JPH1013526A - Control circuit for voice switch - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent the production of howling by blocking echo circulation and to attain high speed changeover of a voice switch not attended with reception/transmission blocking with respect to the improvement of the control circuit for a voice switch used to build up a hand-free speech system. SOLUTION: The attenuation amount is controlled among a reception state, a transmission state, according to the presence of a voice signal, a reception reverberation state and a transmission reverberation state being intermediate states while being transited among them, and the attenuation of a transmitter side speech circuit in the case of transiting to the reception reverberation state is adjusted and controlled by referring to output of a circuit simulating the reverberation characteristic of a room. It is controlled that a level of a leaked echo because of a small attenuation of the transmitter side adjusted by referring to an output of a circuit simulating the reverberation characteristic of a room and a receiver side signal level are compared and the voice switch is set to the transmission state when the former is higher and the the voice switch is set to the reception state when the latter is higher.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an improvement of a voice switch for constructing a hands-free communication system that allows a user to talk using a speaker and a microphone without having a handset.

Generally, in a telephone or a video conference system having a hands-free function for making a call using a speaker and a microphone instead of a handset, an acoustic coupling between the speaker and the microphone is added (in a system including a soundproofing circuit, a sidetone coupling is added). ) Is required to prevent howling that occurs when the gain of the one-cycle closed circuit constituted by (1) exceeds 1.

A voice switch is provided as one of the means. FIG. 7 is a diagram showing the basic concept of this voice switch. The voice switch includes a transmission attenuation circuit 1N for attenuating transmission voice to a transmission side circuit in each of the near-end speaker terminal NE and the far-end speaker terminal FE connected via a transmission path. 1F, and a receiving-side circuit are provided with receiving attenuating circuits 2N, 2F for attenuating the receiving voice. From the comparison between the signal powers of the receiving side and the transmitting side, the larger one is determined to be the uttering speaker, and the attenuation amount of the transmitting attenuation circuit 1N of the uttering side (for example, the near end speaker terminal NE) is set to 0. The occurrence of howling is prevented by inserting a large amount of attenuation into the transmission attenuation circuit 1F on the opposite side of the speaker and controlling the gain of the one-cycle closed loop to always be less than 1.

[0004] The problem with this voice switch prevention method is that the attenuation provided by the voice switch is always one-way since the attenuation is always inserted on either side. It is. Usually, conversations are alternating speakers, so this one-way nature of the conversation does not make the conversation impossible. However, in order to provide a smooth conversation, the switching must be as quick as possible.

[0005]

2. Description of the Related Art A conventional voice switch is usually constructed as shown in FIG. That is, a transmission attenuation circuit 1 that inserts an attenuation amount into a transmission voice, a reception attenuation circuit 2 that inserts an attenuation amount into a reception voice, a speaker 8 that emits a reception voice, a microphone 9 that inputs a transmission voice, a transmission voice. Power P _T
Power calculation filter 21 for calculating the power calculation filter 22 to calculate the power P _R of the received voice control for controlling the insertion of attenuation in transmission attenuation circuit 1 and the reception attenuation circuit 2 on the basis of the power P _T and P _R It is configured to include a circuit 34.

[0006] In the control circuit 34, in the comparison of the power P _T of the transmission voice and power P _R of the received voice, aP _R
If ≧ _PT, it is determined that the receiver is in the receiving state, the attenuation of the receiving attenuation circuit 2 is 0, and the attenuation of the transmitting attenuating circuit 1 is large. If aP _R < _PT , the transmitting state is determined and the transmitting attenuation circuit 1 is determined.
The communication direction is switched by controlling the attenuation of the reception attenuation circuit 2 to 0 and the attenuation of the reception attenuation circuit 2 to a large value. Here, a is a constant determined from the magnitude of the acoustic coupling assumed in advance.

