JPH0514476A - Loudspeaker telephone set - Google Patents

Abstract

PURPOSE:To improve the talking quality by prevent difficulty of talking in a sense of echo or howling caused when a drive noise or external noise in an automobile telephone set or the like or a background noise in a conference telephone system or the like is large. CONSTITUTION:The telephone set is provided with an acoustic echo/howling canceller 30 having a near end talker voice signal detection and howling detection function and a line echo/howling canceller 40. Then the acoustic echo/ howling canceller 30 cancels an acoustic echo EC1 and howling and the line echo/howling canceller 40 cancels a line echo EC2 and howling to attain hand- free talking with excellent naturality and high quality.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a mobile telephone such as an automobile, a conference telephone system, and the like, and to a loudspeaker telephone having a hands-free call function capable of making a call using a speaker and a micro without using a handset. Is.

[0002]

2. Description of the Related Art Conventionally, as this type of public address telephone, for example, there is one described in the following documents.

IEICE Technical Committee Material EA84-
41 (1984) Yamamori / Matsui / Makino, “Method for constructing a loudspeaker telephone system using microprocessor control” p. 15-2
1 For example, in a car phone, a driver may make a call while traveling, and from the viewpoint of safety and operability, a hands-free call function that uses a microphone and a loudspeaker to talk without a handset is provided. The situation is becoming essential. In addition to the video conference system and the conference telephone system, the hands-free call function is desired not only in ordinary telephones, but it is expected that it will become popular in the future.

Generally, in a loudspeaker telephone, due to acoustic echo (acoustic coupling) between a speaker and a microphone, in a loop loop formed by a side echo circuit or a line echo in a two-wire to four-wire conversion hybrid. A howling phenomenon called a melody may occur. Therefore, a loss circuit is inserted in order to reduce the loop gain of this loop loop, and it is controlled so that the howling phenomenon does not occur. An example of this type of public address telephone is shown in FIG.

FIG. 2 is a schematic configuration diagram of the conventional loudspeaker telephone described in the above document.

This loudspeaker telephone has a microphone 1 for inputting a voice signal of a caller (near-end speaker), and the microphone 1 includes an amplifier (hereinafter referred to as an amplifier) 2, a variable loss circuit 3, and an amplifier 4. , And the subscriber line 6 is connected via the anti-sidetone circuit and the two-wire / four-wire conversion hybrid 5. This subscriber line 6 is connected to an amplifier 7 via a sound-proof sound circuit or a two-wire / four-wire conversion hybrid 5, and the amplifier is connected to a partner (far-end speaker) via a variable loss circuit 8 and an amplifier 9. It is connected to the speaker 10 which outputs the audio signal from. The variable loss circuits 3 and 8 are configured to be controlled by the voice switch control circuit 11, the transmission signal smoothing circuit 12, and the reception signal smoothing circuit 13.

Next, the flow of the call voice signal of the public address telephone when the near-end talker makes a call or when the far-end talker receives a call, and howling and echo prevention control will be described.

The voice signal of the near-end speaker is input from the microphone 1, adjusted to a predetermined level by the amplifier 2, passed through the variable loss circuit 3 and amplified by the amplifier 4, and the side-tone sound circuit and
It is sent to the subscriber line 6 via the line-to-line conversion hybrid 5. On the other hand, the voice signal from the far-end speaker is amplified by the amplifier 7 from the subscriber line 6 via the protection side sound circuit and the 2-wire / 4-wire conversion hybrid 5, passes through the variable loss circuit 8, and is leveled by the amplifier 9. Is adjusted and output from the speaker 10,
The near-end speaker will hear.

Control of the variable loss circuits 3 and 8 inserted for preventing howling and echo is performed by the transmission signal smoothing circuit 12
And the reception signal smoothing circuit 13 monitors the power level of the transmission signal and the power level of the reception signal, respectively, and sends the monitoring information to the voice switch control circuit 11.
The voice switch control circuit 11 reduces the loss of the variable loss circuit 3 and increases the loss of the variable loss circuit 8 on the receiving side so that the loop gain does not rise when it determines that the transmission signal is dominant. In the opposite case, when the received signal is dominant, the loss of the variable loss circuit 8 is reduced and the loss of the variable loss circuit 3 on the transmitting side is increased, so that stable communication is controlled. The insertion / extraction of the loss during this period is controlled so that there is no discomfort as much as possible and the beginning or the end of the tail is not cut off as much as possible.

