JP2007288372A - Voice switch control method and device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To increase the sound volume of transmitted and received voice without impairing bidirectionality during double talk in the case that a caller terminal performs hands-free communication and a callee terminal also performs hands-free communication. <P>SOLUTION: The voice switch control device includes: an echo canceler 3 which removes acoustic echo passing around to a microphone from a speaker for converting a reception signal to voice; a power calculation parts which obtains respective powers of a reception input signal from the callee terminal a transmission signal and an error signal from the caller terminal; and a voice switch 20 which determines the presence or the absence of transmitted and received voice in accordance with the reception input signal power and the error signal power and obtains the amount of echo cancellation from a power ratio of the error signal power and the transmission signal power and obtains the amount of transmission loss insertion and the amount of reception loss insertion on the basis of the amount of echo cancellation and multiplies the error signal by the amount of transmission loss insertion to obtain a transmission output signal to the callee terminal and multiplies the reception input signal from the callee terminal by the amount of reception loss insertion to obtain a reception signal of the caller terminal. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a voice switch control method and apparatus. Specifically, it is intended to provide a novel voice switch control method and apparatus in a communication apparatus capable of hands-free communication.

  FIG. 9 is a loop gain diagram for explaining the loop gain. When a hands-free call is made using the call device shown here, it is necessary to design the loop gain formed by reverberation or the like to be 0 dB or less in order to prevent howling. In the case of being configured only by the 4-wire system of the same figure without the hybrid circuit of 2-wire to 4-wire conversion, it is formed by wraparound from the counterpart terminal and wraparound by the acoustic path between the speaker S and the microphone M of the own terminal. The loop gain is set to 0 dB or less. At this time, considering that the other terminal may also make a hands-free call, the gain from the input terminal T1 to the output terminal T2 of the own terminal needs to be 0 dB or less. Here, it is assumed that the acoustic wraparound attenuation amount due to the acoustic path between the speaker S and the microphone M is ZdB.

  FIG. 10 shows a circuit configuration diagram of a conventional communication device. The terminal shown in the figure is used as a terminal of a conference device or a button telephone device capable of hands-free calling. Amplifiers 91 and 96 are inserted so that the speaker can talk even if the speaker is about 0.5 to 1 m away from the speaker S or microphone M of the terminal. Similarly, the amplifiers 92 and 95 are inserted on the partner terminal side. Between the own terminal and the partner terminal, a hybrid circuit that performs two-wire to four-wire conversion (not shown) is interposed in each of the own terminal and the partner terminal.

  Here, assuming that the transmission system gain of the amplifier 91 is YdB, the reception system gain of the amplifier 96 is XdB, and the acoustic wraparound attenuation is ZdB, in general, when the echo canceller 3 and the voice switch 20B are not present, The gain X + Y-Z of the own terminal becomes larger than 0 dB. The same applies to the counterpart terminal side. When such a terminal and the partner terminal are connected, the loop gain described with reference to FIG. 9 becomes larger than 0 dB, so that howling occurs and the call cannot be made.

  Therefore, in order to enable a hands-free call, it is necessary to mount an echo canceller 3 for canceling the echo and a voice switch 20B inserted to give a loss to the voice. When the echo canceling amount of the echo canceller 3 is α dB and the loss insertion amount of the voice switch 20B is β dB, the design is such that X + YZ-α-β <0. The acoustic wraparound attenuation ZdB of the terminal itself and the maximum echo cancellation amount αdB of the echo canceller 3 are obtained by measurement. Finally, it is necessary to determine the values for the transmission system gain YdB, the reception system gain XdB, and the loss insertion amount βdB of the voice switch 20B.

  FIG. 11 shows a loss insertion point of the conventional voice switch 20B. When both transmission and reception are silent, a loss is inserted into the transmission side. In the case of single talk where the transmission is silent and the reception is only voiced, a loss is inserted into the transmission side. In the case of a single talk in which the incoming call is silent and only the outgoing voice is sent, a loss is inserted into the receiving side. In other words, there is no problem because a loss is inserted on the silent side during single talk.

  However, since it is possible to set in advance whether or not to prioritize the incoming call or the outgoing call in the call device, if the incoming call priority method is used in the case of double talk where both the incoming and outgoing voices are voiced, Since the loss is inserted into the transmission side, the influence of the insertion appears on the other party's reception, and when the transmission priority method is used, the loss is inserted into the reception side. Therefore, the effect of insertion appears in the reception on the terminal side.