The problem is that the received voice transmitted from the speaker 8 becomes an echo due to the echo of the room and becomes
, Which is derived from the fact that this echo cannot be distinguished from the transmitted voice. That is, considering that it takes a little time for the echo to reach the microphone 9, the power fluctuates greatly with the passage of time of the voice. If the aP _R <P _T from variation (transmission state) (1) are also fully contemplated. In this case, the voice switch is turned to the transmitting side by the same echo (that is, a large amount of attenuation is inserted into the receiving attenuation circuit 2), and as a result, the power of the immediately following received voice increases, and aP _R ≧ P _T ( (Receiving state) A phenomenon called receiving blocking occurs in which the received voice is interrupted until it is determined that (2). The level fluctuation of the voice is large, and even if the far-end speaker is uttering (during reception), the possibility that the relationship of Expression (1) is satisfied and the transmission state is determined is sufficiently high.

In order to suppress the reception blocking by this method, the power of the received voice is considered to be larger than the actual power by increasing the constant a, and the time constant of the filter 21 for calculating the power of the received voice corresponds to the reverberation time. It is necessary to adopt a configuration in which the possibility that aP _R <P _T (transmission state) is reduced. However, increasing the time constant of the power calculation filter 21 is to increase the time during which the current state is forcibly set to the state in which the far-end speaker is speaking (receiving state) even when the received voice becomes silent. Furthermore, the constant a
Increasing means that a high power transmitted voice is required to switch the voice switch to the transmitting side. All of these delay the switching of the voice switch to the transmitting side, making it difficult to smoothly switch to a normal conversation in which the speaker changes without time.

As a means for solving this problem, a method proposed by the present inventors is an audio switch circuit shown in FIG. That is, the power calculation filter is replaced with a voice detection circuit, and the call is switched according to the presence or absence of voice detection. In FIG. 9, the transmission attenuation circuit 1, the reception attenuation circuit 2, the speaker 8, and the microphone 9 are as described above. 4 is a transmitted voice detection circuit for detecting the presence or absence of transmitted voice, 5 is a received voice detection circuit for detecting the presence or absence of received voice,
31 is a transmission reverberation setting circuit, and 32 is a reception reverberation setting circuit.

The switching state transition of the voice switch is performed as shown in FIG. First, the basis of the switching rule is that only the transmitted voice is detected → the transmitting state is determined, and the receiving side attenuation is increased, the transmitting side attenuation is set to 0. Only the received voice is detected → the receiving state is determined. The attenuation on the sending side is large, the attenuation on the receiving side is set to 0, both voices are detected → the attenuation is not switched as the current priority, and both voices are silent → the attenuation is not switched as the current priority. .

For example, when it is assumed that the apparatus is currently in the receiving state, according to the above switching rule, the receiving voice wraps around the microphone 9 as an echo, and the power of the echo is much larger than the power of the receiving voice. Even at this time, the receiving state is maintained while the receiving voice is continuously obtained (the state (a) of the receiving state shown in FIG. 10),
The amount of attenuation remains inserted on the transmitting side. As a result, the received voice is provided without interruption and a stable call is provided.

With this basic control alone, a problem arises when the received voice or the transmitted voice is temporarily silenced. First, the case where the received voice becomes silent will be described. That is, according to this control method, the echo of the received voice is immediately determined as the transmitted voice, and the state shifts to the transmitted state. Therefore, as long as the echo is detected, the received voice is blocked even if the received voice is received again with priority given to the current state. Therefore, in the state transition of FIG. 10, a reverberation state of the received voice is provided. When this reverberation state is provided, even if the received voice is temporarily silenced, the voice switch does not immediately shift from the receiving state to the transmitting state due to the echo, but receives the voice via (b) shown in FIG. At the same time as the state moves to the reverberation state, the attenuation amount enters the receiving side. This state is different from the transmitting state in which the switch is kept in priority to the transmitting side, and the attenuation is temporarily set to the receiving side (transmitting state), and if the receiving voice is detected again, The state immediately returns to the receiving state via (c) shown in FIG. 10, and the attenuation can be switched to the transmitting side again. In this way, while the voice switch is in the reverberation state, receiving blocking is prevented.