In the voice telephone of FIG. 2, the voice signal is processed as an analog signal as it is, but it may be replaced with a digital processing configuration. That is, the output of the amplifier 2 is an analog / digital converter (hereinafter referred to as an A / D converter).
Is converted into a digital signal through the variable loss circuit 3, converted into an analog signal by a digital / analog converter (hereinafter referred to as a D / A converter), and amplified by an amplifier 4. Further, the voice signal of the far-end talker from the subscriber line 6 is amplified by the amplifier 7, converted into a digital signal by the A / D converter, and the output thereof is passed to the variable loss circuit 8, and the variable loss circuit Even if the output of 8 is converted into an analog signal by the D / A converter and then amplified by the amplifier 9, the same effect as in FIG. 2 can be obtained.

[0011]

However, the conventional loudspeaker telephone has the following problems.

(A) For example, in an automobile telephone, when a call is made while traveling, if the traveling noise such as engine noise, tire ground contact noise, wind noise, or external noise is large, the signal detection of the transmission signal smoothing circuit 12 is performed. Level up.
Therefore, the voice switch control circuit 11 malfunctions so that the variable loss circuits 3 and 8 cannot be properly inserted / removed, the voice level is undulated, and the beginning or ending of the word is cut off, which is basically one-way communication. Making natural conversations difficult.

(B) Even in a conference telephone system or the like, if the background noise is large, the same problem as in (a) above occurs, and if the loop gain increases for some reason, an echo feeling or a howling phenomenon occurs. Calls may be difficult.

An object of the present invention is to provide a loudspeaker telephone which solves the problem that the above-mentioned prior art has, that the call quality is deteriorated when running noise, external noise, or background noise is large.

[0015]

In order to solve the above problems, a first invention is directed to a microphone for inputting a near-end talker voice signal and a speaker for outputting a far-end talker voice signal from a subscriber line. In a loudspeaker telephone that makes a call with a far-end speaker using and, an acoustic echo / howling canceller,
A line echo / howling canceller is provided. Here, the acoustic echo / howling canceller cancels acoustic echo and howling from the microphone with the far-end speaker voice signal as a reference signal, respectively.
Echo canceller and howling canceller having an adaptive digital filter (hereinafter referred to as ADF), and the first end by detecting the near-end talker voice signal and howling.
And a control unit for performing on / off control of the adaptive operation in the second ADF. The line echo / howling canceller has the same circuit configuration as the acoustic echo / howling canceller and is connected in anti-parallel thereto.

The second invention is the acoustic echo of the first invention /
The howling cancel function between the howling canceller and the line echo / howling canceller is configured by a single circuit.

[0017]

According to the first aspect of the present invention, since the loudspeaker telephone is configured as described above, in the acoustic echo / howling canceller, the echo canceller cancels the acoustic echo from the microphone by the on / off control of the controller. The howling canceller is controlled by the on / off control of the control unit.
Cancel howling. Similarly, in the line echo / howling canceller, the echo canceller therein cancels the line echo, and the howling canceller cancels the howling. As a result, the acoustic echo and the line echo are eliminated, and the howling state is further suppressed.

According to the second aspect of the invention, the howling cancel function is constituted by a single circuit, so that the circuit configuration can be simplified. Therefore, the above problem can be solved.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a schematic block diagram of a loudspeaker telephone showing an embodiment of the present invention.

This loudspeaker telephone has a microphone 20 for inputting a voice signal of a near-end speaker, and an amplifier 21, an A / D converter 22, an acoustic echo / howling canceller 30, a line echo / line echo / on its output side. Howling canceller 4
0, the D / A converter 50, the amplifier 51, and the side line circuit and the two-wire / four-wire conversion hybrid 52, and the subscriber line 5
3 is connected. On the other hand, an amplifier 54 is connected to the subscriber line 53 via a sound-proof side sound circuit or a two-wire / four-wire conversion hybrid 52, and the output side of the amplifier is an A / D converter 55, a line echo / howling canceller 40, Acoustic echo / howling canceller 30, D / A converter 60,
A speaker 62 that outputs a far-end talker signal from the subscriber line 53 is connected via the amplifier 61.