  Normally, experience shows that even if a loss of about 6 dB is inserted during a double talk, the caller is hardly noticed and can maintain bidirectionality. However, if a loss of 6 dB or more is inserted, the voice switch 20B The presence of the operation becomes prominent, and a switching feeling (abrupt change in volume) exceeding the loss insertion amount is obtained. When the loss insertion is about 12 dB, the influence of insertion becomes strong and double talk becomes difficult. When it becomes a loss insertion of about 15 dB, the influence of the insertion appears more strongly and it almost feels like a single talk.

  When placing importance on the quality of calls during double talk, the loss insertion amount of the voice switch 20B should be set to about 6 dB when the echo canceller 3 converges so that the loss insertion is hardly noticed by the caller. It is common. Since the echo canceling amount of the echo canceller 3 is likely to fluctuate depending on the call environment, the design is made with a certain margin in order to determine the convergence.

  If the margin is reduced, the volume of the transmission / reception can be increased. However, if the operation of the echo canceller 3 becomes unstable due to fluctuations in the communication environment and does not converge, the loss insertion amount of the voice switch 20B is reduced. Since the volume is increased, the volume of the incoming call at the other terminal is reduced and the call quality deteriorates. Also, if the margin is increased, the loss insertion amount of the voice switch 20B can be kept small even if the echo cancellation amount of the echo canceller 3 is slightly changed. The volume will be insufficient.

Patent Document 1 discloses an echo suppressor including a continuous tone switch and a multiplier. According to this echo suppressor control method and apparatus, when the continuity of the received signal is detected, the transmitted signal is suppressed regardless of whether or not the transmitted signal exists. As a result, the dial tone is not interrupted or fluctuated. When the continuity of the received signal is not detected, the received signal is suppressed during the period in which the transmitted signal is present to prevent the occurrence of echo due to the wraparound of the transmitted signal (Patent Document 1, paragraph 0012).
JP 2005-260763 A

  When the own terminal makes a call using the hands-free call device, the other terminal also makes a call using the hands-free call device, and the other terminal makes a call using the handset call device. In both cases, when the operation of the echo canceller 3 becomes unstable due to fluctuations in the call environment and does not converge, the voice switch quality is deteriorated because the loss insertion amount of the voice switch is increased. Further, if the loss insertion amount of the voice switch is reduced, the amount of echo cancellation becomes insufficient, and the volume of transmission / reception is suppressed, resulting in insufficient volume.

  Accordingly, the problem to be solved by the present invention is that the bidirectionality at the time of double talk is lost when the terminal performs a call using a hands-free call device and the partner terminal also performs a call using a hands-free call device or a handset device. In other words, the volume of the transmitted / received speech is made larger than that of the conventional example.

An echo canceller means for removing acoustic echo that circulates from the speaker that converts the received signal into speech and circulates into the microphone;
Power calculation means for obtaining the power of the incoming input signal from the counterpart terminal side, the transmission signal from the own terminal side and the error signal;
Judgment of presence / absence of transmitted / received sound based on received signal power and error signal power, echo cancellation amount obtained from power ratio of error signal power and transmitted signal power, and transmission loss insertion based on this And the received loss insertion amount, the error signal is multiplied by the transmission loss insertion amount to obtain the transmission output signal to the partner terminal side, and the reception loss insertion amount is set in the reception input signal from the partner terminal side. A voice switch control device including voice switch means for multiplying and obtaining a reception signal on the own terminal side is constructed.

  Even if the loss insertion amount of the voice switch during double talk is a fairly large value of about 6 to 12 dB, the loss insertion amount is distributed to both transmission and reception. As a result, the loss insertion amount on the transmitting side or the receiving side can be suppressed to about 6 dB to 8 dB. Therefore, even when the other terminal makes a call using a hands-free call device, or a call using a handset call device. Even in the case where the call is performed, since the bidirectionality and the transmission / reception gain are improved, call quality in the hands-free call is improved.

An echo canceller means for removing acoustic echo that circulates from the speaker that converts the received signal into speech and circulates into the microphone;
Power calculation means for obtaining the power of the incoming input signal from the counterpart terminal side, the transmission signal from the own terminal side and the error signal;
The presence / absence of transmitted / received sound, that is, silence, single talk or double talk, is judged from the received signal power and error signal power, and the echo cancellation amount is obtained from the power ratio of the error signal power and the transmitted signal power. Based on this, the transmission loss insertion amount and the reception loss insertion amount are obtained, and the error signal is multiplied by the transmission loss insertion amount to obtain the transmission output signal to the partner terminal side. A voice switch control device is provided that includes voice switch means for multiplying the received input signal from the side by the received loss insertion amount to obtain the received signal on the terminal side.

