JPH0329424A - Echo canceller circuit for hand-free telephone call - Google Patents

Abstract

PURPOSE:To ensure hand-free communication with high quality by providing a line echo canceller synchronizing a cancel signal of a line echo synchronously with the line echo so as to erase effectively the line echo including a time change element such as frequency offset or phase jitter thereby suppressing howling or the line echo. CONSTITUTION:A pseudo signal to cancel an acoustic echo generated from a received sound outputted from a speaker reflected indoor and inputted from a microphone 1, is generated, and an acoustic echo canceller 20 is provided to subtract the simulated signal from the transmission signal. Moreover, a line echo canceller 21 is provided, in which simulation signal cancelling the line echo being the transmission signal inputted from the microphone 1 and returned while being reflected in the station is generated synchronously with the line echo and the simulation signal is subtracted from the reception signal. Thus, even in the presence of the line echo having the time charge element component, the effect by the time change element is not caused, not only the sound echo but also the line echo is cancelled effectively to realize the excellent communication quality.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to an echo canceller circuit used in a hands-free telephone for making calls indoors using a speaker and a microphone, and particularly relates to an echo canceller circuit that generates a pseudo echo. Regarding improved performance.

[Conventional technology] Conventionally, as an echo canceller for hands-free telephones,
For example, there was one disclosed in ``Hands-Free Car Telephone Terminal Using Echo Canceller'' (1985, National Conference of the Institute of Electronics and Communication Engineers). The principle will be briefly explained using FIG.

This figure shows a case where a speaker in a car (hereinafter referred to as a mobile speaker) is talking to a general subscriber (hereinafter referred to as a far-end speaker) in a hands-free passage.

The voice signal R of the far end speaker entering from the terminal 12 is transmitted to the public communication network P S T N via the hybrid node 11 in the subscriber station.
(Public Switching Teleph
from the base station 9 via the antennas 13 and 14 to the radio device 6 in the vehicle, and is output from the speaker 2 as the far-end speaker's voice to be heard by the mobile speaker.

Further, this far-end speaker's voice outputted from the speaker 2 is reflected by the window glass inside the car, becomes an acoustic echo E, and is inputted from the microphone 1. This acoustic echo signal S converted into an electrical signal is superimposed on the mobile speaker's voice signal N and is transmitted from the terminal 12 to the far end speaker via the radio 6, antennas 7 and 8, base station 9, PSTNIO, and hybrid 1l. reach At this time, the far-end speaker has difficulty communicating due to communication interference caused by the acoustic echo EA in which his/her own voice returns after making one loop.

In extreme cases where the mobile speaker's voice signal N manually input from the microphone 1 is reflected by the subscriber station's hybrid 1l and returns to the receiving side, the line echo EC is particularly large, and the loop gain exceeds 1, causing howling. may occur.

Therefore, by introducing the Echo Ushi Jansera 5, the Acoustic Echo EA
cancel.

In terms of operation, a pseudo echo signal S for canceling the acoustic echo signal S is generated by the adaptive filter 4 which receives the far-end speaker voice signal R as an input signal, and in addition 2g3, an acoustic signal is added to the mobile speaker voice signal N. The acoustic echo is canceled by subtracting the pseudo echo signal S from the superimposed signal on which the echo signal S is superimposed.

A well-known learning identification method is used for adaptive control of the adaptive filter 4 so that the residual signal e remaining without being removed in the so-called addition 3 is minimized. That is, the echo canceller 5 estimates and controls the transfer characteristic (-impulse response) from the speaker 2 to the microphone 1 one after another every sampling period.

[Problems to be Solved by the Invention] However, the above-described conventional method has the following drawbacks.

(1) The echo canceller only cancels acoustic echoes, but if the line echo is large, the acoustic echo characteristics change and the amount of echo cancellation decreases even if you cancel only the acoustic echoes. However, even if the voice does not cause howling, the mobile speaker's own voice will be heard again, and the situation will be as if they were talking in the bathroom.

