JPH0481133A - echo canceller - Google Patents

Abstract

PURPOSE:To ensure prescribed echo cancelling quantity regardless of level fluctuation of a noise component and an echo component by estimating a level ratio of the echo component and the noise component from a reception signal and a transmission signal and controlling the revision speed of a tap coefficient of an adaptive filter with respect to an object value of echo cancellation quantity. CONSTITUTION:A revision speed control circuit 5 estimates an echo component and a noise component from a reception signal Rin and a transmission signal Sout of an echo canceller and calculates a revision speed alpha of a tap coefficient of an adaptive filter 4 based on the relative relation (ratio) for the purpose of control. Thus, the value alpha is continuously controlled in response to the estimate level of the echo component and the noise component and when the echo component is smaller than noise to decrease the value alpha for aiming at the assurance of a prescribed echo cancellation thereby minimizing the deterioration in the estimate accuracy while keeping the adaptive speed to a certain extent.

Description

[Detailed Description of the Invention] [Summary] This invention relates to an echo canceller that uses a learning identification method to adaptively cancel echo components in an acoustic system from a speaker to a microphone. In an echo canceller that performs control to adaptively cancel the echo components of the acoustic system, the aim is to achieve a certain target amount of echo cancellation without being affected by the noise level or echo component level on the microphone side. The adaptive filter is configured to include an update rate control circuit that estimates a level ratio of an echo component and a noise component from a transmitted signal and controls an update rate coefficient value of a tap coefficient of the adaptive filter with respect to a target value of an amount of echo cancellation.

[Industrial application field]

The present invention relates to an echo canceller, and more particularly to an echo canceller that performs control to adaptively cancel echo components of an acoustic system from a speaker to a microphone using a learning identification method.

Acoustic echo cancellers are used in various systems such as video conferences, long-distance telephones, and loudspeaker telephones, to reduce the echo that goes around from the speaker to the microphone, thereby accurately transmitting the original near-end speaker's signal to the other party. The learning identification method, which is a modification of the LMS algorithm, is most suitable as the algorithm to be applied to this echo canceller, since it is simple and provides relatively good echo cancellation characteristics.

[Conventional technology]

Figure 4 shows a conventional echo canceller.
1 is the speaker, 2 is the echo path, 3 is the microphone, and 4
is an adaptation that cancels the echo component from the transmitted signal by estimating the characteristics of the echo path 2 from the received signal Rin and the transmitted signal 5out, generates a pseudo echo S'' (echo replica), and subtracts this from the echo component S. It is a filter (ADF).

In such an echo canceller, pseudo echo S
゛ is estimated control (tap coefficient update) so that the received signal Rin has characteristics equal to the impulse response of echo path 2.
The adaptive algorithm is obtained by passing the echo canceller output 5out through the adaptive filter 4.
The i (i·1..., n)-th stamp coefficient H(i) is updated for each sample by feedback control according to the learning identification method shown in the following equation (2) so that H(i) is minimized.

However, X(i) is the received signal of the i-th tamper, and e is the residual component (transmitted signal 5o in Fig. 4) that cannot be completely canceled.
ut), α indicates a constant indicating the update rate of the tap coefficient.

(Problems to be Solved by the Invention) By the way, if the adaptive algorithm of the echo canceller is a sequential estimation type algorithm such as the learning identification method described above, the transmission side (near end), that is, the microphone 3, is If there is such a noise source 10, the accuracy with which the adaptive filter 4 estimates the echo path 2 will deteriorate due to its influence, and therefore the amount of echo cancellation by the echo canceller will also become smaller.

In other words, the echoes (residual echoes) leaking into the transmission signal 5out will increase.

Now, assuming that the level of IIf is N and the level of the echo component is S, the update rate coefficient α in the above learning identification method
The target echo cancellation amount ERL obtained according to the change in
E becomes the value shown by the following formula (■).

This formula ■ is published in the Journal of the Institute of Electronics and Communication Engineers '77/11 Vol.
, J60-A No. 11, is a well-known formula published in the paper No. 11.

Since conventional echo cancellers use a fixed value for α, the change in S/N is directly reflected in the echo cancellation amount E.
For example, as shown in FIG.
When the level of noise N decreases and the level of echo component S decreases (section B in the same figure (a)), or when the level of noise N increases (section D in the same figure), the amount of echo cancellation decreases. E
There was a problem that RLE decreased.

Therefore, the present invention provides an echo canceller that adaptively cancels the echo components of the acoustic system from the speaker to the microphone using such a learning identification algorithm. The aim is to achieve a certain amount of echo cancellation.

[Means to solve the problem]

FIG. 1 shows the principle of the echo canceller according to the present invention. In the present invention, the level ratio of the echo component and the noise component is estimated from the received signal and the transmitted signal, and the echo canceller is adapted to the target value of the amount of echo cancellation. The present invention is characterized in that an update speed control circuit 5 is provided to control the update speed coefficient value α of the tap coefficient of the filter 4.

[For production]

Transforming the above formula ■, we get becomes.

Here, ERLE is the required amount of echo cancellation and is a constant value.