Acoustic echo / howling canceller 30
Is for canceling acoustic echo EC1 and howling between the speaker 62 and the microphone 20, and the first ADF
An echo canceller composed of 31 and an adder 32,
A howling canceller including a second ADF 33 and an adder 34, and an ADF control unit 35 that controls these two cancellers are provided.

The first ADF 31 uses the voice signal Rin (j) (j; time) of the far-end speaker output from the A / D converter 55 as a reference signal, and mainly outputs sound from the speaker 62 to the microphone 20. The transfer characteristic of the echo path is estimated, the acoustic echo cancellation signal CSin (j) is output to the adder 32, and the adder 32 outputs the output signal Si of the A / D converter 22.
It has a function of eliminating the acoustic echo EC1 contained in n (j).

Like the first ADF 31, the second ADF 33 uses the far-end talker voice signal Rin (j) as a reference signal and the howling cancellation signal CRes (j) as an adder 34.
And has a function of eliminating the howling component (tone-like signal) included in the estimated residual Res (j) output from the adder 32 by the adder 34.

The ADF control unit 35 uses the signals Sin (j),
The presence or absence of the input of the near-end talker voice signal Sin (j) is detected by monitoring the levels of Res (j) and Rin (j) to turn on / off the adaptive operation of the first and second ADFs 31 and 33. It has a control function.

Line echo / howling canceller 40
Has a function of canceling the line echo EC2 and howling from the amplifier 51 to the two-wire / four-wire conversion hybrid 52 to the amplifier 54, and has the same circuit configuration as the acoustic echo / howling canceller 30. It is connected in an anti-parallel state with respect to 30. That is, the circuit configuration is such that the adder 42 on the line echo canceller side and the adder 44 on the howling canceller side are connected to the output side of the A / D converter 55.

FIG. 3 is a block diagram of the first ADF 31 in FIG.

The first ADF 31 is a transversal type circuit, and includes delay elements 70-0 to 70- (n-1) for sequentially delaying the far-end talker voice signal Rin (j), and an estimated residual. Res (j) is input to tap coefficients h _{0 to} h _n−1 representing the impulse response (transfer characteristic) of the echo path and taps to the input / output signals of the delay elements 70-0 to 70- (n−1). Multiplier 71-0 for multiplying the coefficients h _{0 to} h _n-1
~ 71- (n-1) and each of the multipliers 71-0 to 71- (n
-1) output is added to cancel acoustic echo cancellation signal CSin
And an adder 72 that outputs (j) to the adder 32.

The number of taps of this ADF 31 is equal to that of the speaker 6
It is determined according to the impulse response length between 2 and the microphone 20. Then, the tap coefficients h _{0 to} h _n−1 are adaptively controlled by a learning identification method, a LMS (least mean square) method, or the like so that the mean square value of the estimated residual Res (j) is minimized, and the acoustic echo The cancellation signal CSin (j) is output from the adder 72, and the adder 32 outputs the signal Sin (j).
By subtracting the acoustic echo cancellation signal CSin (j) from (j), it functions to cancel the acoustic echo EC1.

The second ADF 33 of FIG. 1 also has the same circuit configuration as that of FIG. FIG. 4 shows the ADF control unit 3 in FIG.
It is a block diagram of FIG. The ADF control unit 35 performs on / off control for the adaptive operation of the first and second ADFs 31 and 33, and detects the input of the near-end speaker voice signal Sin (j). Detector 80 and estimated residual R
The output of the howling detector 81 for detecting the howling state by paying attention to the tone property and level amount which are the characteristics of howling using es (j), and the output of the near-end speaker voice signal detector 80 and the howling detector 81. Based on the first and second AD
The control signal generator 82 outputs control signals S82a and S82b for performing on / off control for the adaptive operation of F31 and 33.