The silence of the received voice is equivalent to the silence of the transmitted voice at the far end. Therefore, in this case, the voice switch on the far end side is in the transmitting state. However, when the voice of the far-end speaker returns as an echo from the near end after the silence, the voice switch on the far end switches to the receiving side. At this point, if the voice switch is allowed to transition from the transmitting state to the receiving state, a large amount of attenuation is inserted into the transmitting side even if the far-end speaker continues to utter while the echo continues. Therefore, the sound is not transmitted to the near end. That is, transmission blocking occurs.

As a countermeasure against this, the voice switch has a rule that a reverberation state is provided between a transmission state and a reception state, and the state returns to the transmission state as soon as a transmission voice is detected. However, this causes howling. That is, the voice switch of the far-end talker side terminal FE determines the echo from the near-end talker side terminal NE as the received voice, and immediately shifts to the transmission reverberation state. 2 is set to 0, the attenuation of the transmission attenuation circuit 1F on the transmission side is increased, and the echo is output from the speaker 8F. The voice switch on the far end side outputs the echo output from the speaker 8F. The sound is picked up by the microphone 9F and determined to be the voice of the utterance of the far-end speaker, and the state shifts to the transmission state and the attenuation of the transmission attenuation circuit 1F is returned to zero. As a result, the echo is sent back to the voice switch of the near-end talker terminal NE as the received voice. Naturally, the voice switch of the near end speaker terminal NE returns to the receiving state according to the receiving voice rule, and outputs the echo from the speaker 8N. That is, the echo circulates through the communication circuit many times, and the danger of howling increases.

As means for solving this problem, the present inventors have proposed the following method. That is, a reverberation monitoring filter (low-pass filter) having an appropriate rise time constant is provided on the microphone 9 side.
The reverberation monitoring filter outputs an echo following the ending of the received voice when the received voice is interrupted when the voice switch on the near end speaker side is in the receiving state and becomes silent.

At this point, when there is a near-end speaker's utterance and the transmitted voice is mixed with the echo of the received voice, the transmitted voice mixed with the echo is reproduced in order to secure simultaneous communication. It is desirable to transmit to the far end speaker. But,
The transmission is not preferable from the viewpoint of suppressing reverberation and preventing howling.

Therefore, the near-end talker side terminal NE controls the echo of the received voice returned to the far end side to a level at which the far-end talker side does not judge the input of voice. That is, at the near end speaker terminal NE, the received voice is temporarily silenced and its echo is
In the reverberation state of the reverberation that wraps around from N, the attenuation amount that the attenuation amount of the transmission attenuation circuit 1N is originally set to 0 is reduced to such an extent that the echo is attenuated to the ambient noise level in the far end talker side terminal FE. insert. The insertion of this attenuation is limited to the period during which the reverberation of the echo persists. If the attenuation of the transmission attenuation circuit 1N is gradually reduced in accordance with the reverberation characteristic with reference to the output of the reverberation monitoring filter, the echo is not detected as voice by the far-end speaker terminal FE, and is not detected from the speaker 8F. It will not be output. In other words, no reverberation is generated for the far end speaker, and no howling occurs. At the same time, the attenuation of the transmission attenuation circuit 1N in the near-end talker side terminal NE becomes 0 after the echo of the received voice is attenuated. The voice of the near end speaker is detected by the voice switch of the end speaker side terminal FE, and the voice of the near end speaker is output from the speaker 8F of the far end speaker side terminal FE.

When returning from the reverberation state to the transmission state, there is a possibility that the same phenomenon as described above may occur. In this case as well, the transmission is performed by referring to the output of the reverberation monitoring filter. The attenuation amount of the speech attenuation circuit 1 is gradually reduced in accordance with the reverberation characteristics, and an attenuation amount is inserted so that the echo circulating to the microphone 9F is attenuated to such an extent that the other party's terminal can be regarded as the ambient noise level.

