plurality of types α ₀ and α _{1 of the} constant α. The specific procedure at time n is as follows.

First, using α ₀ , the impulse response estimation unit 25a estimates h _{n + 1} (0). And the erase circuit 18
E _{n + 1} (0) is obtained as the output of a. Next, using another α ₁ , the impulse response estimation unit 25b
Then, h _{n + 1} (1) is estimated by using e as the output of the erasing circuit 18b.
_{Find n + 1} (1). The evaluation selection unit 56 compares e _{n + 1} (0) and e _{n + 1} (1), and uses the smaller α to calculate h _{n + 1 of the} pseudo echo paths 24a and 24b.
And the minimum of e _{n + 1} (0) and e _{n + 1} (1) is obtained. Rewrite the state of. After that, the process proceeds to the next point.

In this way, even if the delay is only one sample, those having a large echo canceling effect are sequentially selected, so that the canceling effect is improved. When high-efficiency coding of speech is performed on a block-by-block basis as in the CELP method described above, the speech is used for coding, and a delay occurs on a coding-block basis. In other words, if delays are allowed to occur in such encoding and decoding, the impulse response is estimated by multiple impulse response estimators within the delay of the block unit, and the minimum based on the estimation. If the echo canceling means that is the residual echo amount is selected, the delay due to the selection of the minimum residual echo amount does not occur. Further, the constant α is not limited to the different impulse response estimation method.

For example, as shown in FIG. 2 by attaching the same reference numerals to the portions corresponding to those in FIGS. 1, 3 and 5, the echo canceling means 55 is shown.
In a, the decoded speech signal from the decoder 13 and the output of the cancellation circuit 18a are input to the impulse response estimation unit 25a, and in the echo canceller 55b, the decoding excitation signal from the decoding excitation source 49 is expanded in bandwidth. The signal is input to the impulse response estimation unit 25b through the filter 59, and the impulse response estimation unit 25b
The impulse response is estimated with the output of 8b. The filter coefficient of the bandwidth expansion synthesis filter 59 is controlled according to the decoded spectrum envelope parameter of the spectrum envelope decoder 47 to give a gentle unevenness to the spectrum envelope of the excitation signal whose frequency characteristic is substantially flat, and to view it in units of coding blocks. Impulse response estimation unit 25b as a signal trivialized with
Is supplied to. The delays in the buffers 57a and 57b are set in coding block units, and the evaluation selecting unit 56 selects the echo canceling means that minimizes the average residual echo amount in one coding block as a whole.

Although the echo canceling means is two in the above description, it may be three or more. The present invention can be applied to the echo canceller shown in FIG.

[0019]

As described above, FIR filters having 512 taps are used as the pseudo echo paths 24a and 24b, the coding block length is 160 samples, and the sampling frequency is 8 kHz.
z, the correction step size at the time of impulse response estimation is 1.0, normal voice is input, and the echo cancellation rate is estimated by using the conventional learning identification method by using the echo that convolves the impulse response of a normal room. The simulation results for the case of using the two echo cancellers and the case of using the learning identification method in the embodiment shown in FIG. 2 are as follows.

20 ms 200 ms 2s 4s Conventional method 12.1 19.0 24.3 26.9 Present invention (1) 18.8 20.3 25.4 27.9 Present invention (2) 18.7 20.8 25 92.3 In the present invention (1), the impulse response estimation is updated up to 8 samples later, and in the present invention (2), the impulse response estimation is updated up to 1 sample later.

The echo canceling means 55b quickly follows the impulse response and has a large echo canceling effect, but in some cases it becomes unstable and the residual echo may increase sharply. In this case, the stable echo canceling means 55a is always selected automatically, and the stable echo canceling performance is improved as a whole. As described above, according to the present invention, by using a plurality of echo canceling means having different impulse response estimators,
Since the best one is selected, the echo canceling effect is great.

Further, when combined with the voice encoding of the block processing, the delay unavoidable in the voice encoding can be effectively used. The calculation amount increases because a plurality of echo cancellers are prepared, but parallelization is easy and the processing time does not become long.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a block diagram showing another embodiment of the present invention.

FIG. 3A is a block diagram showing a conventional acoustic echo canceller in a loudspeaker type communication terminal, and B is a block diagram showing a conventional line echo canceller.

FIG. 4 is a block diagram showing a conventional echo canceller having a remote echo path.

5A is a block diagram showing a high-efficiency encoder, and B is a block diagram showing its decoder. FIG.

Front Page Continuation (72) Inventor Yutaka Kaneda 1-1-6 Uchisaiwaicho, Chiyoda-ku, Tokyo Nihon Telegraph and Telephone Corporation

Claims (2)

[Claims] 1. A signal to the echo path is passed through the pseudo echo path to convolute an impulse response, and the output of the pseudo echo path is subtracted from the signal from the echo path by an erasing circuit, and the output of the erasing circuit and the A signal to the echo path is input to the impulse response estimation means to estimate the impulse response of the echo path, and the estimation result is set to the pseudo echo path in the echo canceller, in which the pseudo echo path and the cancellation circuit are provided. A plurality of echo canceling means including the impulse response estimating means are provided with the impulse response estimating means being different from each other, and means for selecting the one having the smallest residual echo amount from these canceling means. Echo canceller that is characterized. 2. An encoder and a decoder for performing encoding and decoding which cause a delay with respect to a voice signal transmitted and received on the echo path, are provided, and between the encoding and decoding delays,
The echo canceller according to claim 1, wherein the impulse response is estimated and the selection is performed.