The operation of the public address telephone configured as described above will be described.

In FIG. 1, when the near-end speaker makes a call,
The voice signal of the near-end speaker is input from the microphone 20, amplified by the amplifier 21 to a predetermined level, and converted to A /
The digital signal is converted by the D converter 22. From the converted signal Sin (j) (j; time), an acoustic echo EC is output by the acoustic echo / howling canceller 30.
1 and howling are canceled. The canceled signal passes through the line echo / howling canceller 40, is converted into an analog signal by the D / A converter 50, and is amplified to a predetermined level by the amplifier 51. Is sent to the subscriber line 53 via.

On the other hand, when a call is received from the far-end speaker, the voice signal from the far-end speaker is amplified from the subscriber line 53 via the protection side sound circuit and the two-wire / four-wire conversion hybrid 52. After being amplified to a predetermined level by 54, it is converted into a digital signal by the A / D converter 55. From this digital signal, the line echo EC2 and howling are canceled by the wiring echo / howling canceller 40, and the canceled far-end talker voice signal Rin (j) passes through the acoustic echo / howling canceller 30 and is D / A. It is converted into an analog signal by the converter 60. The converted analog signal is level-adjusted by the amplifier 61 and output from the speaker 62, and is heard by the near-end speaker.

The signal Si output from the A / D converter 22
n (j) can be regarded as an acoustic echo signal in the absence of near-end speaker speech. This signal Sin (j)
Is an echo canceller including a first ADF 31 and an adder 32 in the acoustic echo / howling canceller 30, and the acoustic echo EC1 is canceled. First A
The DF 31 uses the voice signal Rin (j) of the far-end speaker as a reference signal, estimates the transfer characteristic of the acoustic echo path mainly from the speaker 62 to the microphone 20, and adds the acoustic echo cancellation signal CSin (j) to the adder 32. Output to. Adder 3
2, the acoustic echo cancellation signal C from the signal Sin (j)
Subtract Sin (j) to eliminate acoustic echo EC1,
The estimated residual Res (j) is sent to the adder 34 that constitutes the howling canceller.

Howling detector 8 in ADF controller 35
When 1 detects the howling state, the adaptive control of the first ADF 31 is stopped by the control signal S82a output from the control signal generation unit 82 in the ADF control unit 35. Therefore, in the second ADF 33, the ADF control unit 3
Control signal S8 output from the control signal generator 82 in FIG.
2b, the same operation as that of the first ADF 31 is performed for a short time, and the howling cancellation signal CRes (j) is added to the adder 34.
Output to. In the adder 34, the howling cancellation signal CRes (j) is subtracted from the estimated residual Res (j), the howling component (tone signal) in the estimated residual Res (j) is deleted, and the estimated residual Output res (j). As a result, the public address telephone can be taken out of the howling state. Here, the operation of the ADF control unit 35 will be described with reference to FIG.

In the ADF control unit 35, the near-end speaker voice signal detector 80 outputs signals Sin (j), Rin (j),
By monitoring the level of Res (j), the input of the near-end talker voice signal Sin (j) is detected, and the detection result is notified to the control signal generation unit 82. The control signal generator 82 outputs the control signal S82a to output the first ADF3.
The adaptive operation of 1 is stopped. On the other hand, when the near-end talker voice signal detector 35 detects that the near-end talker voice signal Sin (j) has disappeared, the near-end talker voice signal detector 35 notifies the control signal generator 82 of the detection result, and the control signal generator 82 is the control signal S
82a causes the adaptive operation of the first ADF 31 to be performed.

The howling detector 81 calculates the estimated residual Res.
Using (j), the howling state is detected by paying attention to the tone property and the amount of level which are the characteristics of howling, and the detection result is notified to the control signal generation unit 82. Control signal generator 8
2 stops the adaptive operation of the first ADF 31 by the control signal S82a, and the second ADF 31 by the control signal S82b.
The adaptive operation of the DF33 is started.