  FIG. 1 is a block diagram showing an embodiment of the present invention. Corresponding symbols are attached to components corresponding to the components shown in FIG. Here, the description of the hybrid circuit that performs 2-wire to 4-wire conversion, the microphone M, the speaker S, the amplifiers 91 and 96, and the counterpart terminal side is omitted. An echo canceller 3, a power calculator 10 and a voice switch 20 are included.

  The echo canceller 3 includes an adaptive filter 5 and a subtracter 9. The error signal e and the received signal x are applied to the adaptive filter 5, and a pseudo echo signal y 'is created by a known method. The subtracter 9 receives the transmission signal y and the pseudo echo signal y ′ from the microphone M (not shown), subtracts the pseudo echo signal y ′ from the transmission signal y, and generates an error signal e (= y− y ′) is output. The transmitted signal y includes an acoustic echo component that circulates as sound from a speaker S (not shown) of the received signal x.

A transmission signal y, an error signal e, and a reception input signal r _in are applied to the power calculation unit 10, and each power is constantly calculated at intervals of 125 μs, for example, in the case of 8 kHz sampling, and the transmission signal power is calculated. P _y, error signal power _{P e} and received input signal power _{P rin} is obtained as an output. The voice switch 20 includes a transmission multiplier 21, a reception multiplier 22, and a loss insertion amount calculation unit 25. The loss insertion amount calculation unit 25 outputs the transmission loss insertion amount L _t and the reception loss insertion amount L _r .

The transmission multiplier 21 outputs a transmission output signal t _out (= e × L _t ) by multiplying the error signal e by a transmission loss insertion amount L _t which is 1 or less. The reception multiplier 22 multiplies the reception input signal r _in by a reception loss insertion amount L _r which is a value equal to or less than 1, and outputs a reception signal x (= r _in × L _r ).

FIG. 2 is a circuit configuration diagram showing details of the internal configuration of the power calculation unit. The received signal power calculator 12, the transmitted signal power calculator 13, and the error signal power calculator 14 are included therein. The received input signal power calculator 12 received input signal r _in is applied, received input signal power P _rin its power is obtained as an output. The transmission signal power calculator 13 transmission signal y is applied, the resulting transmission signal power P _y is its power as output. An error signal e is applied to the error signal power calculator 14, and an error signal power _{Pe that} is the power is obtained as an output.

FIG. 3 is a circuit configuration diagram showing details of the internal configuration of the loss insertion amount calculation unit 25. As shown in FIG. There are a call state determination unit 27 and a loss insertion amount determination unit 28. The loss insertion amount determination unit 28 includes a power ratio calculator 31, an echo canceller 32, and a loss insertion amount distribution unit 33. The call state determination unit 27 received input signal power P _rin and the error signal power P _e is applied, silence, receiving only sound of the receiver single talk state, transmitting only voice of the transmitter single-talk state, or A call state determination signal 35 is output by determining which of the double talk state of the transmission / reception voice is in the call state.

The power ratio calculator 31 included in the loss insertion amount determining section 28, a call state determination signal 35, the error signal power P _e and transmission signal power P _y is applied, when the receiver single talk state, the error signal The power ratio (P _e / P _y ) between the power P _e and the transmission signal power P _y is calculated, and the power ratio signal 36 is output at regular intervals of 1 ms, for example. The power ratio signal 36 is applied to the echo canceller 32, and it is determined that the operation of the echo canceller 3 has not converged until it falls below a predetermined threshold value. Notify 33. In the loss insertion amount distribution unit 33 that has received the echo cancellation amount signal 37 and the call state determination signal 35, the maximum value is set as the initial value of the loss insertion amount.

When the value indicated by the echo cancellation amount signal 37 is equal to or smaller than a predetermined threshold value, the loss insertion amount distribution unit 33 determines the transmission loss insertion amount L _t and the reception loss according to the call state indicated by the call state determination signal 35. The insertion amount _Lr is determined and output. The transmission loss insertion amount L _t is applied to the transmission multiplier 21, and a transmission output signal t _out is obtained by multiplying the error signal e by the value indicated by the transmission loss insertion amount L _t . The received loss insertion amount L _r is applied to the receiver multiplier 22, to obtain a received signal x by multiplying a value indicated by the received loss insertion amount L _r to the received input signal r _in.