[0019]

The problem with this control method is that the condition for preventing blocking is that "the echo transmitted to the far end is completely suppressed to the ambient noise level". In other words, if there is an error in the estimation of the reverberation characteristic and even a small amount of echo leaks to the far end side, circulation of the echo is inevitable. This control also suppresses the transmitted voice at the same time as the echo, so that the voice of the near-end speaker cannot be heard from the far-end speaker during that time. This means that call switching is delayed.

Accordingly, it is an object of the present invention to prevent the occurrence of howling by preventing the echo from circulating, and to enable the voice switch to be switched at a high speed without blocking reception or transmission. .

[0021]

In order to solve the above-mentioned problems, in one aspect of the present invention, howling is generated by alternately inserting an attenuation amount into a communication circuit on a transmitting side and a receiving side. In a voice switch used for a hands-free function that realizes a telephone conversation with a microphone and a speaker, the amount of attenuation is determined according to the presence or absence of voice in the receiving state, the transmitting state, and the reverberation state, A control circuit that controls by transiting between reverberation states and, when transiting to the reverberation state, controls the amount of attenuation in the transmitting side speech circuit by referring to the output of the circuit that simulates the reverberation characteristics of the room. There is a small attenuation on the transmitting side, which is adjusted with reference to the output of the circuit that simulates the reverberation characteristics of the room, and the level of the leaked echo is compared with the signal level on the receiving side. The voice switch to the sending state in the case,
A control circuit for an audio switch, characterized in that control is performed to set a receiving state when the latter is large. This voice switch is suitable for use in a hands-free communication system using a four-wire transmission line.

In the present invention, as another form,
The amount of attenuation is reduced by inserting the amount of attenuation into the communication circuits on the transmitting side and the receiving side alternately to suppress the occurrence of howling and using the voice switch used for the hands-free function to realize a telephone conversation with a microphone and a speaker. Depending on the presence or absence, the receiving state, the transmitting state, and the intermediate state of the receiving reverberation state and the transmitting reverberating state are transitioned and controlled, and when the state transits to the transmitting reverberating state, the attenuation amount of the receiving side speech circuit is reduced. A control circuit for controlling so as to refer to the output of a circuit simulating the reverberation characteristic at the far end, wherein the attenuation on the receiving side is adjusted by referring to the output of the circuit simulating the reverberation characteristic at the far end. Comparing the level of the leaked echo with a small amount and the signal level of the transmitting side, and setting the voice switch to the receiving state when the former is large, and performing the transmitting state when the latter is large. That the control circuit of the voice switch is provided. This voice switch is suitable for use in a hands-free communication system using a two-wire transmission line.

In the above-mentioned voice switch, the attenuation circuit for inserting the amount of attenuation into the communication circuits on the transmitting side and the receiving side may be provided with a delay circuit in front of a delay corresponding to the time required for voice detection by the voice switch. it can.

[0024]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows an embodiment in which the present invention is applied to a hands-free communication system using a four-wire transmission line (a transmission line separated into a transmitting side and a receiving side).

In FIG. 1, reference numeral 1 denotes a transmission attenuation circuit for attenuating a transmission voice, 2 denotes a reception attenuation circuit for attenuating a reception voice, 8 denotes a speaker for generating a reception voice, 9 denotes a microphone for inputting a transmission voice, 4 is a transmitted voice detection circuit for detecting the presence or absence of transmitted voice, 5 is a received voice detection circuit for detecting the level of received voice B, and 6 is a transmitted voice for detecting the level of output voice A of the transmission attenuation circuit 1. A detection circuit 7 is a reverberation monitoring filter, and 3 is a determination circuit for controlling insertion of an attenuation amount in the transmission attenuation circuit 1 and the reception attenuation circuit 2. As described above, the reverberation monitoring filter 7 is constituted by a low-pass filter having an appropriate rising time constant, and when the voice switch on the near end speaker side is in the receiving state, the received voice is interrupted and becomes silent. At that point, an echo following the ending of the received voice is output.