Then, the second ADF 33 outputs the howling cancellation signal CRes (j) to the adder 34, and the howling is stopped by the subtraction operation of the adder 34. When the howling detector 81 detects that the howling state has disappeared, the control signal generating unit 82 is notified of the detection result, and the control signal generating unit 82 stops the adaptive operation of the second ADF 33 by the control signal S82b. Second AD
The output of the howling cancellation signal CRes (j) is stopped from F33. Then, the control signal generation unit 82 passes the control entry of the first ADF 31 to the near-end talker voice signal detector 35.

By the operation of the voice echo / howling canceller 30 as described above, the loudspeaker is controlled, and there is no echo feeling, and even if a howling state occurs, it can be quickly settled and good communication performance can be maintained.

The line echo / howling canceller 40 has the same circuit configuration as the acoustic echo / howling canceller 30 and is connected in a state in which the acoustic echo / howling canceller 30 is reversed. Similarly to the acoustic echo / howling canceller 30, the line echo EC2 and the howling are erased instead of the acoustic echo EC1.

In this embodiment, the acoustic echo / howling canceller 30 and the line echo / howling canceller 40 are used.
Since it is provided, acoustic echo EC1 and line echo EC2
, And completely suppress the unexpected howling state. Therefore, a natural and high quality hands-free call becomes possible, and a high performance loudspeaker can be provided.

The present invention is not limited to the above embodiment,
Various modifications are possible. The following are examples of such modifications.

(I) In FIG. 1, howling cancellers are provided in the acoustic echo / howling canceller 30 and the line echo / howling canceller 40, respectively. However, this howling canceller is provided in either one,
The howling canceller may be commonly used. As a result, the circuit configurations in the acoustic echo / howling canceller 30 and the line echo / howling canceller 40 can be simplified. (Ii) The first and second ADFs 31 and 33 shown in FIG.
In the above, the transversal type is used, but other configurations may be used, and adaptive control of the tap coefficient may be performed using a method other than the learning identification method and the LMS method.

[0043]

As described in detail above, according to the first aspect of the present invention, since the voice echo / howling canceller and the line echo / howling canceller are provided, the acoustic echo and the line echo can be canceled and expected. Howling can be completely suppressed. Therefore, even if there is a large amount of running noise or external noise in an automobile telephone or background noise in a conference telephone system or the like, a natural and high quality hands-free call becomes possible, and the performance of the loudspeaker telephone is significantly improved.

According to the second invention, since the howling canceller functions of the acoustic echo / howling canceller and the line echo / howling canceller are constituted by a single circuit, the circuit constitution can be simplified.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram of a loudspeaker telephone showing an embodiment of the present invention.

FIG. 2 is a block diagram of a conventional public address telephone.

FIG. 3 is a configuration diagram of an ADF in FIG.

FIG. 4 is a configuration diagram of an ADF control unit in FIG.

[Explanation of symbols]

20 microphones 22,55 A / D converter 30 Acoustic Echo / Howling Canceller 31, 33 First and second ADF 32,34 adder 35 ADF control unit 40 line echo / howling canceller 42,44 adder 50,60 D / A converter 52 2-wire 4-wire conversion hybrid 62 speakers EC1 acoustic echo EC2 line echo 80 Near-end speaker voice signal detector 81 Howling detector 82 Control signal generator

Claims (2)

[Claims] 1. A loudspeaker telephone for making a call with a far-end talker using a microphone for inputting a near-end talker voice signal and a speaker for outputting a far-end talker voice signal from a subscriber line. An echo canceller and a howling canceller having first and second adaptive digital filters for canceling acoustic echo and howling from the microphone, respectively, using the far-end speaker voice signal as a reference signal, and the near-end speaker voice signal and An acoustic echo / howling canceller having a controller for detecting howling and performing on / off control of adaptive operation in the first and second adaptive digital filters; and an acoustic echo / howling canceller having the same circuit configuration. A loudspeaker system characterized in that it is provided with a line echo / howling canceller connected in antiparallel. Talk machine. 2. A loudspeaker telephone characterized in that the acoustic echo / howling canceller and the line echo / howling canceller according to claim 1 are provided with a howling cancel function in a single circuit.