  FIG. 4 is a loss insertion location diagram showing the loss insertion location of the voice switch. That is, the operation of the loss insertion amount distributor 33 in FIG. 3 is represented. First, the loss insertion amount is determined from the value indicated by the echo cancellation amount signal 37. In this example, it is divided into three stages of 0 to 6 dB, 6 to 12 dB, and 12 dB or more. Here, when the call state indicated by the call state determination signal 35 is transmission / reception silence, reception single talk, and transmission single talk, the same loss insertion location as the conventional example shown in FIG. 11 is obtained. Even when the loss insertion amount at the time of double talk is 0 to 6 dB and 12 dB or more, the loss insertion position is the same as that of the conventional example shown in FIG. That is, the loss insertion location is only one of the transmission side and the reception side.

  FIG. 5 is a loss insertion amount distribution diagram in which the loss insertion amount is 6 dB to 12 dB in FIG. The amount of insertion is switched between the receiving side and the transmitting side depending on whether the call device is set by selecting the receiving priority method or by selecting the transmitting priority method. For example, when the loss insertion amount is 6 dB, the transmission side insertion amount is 6 (0) dB and the reception side insertion amount is 0 (6) dB in the reception (transmission) speech priority method.

  For example, when the loss insertion amount is 9 dB, the transmission side insertion amount is 6 (3) dB and the reception side insertion amount is 3 (6) dB in the case of the reception (transmission) speech priority method. Is 12 dB, in the case of the reception (transmission) speech priority system, the transmission side insertion amount is 8 (4) dB, and the reception side insertion amount is 4 (8) dB. The reason why the loss insertion amount is not distributed when the loss insertion amount at the time of double talk is 12 dB or more is that, if it is distributed, the call quality is significantly deteriorated for both transmission and reception. This is because if one is not distributed, only one of the transmission / reception is prioritized, and the prioritized one can maintain good call quality.

FIG. 6 is a call state determination diagram for determining the call state of the voice switch in FIG. This call state determination is performed in the call state determination unit 27 of FIG. Since the call state determination unit 27 received input signal power P _rin and the error signal power P _e is applied, received input signal power P _rin and the error signal power P _e is if both below the predetermined value, determines that the silent state To do. If received input signal power P _rin is the error signal power P _e equal to or higher than the predetermined value or less than a predetermined value, it is determined that the receiver single talk state of the receiving only sound. If received input signal power P _rin error signal power P _e is more than a predetermined value below the predetermined value, it is determined that the transmission single talk state of the voice only transmission. If received input signal power P _rin and the error signal power P _e are both higher than a predetermined value, it is determined that the double-talk state of the handset voice.

7 and 8 are flowcharts showing the flow of the call operation of the apparatus shown in FIG. When the call operation is started, the power calculation unit 10 uses the transmission signal y, the error signal e, and the reception input signal r _in to transmit the transmission signal power P _y , the error signal power P _e, and the reception. The input signal power P _rin is obtained as an output (S1, FIG. 7). Received input signal power P _rin and the error signal power P calls from _e state determination unit 27, the silence or single talk or or the double-talk state or the talk state determination (S2).

If receiver single talk state (S3y), calculates the power ratio calculator 31 power ratio of the transmission signal power _{P y} and the error signal power _{P e} a _{(P e / P y) (} S4). Based on this power ratio (P _e / P _y ), the echo canceller 32 calculates the echo canceling amount (S5), and based on the result, the loss insertion amount distributing unit 33 corresponds to the call state, The amount of loss insertion shown in 4 and 5 is calculated (S6), and it is checked whether or not a double talk state has been reached (S7). If it is not in the double talk state (S7N), the same operation as the conventional example is performed and the process returns to step 1 (S8).

If it is in the double talk state (S7Y), it is selected whether the reception priority or the transmission priority (S9, FIG. 8). If the calculated loss insertion amount is not 6 to 12 dB (S9N), the loss insertion amount distribution unit 33 transmits the transmission loss insertion amount L in order to perform loss insertion on either the transmission side or the reception side shown in FIG. and outputs one of _t or received loss insertion amount L _r is applied to one of the transmitter multiplier 21 or receiving the multiplier 22 (S11), the call is continued (S13).