In the following description, the near end speaker terminal N
When it is necessary to represent the circuit of E, "N" is added after the reference number of each circuit, and when it is necessary to represent the circuit of the far end speaker terminal FE, "F" is appended to the reference number.

This voice switch operates in the above-described transmission state,
A transition is made between four states of a transmission reverberation state, a reception state, and a reception reverberation state according to the presence or absence of the transmission voice and the reception voice. In the reception reverberation state, when the state transits to the reception reverberation state, the attenuation is temporarily inserted into the transmission attenuation circuit 1 for a certain period of time, and thereby the echo transferred to the far end talker side terminal FE is reduced. Is controlled to a level that is not detected as a voice at the far end speaker terminal FE. The same control is performed when returning from the transmission reverberation state to the transmission state.

Here, the voice of the near end speaker, that is, the transmission voice, input to the microphone 9 is transferred to the far end speaker side terminal FE via the transmission attenuation circuit 1. In the far-end talker side terminal FE, the voice of the far-end talker, that is, the received voice, is output from the speaker 8F via the reception attenuation circuit 2F. The present invention can prevent blocking of transmission / reception even when the echo attenuating from the speaker 8N to the microphone 9N cannot be suppressed by the transmission attenuation circuit 1N to a level that is not detected as voice on the far end speaker side. to be able to do,
It is intended to provide rules for state transitions from the transmission reverberation state and the reception reverberation state.

First, the output of the reverberation monitoring filter 7 is used for suppressing the echo. The output of the reverberation monitoring filter 7 is used by the determination circuit 3 to calculate the amount of attenuation required to bring the echo to a level that is not detected as voice at the far end speaker. The voice detection circuit 4 detects the transmitted voice, and the received voice detection circuit 5 detects the received voice and provides information necessary for the transition of the call state in the determination circuit 3.

A transmission voice detection circuit 6 provided on the output side of the transmission attenuation circuit 1 is used for comparing the level of the leaked echo A with the signal B of the reception side. The determination circuit 3 determines the larger one as the call direction, and as a result, blocks the reception and the transmission.

FIG. 2 is a state transition diagram of the embodiment according to the present invention. The operation of the voice switch according to the embodiment will be described with reference to this state transition diagram.

When the attenuation of the transmission attenuation circuit 1N, which is adjusted by using the reverberation monitoring filter 7N at the near end talker terminal NE, is small and the echo leaks to the far end talker terminal FE, the far end talk The voice switch of the speaker terminal FE determines that the echo is the voice of the near-end speaker, shifts to the transmission reverberation state, outputs an echo from the speaker 8F, and inputs the echo to the microphone 9F. As a result, the voice switch of the far-end talker side terminal FE maintains the transmission reverberation state, and in this case, naturally, the reception voice is not transmitted to the near end side with respect to the microphone input of the reception voice by the transmission attenuation circuit 1F. Apply attenuation.

Therefore, the echo returned to the near-end talker terminal NE may be considered to have a smaller amplitude than the original echo observed at the output of the near-end transmission attenuation circuit 1N. That is, in the near-end talker side terminal NE in the reverberation state, when the echo that has cycled in the previously proposed method is detected as the received voice, the state immediately returned from the reverberation state to the receiving state is transmitted. When comparing the amplitude or the power of the signal observed by the speech sound detection circuit 6N and the received speech detection circuit 5N, if the former is larger, the voice switch is maintained on the transmitting side (that is, the reception reverberation state is maintained). Maintained) can be determined. As a result, a large amount of attenuation remains inserted in the reception attenuation circuit 2N.
The echo is prevented from circulating and howling is prevented.