(2) In addition, a canceller similar to the acoustic echo canceller is introduced on the line side as a line echo canceller to generate a line echo pseudo signal for canceling the line echo,
By subtracting this, it is possible to eliminate the line echo as well. However, if the line echo component contains time-varying elements such as frequency offset and phase jitter that occur when transmitting and receiving carriers are shifted during modulation and demodulation on the wireless line, line echo may occur on the line side. Even if a canceller is simply introduced, line echoes that include time-varying elements cannot be sufficiently canceled because time-varying elements are not taken into consideration.

An object of the present invention is to separately generate an acoustic echo pseudo signal and a line echo pseudo signal using different filters, and to synchronize the line echo pseudo signal with the line echo signal. Even if there is a line echo with a time-varying element component, it effectively cancels not only the acoustic echo but also the line echo without being affected by the time-varying element, resulting in an excellent communication product. An object of the present invention is to provide an echo canceller circuit for hands-free telephones that can realize high quality.

[Means for Solving the Problems] The echo canceller circuit of the present invention removes echo components superimposed on the call signal by subtracting pseudo echoes from the call signal when a call is made using a speaker and a microphone. It is configured as follows.

In such a configuration, the received sound output from the speaker is reflected inside the room, producing a pseudo signal to cancel the acoustic echo generated by manual input from the microphone, and subtracting this from the transmitted signal. Equipped with a canceller.

In addition, in this acoustic echo canceller, a pseudo signal is generated in synchronization with the line echo to cancel the line echo of the transmitted voice input from the microphone reflected within the station, and this is subtracted from the received signal. Equipped with a canceller.

[Effects] Because the line echo canceller is provided separately from the acoustic echo canceller, even if the acoustic echo characteristics change and the amount of echo cancellation in the acoustic echo canceller decreases, the line echo canceller will eliminate the line echo. is canceled out, so there is no howling, and the speaker does not hear his own voice coming back to him.

In addition, when a received signal with line echo superimposed is input to a line echo canceller, a pseudo signal is generated to cancel the line echo, but since this pseudo signal is synchronized with the line echo, frequency offset, phase jitter, etc. Even if the line echo includes a time-varying element of , the line echo and the line echo pseudo signal match in phase.

Therefore, when the line echo pseudo signal is subtracted from the received signal, the line echo component is also sufficiently canceled from the received signal.

[Example] An example of the present invention will be described below with reference to FIGS. 1 and 3.

FIG. 1 shows an example of an echo canceller circuit for a hands-free car telephone according to the present invention. The present embodiment differs from the conventional example shown in FIG. 2 in the echo canceller, and other parts having the same functions as the conventional example are given the same reference numerals and detailed explanations thereof will be omitted.

The echo canceller is divided into an acoustic echo canceller 20 that cancels acoustic echoes and a line echo canceller 2l that cancels line echoes. The acoustic echo canceller 20 receives as input a branched signal of the far-end speaker's voice signal k that is input to the speaker 2 via the radio device 6 . The acoustic echo canceller 20 is provided with an acoustic echo canceling adaptive filter 20 that receives this branched signal as input, and generates an acoustic echo canceling signal (acoustic echo pseudo signal) SA that minimizes the residual signal eA. to occur. Then, in the adder 3, an acoustic echo cancellation signal S is obtained from a signal obtained by superimposing the acoustic echo signal S on the mobile speaker voice signal N.
The acoustic echo is canceled by subtracting . Since this acoustic echo canceller 20 has the same functions as the echo canceller 5 shown in FIG. 2, the internal configurations are given the same reference numerals. However, since the acoustic echo is specifically canceled, the acoustic echo signal is SA+, the acoustic echo cancellation signal is SA, and the residual signal is eA, and so on.
(Acoustic) has been added.

Further, a line echo canceller 2l provided separately from the acoustic echo canceller 20 manually generates a branch signal of the mobile speaker's voice signal N which is converted into an electrical signal by the microphone l and inputted to the radio device 6. Line echo canceller 2
1, a phase follower circuit 22 as a synchronous circuit is provided, and a branch signal is manually input to the phase follower circuit 22 to generate an error signal e.
. Synchronize with. In addition, an adaptive filter 23 for line echo cancellation using this synchronization signal as a human power is provided, and a line echo cancellation signal (pseudo signal of line echo) for canceling the line echo so as to minimize the error signal ec.
Generates Sc. Then, in the adder 24, the line echo cancellation signal Sc is subtracted from the superimposed signal of the far-end speaker's voice signal R output from the radio device 6 and the line echo signal Sc, thereby canceling the line echo EC. the result,
From the addition field 24, the far end speaker's voice signal R and the error signal ec of the line echo remaining without being removed are obtained.