If the calculated loss insertion amount is 6 to 12 dB (S10Y), the loss insertion amount distribution unit 33 performs the transmission loss insertion amount L _t and the reception in order to perform loss insertion on the transmission side and reception side shown in FIG. The loss insertion amount _Lr is output and the loss insertion amount is distributed and applied to the transmission multiplier 21 and the reception multiplier 22 (S12), and the call is continued (S13). If the power calculation cycle in the power calculator 10 has passed (S14Y) and the call has not ended (S15N), the call operation is repeated after returning to the operation of step S1, and the call is ended (S15Y).

    Because of this operation, even if the loss insertion amount of the voice switch during double talk is a fairly large value of about 6 to 12 dB, for example, the loss insertion amount is distributed to both transmission and reception. The loss insertion amount on the transmission side or the reception side is suppressed to about 6 dB to 8 dB, and the bidirectionality and the transmission / reception gain are improved, so that the call quality in the hands-free call is improved.

It is the circuit block diagram which showed the Example of this invention. (Example 1) FIG. 2 is a circuit configuration diagram showing details of an internal configuration of a power calculation unit, which is an important component of FIG. FIG. 2 is a circuit configuration diagram showing details of an internal configuration of a loss insertion amount calculation unit, which is an important component of FIG. FIG. 2 is a loss insertion location diagram showing a loss insertion location of a voice switch in FIG. FIG. 5 is a distribution diagram of loss insertion amounts to be distributed and inserted during double talk in FIG. FIG. 2 is a call state determination diagram for determining a call state of a voice switch in FIG. It is a flowchart which shows the flow of the telephone call operation | movement by this invention. 8 is a flowchart showing a call operation flow according to the present invention together with FIG. It is a loop gain figure for demonstrating the conventional loop gain. It is the circuit block diagram which showed the prior art example. It is a loss insertion location figure which shows the loss insertion location of the conventional voice switch.

Explanation of symbols

3 Echo canceller
5 Adaptive filter
9 Subtractor
10 Power calculator
12 Received input signal power calculator
13 Transmitter signal power calculator
14 Error signal power calculator
20, 20B voice switch
21 Transmission multiplier
22 Receiver multiplier
25 Loss insertion amount calculator
27 Call status judgment unit
28 Loss insertion amount determination unit
31 Power ratio calculator
32 Echo canceller
33 Loss insertion amount sorter
35 Call status signal
36 Power ratio signal
37 Echo cancellation amount signal
91, 92, 95, 96 Amplifier e Error signal L _r Receive loss insertion amount L _t Transmit loss insertion amount M Microphone P Power R Receiver or speaker r _in Receive input signal S Speaker T Transmitter or microphone t _out Transmit output Signal x Received signal y Transmitted signal

Claims (8)