In the far-end talker terminal FE, the voice switch that has determined that the echo is the received voice at the far-end side transits to the transmission reverberation state.
In order to solve the problem of inserting an amount of attenuation into F and subsequently blocking the transmission of the voice of the far-end speaker who has uttered (sending reverberation state), the attenuation inserted into the sending attenuation circuit 1F. Considering that the amount is controlled to the ambient noise level by making the magnitude proportional to the amplitude or power of the signal detected on the receiving side, the attenuation inserted into the transmitting side for this leaked echo is It can be said that there are few. Therefore, when the far-end speaker utters in this state, the sound is detected to a certain extent at the output of the transmission attenuation circuit 1F. Therefore, if the echo leaked from the near-end speaker terminal NE is suppressed to a level that does not cause reverberation, the far-end talker terminal FE has a transmission attenuation circuit 1F.
It can be expected that the transmitted voice A (the voice of the far-end speaker) passing through is larger than the leaked echo B (input to the reception attenuation circuit 2F). That is, by causing the voice switch to return to the transmission state from the transmission reverberation state at that time, transmission blocking is prevented from occurring.

That is, when the voice switch is in the transmission reverberation state, the output of the transmission attenuation circuit 1 is A and the input of the reception attenuation circuit 2 is B. If A> B, the speech is transmitted from the transmission reverberation state. Transition to the state ・ If A <B, control to maintain the reverberation state of the transmission. The transition from the transmission reverberation state to the reception state is performed by setting a time, and when the transmission reverberation state continues beyond the time (the received voice is detected beyond the reverberation time), the call direction is set. Can be determined to have switched to the receiving side, so that the state may be shifted to the receiving state.

As described above, when the voice switch of the far-end talker side terminal FE can return to the transmitting state during the reception of the echo, the voice switch of the near-end talker side terminal NE uses the voice detection circuit 5N on the receiving side. It may be considered that the detected signal B (that is, the voice of the far-end speaker) is larger than the echo A that leaks on the transmitting side (that is, the detection output of the transmitting voice detecting circuit 6). Therefore, when the output A of the attenuation circuit 1N on the transmission side is compared with the input B of the attenuation circuit 2N on the reception side and the reception side is large, the occurrence of the reception blocking is prevented by transitioning from the reception reverberation state to the reception state. .

If A> B, the reverberation state is maintained. If A <B, the reverberation state is changed from the reverberation state to the reception state. The transition from the reverberation state to the transmission state is also performed by setting a time. If the reception reverberation state continues beyond that time (when the transmitted voice has been detected for more than the reverberation time), a call is made. Since it can be determined that the direction has been switched to the transmitting side, the state may be shifted to the transmitting state.

Next, a detailed configuration example of the voice detection circuit in the above-described embodiment will be described. First, voice detection must be performed as quickly as possible, and a rising time constant that does not cause silence for at least one pitch of a vowel is required. Therefore,
The structure of a low-pass filter for detecting voice is given as shown in FIG.

In FIG. 4, reference numeral 101 denotes a short-time integration circuit for integrating the input signal in a short time, and 102 denotes a short-time integration circuit 10.
The comparator 103 selects the output X of the short-time integration circuit 101 or the output Z of the adder 106 according to the comparison result of the comparator 102, and compares the output Y with the output Y. Switch 104, switch 103
A delay circuit for outputting an output Z obtained by delaying the output Y of the
5 is a multiplier for multiplying the output X of the short-time integration circuit 101 by (1−α), 107 is a multiplier for multiplying the output Z of the delay circuit 104 by α, and 106 is an adder for adding the outputs of the multipliers 105 and 107 And 108 are registers for storing the output of the switch 103 according to the comparison result of the comparator 102.

In FIG. 4, the switch 103 is switched to Y = X if X ≧ Z and Y = X if X <Z. The reason why the short-time integration circuit 101 is inserted in the input portion is to suppress the amplitude fluctuation of the signal X, and is a cumulative value of the amplitude (absolute value or square value) for several sampling periods of the input signal. The fluctuation can be suppressed to a practical size by setting the integration section to about 1 to 2 ms. When the comparator 102 selects X, the register 108 stores and updates the X, and this X is used for the level comparison and the initial value of the reverberation monitoring filter.