Here, the phase tracking circuit 2 in the line echo canceller 21
2 and the adaptive filter 23, the addition 2g2
Error signal e determined by 4. {J' is being paid.

Note that the subscript C of S C+ S C+ eC is the line (
circuit). Further, the mobile speaker voice signal superimposed with the acoustic echo signal is the transmission signal of the present invention, the far end speaker voice signal superimposed with the line echo signal is the reception signal, and the sum of these signals is the call signal. It is.

The adaptive filter 23 is preferably an acyclic sacred digital filter from the viewpoint of convergence and stability. Specifically, it is appropriate to use a transversal type. As for the adaptive control algorithm, since it is connected to the phase tracking circuit 22, it is preferable to use the steepest descent method or the like.

Further, an f-th tracking circuit 22 as a synchronization circuit is provided at the input stage of the line echo canceller adaptive filter 23.
As the method, APC (Automation
Various methods are possible, such as c Phase Control), crystal filter method (ringing method), burst injection lock method, etc., but here the circuit having the configuration shown in FIG. 3 is used.

jij (The mobile speaker signal N led to the line machine 6 is branched into two. One of them is passed through a Hilbert transform filter 30 to generate an orthogonal component, and this is manually input to one multiplier 3l. One multiplier 31 Then, the output S of the function generator 33
Multiply the phase compensation amount by INφ. The other signal is input directly to the other multiplication generator 32 and is then input to the function generator 34.
The phase compensation amount is multiplied by the output COSφ. These multiplication results are then added together by an adder 35 and input to the adaptive filter 23 for line echo cancellation.

Here, COS for phase compensation by function generator 33.34
Adaptive control of φ, SINφ (φ is the frequency of the voice band and includes the frequency offset and the phase jitter compensation component Δφ) is performed using the error signal eC and the line echo cancellation signal Sc.
In the adaptive phase shift 2336 which takes as input and outputs a, it is performed based on the following equation.

$ 1 $ n-1-βF(Sc- ec)ln-1-(
Δφ)...(1) Here, n represents time and I? (S C- e c) is a function of SCXeC, and β is 1 (represents a positive coefficient).

Adaptive control of these adaptive filter 23 and adaptive phase shift 2336 is performed using an error signal e. This is done by widening the steepest descent so that the distance is the minimum.

Next, the operation of the above structure will be explained.

The far-end speaker's voice signal R is guided to the acoustic echo canceller 20 and inputted to the acoustic echo canceller adaptive filter 4, from which an acoustic echo cancellation signal SA is generated.

This acoustic echo L signal SA and the mobile speaker speech signal N on which the acoustic echo signal S has been superimposed are added to an adder 3, and the mobile speaker speech signal N from which the acoustic echo has been removed (the remaining A difference signal eA) is output from the acoustic echo canceller 20.

The mobile speaker's voice signal N from which the acoustic echo has been removed by the acoustic echo canceller 20 is further guided to the line echo canceller 2l and inputted to the phase tracking circuit 22.

Here, during modulation and demodulation over a wireless line, the compensation for time-varying elements such as frequency offset that occurs when the transmitting and receiving carriers are shifted is calculated as described above, and the mobile signal that includes this compensation is calculated. A voice signal is created. This audio signal is further inputted to the line echo canceller adaptive filter 23, which produces a line echo cancellation signal S that follows the line echo signal Sc every 1,000 times. is generated.

On the other hand, the far-end speaker's voice signal R is a line echo signal S generated in the hybrid 11 within the subscriber station. is superimposed and PS
The signal passes through the TNIQ, is input from the base station 9 through the antenna 13 and i4, and is input into the line effect canceller 21 from the in-vehicle transmission line machine 6, and reaches its addition L''B24.

This adder 24 outputs either the phase tracking line echo cancellation signal SC generated by the adaptive filter 23 or the superimposed signal (R1S).

) to obtain the far-end speaker's voice signal R and an error signal eC, which is the residual line echo. The far-end speaker's voice signal R is output from the speaker 2 as the far-end speaker's voice and is heard by the mobile speaker.