A pseudo echo signal (y ′) is created in order to remove the acoustic echo that circulates into the microphone (M) from the speaker (S) that converts the received signal (x) into speech, and transmission from the microphone (M) side Echo canceller processing (3) for subtracting the pseudo echo signal (y ′) from the signal (y) to obtain an error signal (e = y−y ′),
In response to the reception input signal (r _in ), the transmission signal (y), and the error signal (e) from the partner terminal, the reception input signal power (P _rin ) and transmission signal power ( P _y ) and error signal power (P _e ) to obtain power calculation processing (10),
The presence / absence of a transmitted / received sound is determined from the received input signal power (P _rin ) and the error signal power (P _e ), and the error signal power (P _e ) and the transmitted signal power (P _y ) An echo cancellation amount is obtained from the power ratio (P _e / P _y ), and a transmission loss insertion amount (L _t ) and a reception loss insertion amount (L _r ) are obtained based on the echo cancellation amount, The error signal (e) is multiplied by the transmission loss insertion amount (L _t ) to obtain a transmission output signal (t _out ) to the counterpart terminal, and the reception loss is added to the reception input signal (r _in ). A voice switch control method of performing voice switch processing (20) for obtaining the received signal (x) by multiplying an insertion amount (L _r ). A pseudo echo signal (y ′) is created in order to remove the acoustic echo that circulates into the microphone (M) from the speaker (S) that converts the received signal (x) into speech, and transmission from the microphone (M) side Echo canceller means (3) for subtracting the pseudo echo signal (y ′) from the signal (y) to obtain an error signal (e = y−y ′);
In response to the reception input signal (r _in ), the transmission signal (y), and the error signal (e) from the partner terminal, the reception input signal power (P _rin ) and transmission signal power ( Power calculation means (10) for obtaining P _y ) and error signal power (P _e );
The presence / absence of a transmitted / received sound is determined from the received input signal power (P _rin ) and the error signal power (P _e ), and the error signal power (P _e ) and the transmitted signal power (P _y ) An echo cancellation amount is obtained from the power ratio (P _e / P _y ), and a transmission loss insertion amount (L _t ) and a reception loss insertion amount (L _r ) are obtained based on the echo cancellation amount, The error signal (e) is multiplied by the transmission loss insertion amount (L _t ) to obtain a transmission output signal (t _out ) to the counterpart terminal, and the reception loss is added to the reception input signal (r _in ). A voice switch control device including voice switch means (20) for multiplying an insertion amount (L _r ) to obtain the received signal (x). The power calculation means (10)
Receiving input signal power calculating means (12) for receiving the receiving input signal (r _in ) and obtaining the receiving input signal power (P _rin );
A transmission signal power calculating means (13) for receiving the transmission signal ( _y ) and obtaining the transmission signal power (P _y );
3. The voice switch control device according to claim 2, further comprising error signal power calculation means (14) for receiving the error signal ( _e ) and obtaining the error signal power (P _e ). The voice switch means (20)
Single talk state of received speech from the received input signal power (P _rin ), single talk state of transmitted speech from the error signal power (P _e ), double talk state when received speech and transmitted speech are present When no incoming call sound and no outgoing voice are detected, it is determined that there is no sound, and when it is determined that the received voice is in a single talk state, the error signal power (P _e ) and the transmitted signal power (P _The echo cancellation amount is obtained from the power ratio (P _e / P _y ) of _y ), and the transmission loss insertion amount (L _t ) and the reception loss insertion amount (L _r ) are calculated based on the echo cancellation amount. Loss insertion amount calculation means (25) for obtaining
The error signal (e) is multiplied by the transmission loss insertion amount (L _t ) to obtain the transmission output signal (t _out ) to the counterpart terminal, and the reception input signal (r _in ) 3. The voice switch control device according to claim 2, further comprising transmission / reception multiplication means (21, 22) for multiplying a reception loss insertion amount (L _r ) to obtain the reception signal (x). The loss insertion amount calculation means (25)
If the received input signal power (P _rin ) is larger than a predetermined received voice threshold, the received voice has a single talk state, and the error signal power (P _e ) is higher than a predetermined transmitted voice threshold. If it is large, it is a single talk state with a transmission voice, a double talk state when there is a reception voice and a transmission voice, and a silence state judgment when no reception voice and transmission voice are detected. Means (27);
When it is determined that the state is the double talk state, whether to give priority to reception or transmission priority is selected according to a predetermined procedure, and the power of the error signal power (P _e ) and the transmission signal power (P _y ) An echo cancellation amount is obtained from the ratio (P _e / P _y ), and the transmission loss insertion amount (L _t ) and the reception loss insertion amount (L _r ) are determined based on the echo cancellation amount. 5. The voice switch control device according to claim 4, further comprising loss insertion amount determining means (28). The loss insertion amount determining means (28)
Power ratio calculation means (31) for calculating the power ratio (P _e / P _y ) from the error signal power (P _e ) and the transmission signal power (P _y );
Echo cancellation means (32) for obtaining an echo cancellation amount from the power ratio (P _e / P _y );
5. The voice according to claim 4, further comprising loss insertion amount distribution means (33) for distributing the transmission loss insertion amount (L _t ) and the reception loss insertion amount (L _r ) based on the echo cancellation amount. -Switch control device. The transmission / reception multiplication means (21, 22)
Transmission multiplier means (21) for multiplying the error signal (e) by the transmission loss insertion amount (L _t ) to obtain the transmission output signal (t _out ) to the counterpart terminal;
5. A voice switch according to claim 4, further comprising: reception multiplication means (22) for multiplying the reception input signal (r _in ) by the reception loss insertion amount (L _r ) to obtain the reception signal (x). Control device. The loss insertion amount distribution means (33)
A loss insertion amount distribution chart for distributing a transmission loss insertion amount (L _t ) and a reception loss insertion amount (L _r ) set in advance corresponding to the echo cancellation amount in the double talk state; The voice switch control device according to claim 6.