When the presence or absence of sound is determined from the output of the filter, the noise level is also calculated so that the noise picked up by the microphone is not mistaken with the sound, and the sound detection threshold value is determined from the noise level. Is common. Furthermore, as a method of calculating the noise level,
A low-pass filter having a long rising time constant and a short falling time constant is used. That is, it is important to have a characteristic that the sound level does not easily rise even when the sound is input, and returns to the noise level immediately after the sound is lost. FIG. 5 shows an example.

In FIG. 5, reference numeral 201 denotes a short-time integration circuit for integrating the input signal in a short time, and 202 denotes a short-time integration circuit 20.
1 is compared with the output M of the delay circuit 207, 203 is a counter U, 204 is a counter D, 205 is a switch, 206 is the output of the switch 205 and the delay circuit 20.
An adder 207 for adding the outputs of 7 and 207 is a delay circuit for delaying the output N of the adder 206 to obtain an output M.

First, the comparator compares the magnitudes of the signals X and M. If X ≧ M, the value of the counter U is increased by one, and if X <M, the value of the counter D is increased by one. And
When the counters 203 and 204 have counted the predetermined numerical values U and D, the switch 205 determines that the counter 203 has reached the predetermined value U. When the counter 204 has reached the predetermined value D, -β. If both counters have not reached the predetermined value, 0 is selected and added to M. At this time, if the predetermined value U counted by the counter 203 is selected to be large and the predetermined value D counted by the counter 204 is selected to be small, a filter having a slow rise and a fast fall is configured. That is, the sound is output from the noise level detection filter N and the output Y from the sound level detection filter.
If Y ≧ bN, it is determined that there is a voice. If Y <bN, it is determined that there is no voice. Here, b is a margin for preventing noise from being mistaken for voice.

However, even if the rise of the voice detection filter is quick, the time required for the short-time integration circuit 101 is required, and the delay cuts off the head of the voice. Therefore, as shown in FIG. 6, a delay corresponding to the delay time is inserted into the communication circuit by a delay circuit to prevent the head of speech from being cut off.

The reverberation monitoring filter sets the output of the register 108 of the filter shown in FIG. 4 as an initial value when shifting to the reverberation state, and calculates the output as P = Pα + N (1-α). Here, N is the ambient noise level, and the echo including the ambient noise is inputted in consideration of gradually approaching the ambient noise level. In this case, the attenuation to be inserted into the transmitting side is calculated as L _T = P / KN. Here, K is a constant for correcting a value that appears lower than the actual value because the rise time constant and the fall time constant of the noise level detection filter are different. As a result, the echo leaking to the output of the attenuation circuit is suppressed to the ambient noise level. This attenuation is inserted in the reverberation state of the received speech.

Similarly, in order to suppress the echo input to the microphone in the receiving state to the ambient noise level, use the output of the voice detection circuit provided on the transmitting side and set L _S = P / KN. If so, no echo is transmitted to the far end.

Accordingly, the received voice becomes silent, and the echo input to the microphone is suppressed by L _S , transmitted to the far end, and the voice switch on the far end shifts to the reverberation state of the transmission. At that time, the voice switch at the far end gives L _S attenuation to the echo on the transmitting side, so that the echo leaking to the near end is further attenuated. Therefore, the level comparison does not return to the transmission state. However, since the amount of attenuation is given as a magnitude to be compared with the signal level, the amount of attenuation given to the leaked echo is small. This means that if there is a far-end speaker's utterance, the sound is detected at a high level at the output of the attenuation circuit. That is, by comparing the two levels, it is possible to return to the transmission state.

On the other hand, on the near end side where the reverberation state is reached,
The echo is compared with the echo returned from the far end, and the former is larger, so the reverberation state is maintained, and when the voice of the far end speaker is input, the comparison result is reversed, and the receiver returns to the receiving state. You can do it.