The obtained error signal e. is inputted to the adaptive filter 23 and the phase tracking circuit 22, and is adaptively controlled by these so that the error signal ec is minimized.

This adaptive control can use the steepest descent method, etc., as described above.

In this way, the line echo canceller 21 inputs the mobile speaker's voice signal N, and firstly receives the mobile speaker's voice signal N.
A pseudo echo So is generated that follows the phase of the line echo signal SC that is reflected back at the line echo signal SC. Next, this pseudo echo So
is subtracted in the IJcJ calculation 24, and the line echo signal SC containing the frequency offset is deleted from the far end speaker's voice signal R.

As described above, according to the present embodiment, in addition to the acoustic echo canceller 20, a line echo canceller 2l for removing line echo Ec is provided, and this line echo canceller 2l is used to eliminate time-varying element components such as frequency offset and phase jitter. a phase tracking circuit 22 for phase-following the mobile speaker voice signal N to a line echo having a line echo; and a line echo cancellation signal Sc for canceling the line echo from the mobile speaker voice signal N phase-following the line echo. It consists of an adaptive filter for a line echo canceller that generates

Therefore, when the line echo is large, the acoustic echo characteristics change and the echo canceller 20
Even when l1 decreases, line echo canceller 2
1 effectively cancels out the acoustic echo that circulates as a line echo, and therefore prevents the sound from the speaker 2 from entering the microphone 1 and being amplified, resulting in an oscillation state. As a result, there will be no feedback or your own voice will come back to you, making it seem like you are talking in the bathroom.

Furthermore, the line echo cancellation signal SC is made to follow the phase of the line echo signal Sc, which has time-varying components such as frequency offset and phase jitter, so that both signals are completely synchronized. Line echoes having variable element components can be sufficiently canceled.

In the above-described embodiment, a case has been described in which the phase tracking circuit is provided at the input stage of the line echo canceller adaptive filter, but it may also be provided at the output stage of the adaptive filter.

In addition, although the case has been described in which the control signal to the phase tracking circuit is an error signal that is a line echo component that remains without being removed, the present invention is not limited to this, and may include time-varying elements. For example, a far-end speaker's voice signal superimposed with a line echo signal may be used.

Furthermore, the present invention is not limited to application to the car telephone described in the embodiment, but can be widely applied to telephone calls made indoors using a microphone and a speaker.

[Effects of the Invention] According to the present invention, there is provided a line echo canceller that synchronizes a line echo cancellation signal with the line echo to effectively cancel line echo including time-varying elements such as frequency offset and phase difference. This feature suppresses howling and line echo, ensuring high-quality hands-free calls.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing an embodiment of the echo canceller circuit for a hands-free car phone according to the present invention, Fig. 2 is a block diagram showing a conventional example of an echo canceller circuit for a hands-free car phone, and Fig. 3 1 is a flow diagram showing an example of a phase tracking circuit used in this embodiment. 1 is a microphone, 2 is a speaker, 20 is an acoustic echo canceller, 2l is a line echo canceller, 22 is a phase tracking circuit, 23 is an adaptive filter for line echo cancellation, EA is an acoustic amplifier, SA is an acoustic echo signal, and SA is an acoustic echo. A pseudo signal of the signal, N is a mobile speaker voice signal which becomes a transmission signal together with SA, Ec is a line echo, Sc is a line echo signal, Sc is a pseudo signal of a line echo signal, R
is the far end speaker's voice signal which becomes the received signal together with SC. 7'Oy diagram of an example of a phase tracking circuit Figure 3 Procedural official document (spontaneous) October 30, 1989

Claims (1)

[Claims] An echo canceller circuit for a hands-free telephone that removes echo components superimposed on the call signal by subtracting pseudo echoes from the call signal when a call is made using a speaker and a microphone. There is an acoustic echo canceller that generates a pseudo acoustic echo signal that is generated when the output received sound is reflected indoors and input from the microphone, and subtracts it from the transmitted signal. An echo canceller circuit for a hands-free telephone, comprising a line echo canceller that generates a line echo pseudo signal that is reflected back in synchronization with the line echo, and subtracts this from a received signal.