Also, the units of the echo cancellation amounts ERLE and S/H are desohers, and if the corresponding numbers are K and s/n, respectively, then ERLE=10 log, 0(K)
・=■S/N ”” 101og+o(s
/n) ”'■.

Substituting these formulas ■ and ■ into formula ■ gives 1010g+.

(--1) α = 1101o+o(K) -101ogto(s/n
)=101og++(K-n/s).

Therefore, if we find α from 2〇, we get ...■ ...■ 1+K(n/s) becomes.

Therefore, the update rate control circuit 5 shown in FIG. 1 estimates the level of the echo component and the noise level from the received signal R4n and the transmitted signal 5out of the echo canceller, and determines the tap coefficient of the adaptive filter 4 from the relative relationship (ratio). Update speed α
The update rate control circuit 5 calculates and controls the value of .

The operation of the present invention will be explained with reference to FIG. 2. First, in section A, the S/N value is large and α is set to a large value α.

If the level of the received signal R4n becomes low and the level of the echo component decreases, as in section B of the figure (a), or if the noise level increases, as in section B of the same figure, S/n decreases (n/s increases), but in that case, the value of α is set to a value α2, α, α, which is smaller than α according to the above formula
Set to .

As a result, the value of the second term on the right side of the above equation ■ becomes thicker (becomes
Even if the S/N decreases, it is possible to prevent the echo cancellation amount HRLE from decreasing.

In this way, the α value is continuously controlled according to the estimated level ratio of the echo component and noise, and the α value is lowered when the echo component is small compared to the noise, with the goal of ensuring a constant amount of echo cancellation. By doing so, deterioration in estimation accuracy is suppressed as much as possible while maintaining a certain degree of adaptation speed.

[Embodiment] FIG. 3 shows an embodiment of the echo canceller according to the present invention shown in FIG. a squaring circuit 51 that squares the output of the squaring circuit 51; an averaging circuit 52 that averages the output of the squaring circuit 51; a multiplier 53 that multiplies the output of the averaging circuit 52 by the gain (attenuation rate) G of the echo path 2; A squaring circuit 54 inputs and squares 5out, an averaging circuit 55 averages the output of the squaring circuit 54, and an α value is calculated from the output of the multiplier 53, the output of the averaging circuit 55, and the target echo cancellation amount. The α value calculation unit 56 calculates the α value and provides it to the adaptive filter 4. In addition, other components are as follows.
Same as Figure 1.

In an embodiment having such a configuration, the squaring circuit 51
and 54, the received signal Rin and the transmitted signal 5, respectively.
The instantaneous power is obtained by squaring out, and then the power averaged by averaging circuits 52 and 55 is obtained.

Then, the output of the averaging circuit 52 is multiplied by the gain G of the echo path 2 in a multiplier 53, and the estimated level S of the echo component S is sent to the calculation unit 56.

On the other hand, regarding the noise level N, the echo component is removed in the operating state of the echo canceller, and the transmitted signal 5ou
Since the signal appearing as t can be estimated to be approximately equal to the near-end signal of the microphone 3, that is, the noise level n, the output of the averaging circuit 55 is sent as is to the calculation unit 56.

Then, the calculation unit 56 calculates and gives an update rate constant α of the tap coefficients of the adaptive filter 4 using the echo cancellation amount, which is a constant value, and the input signals S and n, according to the above equation (2).

As a result, the adaptive filter 4 executes adaptive control by updating the tap coefficients using the learning identification method of the above-mentioned equation (2).

Incidentally, as the above-mentioned squaring circuit, for example, a multiplier can be used, and as the averaging circuit, a leaky integration circuit can be used.

〔Effect of the invention〕

As described above, according to the echo canceller according to the present invention, the level ratio of the echo component and the noise component is estimated from the received signal and the transmitted signal, and the update speed of the tap coefficient of the adaptive filter is controlled with respect to the target value of the amount of echo cancellation. Since an update speed control circuit is provided, even when the echo canceller is operated in an environment with near-end noise such as indoor noise, a constant amount of echo cancellation can be ensured regardless of level fluctuations of noise components and echo components.

[Brief explanation of drawings]

FIG. 1 is a block diagram of the principle configuration of an echo canceller according to the present invention, FIG. 2 is a time chart diagram for explaining the present invention in detail, and FIG. 3 is an embodiment of the echo canceller according to the present invention. FIG. 4 is a block diagram showing a conventional example. FIG. 5 is a time chart diagram for explaining the operation of the conventional example. In FIG. 1, l... Speaker, 2... Echo path, 3... Microphone, 4... Adaptive filter, 5... Update rate control circuit. In the figures, the same reference numerals indicate the same or corresponding parts.

Claims (1)

[Claims] The characteristics of the echo path (2) are estimated from the received signal to the speaker (1) and the transmitted signal from the microphone (3), a pseudo echo is generated, and this is subtracted from the echo component. a learning identification type adaptive filter (4) that cancels the echo component from the signal; and an adaptive filter (4) that estimates the level ratio of the echo component and the noise component from the received signal and the transmitted signal and adjusts the target value of the amount of echo cancellation. ) update speed coefficient value (α) of tap coefficient
An echo canceller comprising: an update speed control circuit (5) for controlling; and an echo canceller.