FIG. 3 shows another embodiment of the present invention.
This embodiment is a configuration in the case where the present invention is applied to a hands-free communication system of a two-wire transmission line (a transmission line is shared between a transmitting side and a receiving side). In this two-wire transmission path system, the echo leaked on the transmitting side goes around to the receiving side through the soundproofing circuit, but this leaked echo is similar to the echo that has made a round in the previous embodiment. Since the signal is not attenuated in the speaker terminal FE, it is necessary to consider this point. In other words, in the apparatus of this embodiment, a voice detection circuit 10 for detecting the output level of the reception attenuation circuit 2 and a reverberation monitoring filter 11 are added in place of the function of the far end talker side terminal FE. For the leak echo circulating through the sound circuit, the reception attenuation circuit 2 inserts an attenuation amount that makes it equal to or lower than the ambient noise level.

The reverberation monitoring filter 11 provided on the receiving side monitors the echo circulating around the soundproofing circuit, and controls the echo so as not to be output from the speaker 8. The voice detection circuit 10 detects the leaked echo (that is, the output of the transmission attenuation circuit 1) and the signal on the receiving side (that is, the output of the reception attenuation circuit 2).
This is used to determine the level of speech communication and to prevent blocking of reception and transmission.

When the soundproofing circuit is interposed in the transmission path, the echo leaking from the near end to the far end returns to the receiving side via the soundproofing circuit. That is, the state is similar to that of the near-end voice switch 4-wire type far-end side. When the transmitted voice becomes silent, the return signal is determined to be the received voice, the voice switch is switched to the reverberation state of the transmitted voice, and the call direction is switched to the receiving side,
Transmission blocking occurs in which the sound of the subsequent near-end speaker is blocked. However, the amount of attenuation inserted into the receiving side in the receiving attenuation circuit 2 is adjusted so that the signal sent to the speaker 8 is equal to or lower than the ambient noise level with reference to the output of the reverberation monitoring filter 11. Therefore, no sound is detected by the sound detection circuit 10. However, when the far-end speaker utters the voice, the voice is detected when the amount of attenuation decreases, so that the voice switch can be set to the receiving side.

When the echo leakage is permitted, the voice switch is switched to the receiving side using the echo as the receiving voice, so that the signal level of the voice detecting circuit 10 and the signal level of the transmitting voice detecting circuit 4 on the microphone side are compared. If the latter is larger, the transmission state is set, and if the former is larger, the transmission reverberation state is set.

[0053]

As described above, according to the present invention, it is possible to prevent the occurrence of howling by preventing the echo from being circulated, and it is possible to switch the voice switch at high speed without causing blocking of reception or transmission.

[Brief description of the drawings]

FIG. 1 shows an embodiment in which the present invention is applied to a hands-free communication system using a four-wire transmission line.

FIG. 2 is a diagram illustrating a state transition of a voice switch according to an embodiment.

FIG. 3 shows an embodiment in which the present invention is applied to a hands-free communication system using a two-wire transmission line.

FIG. 4 is a diagram illustrating a detailed configuration of a voice detection circuit in the voice switch according to the embodiment;

FIG. 5 is a diagram illustrating a detailed configuration of a noise level detection filter in the voice switch according to the embodiment.

FIG. 6 is a diagram showing an example of the configuration of a voice switch that eliminates talk start disconnection.

FIG. 7 is a diagram illustrating a one-cycle closed circuit constituted by the hands-free communication system.

FIG. 8 is a diagram showing a conventional example of a voice switch.

FIG. 9 is a diagram illustrating an example of a voice switch having a function of preventing blocking due to a reception / transmission reverberation state.

FIG. 10 is a diagram illustrating a state transition of a voice switch having a reception / transmission reverberation state.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 transmission attenuation circuit 2 reception attenuation circuit 3 determination circuit 4 transmission voice detection circuit 5 reception voice detection circuit 7 reverberation monitoring filter 8 speaker 9 microphone 10 voice detection circuit 11 reverberation monitoring filter

Claims (3)

[Claims] An audio switch used in a hands-free function for suppressing the occurrence of howling by inserting an amount of attenuation alternately into a communication circuit on a transmitting side and a receiving side to realize a telephone conversation with a microphone and a speaker. The amount of attenuation is controlled by transiting between the receiving state, transmitting state, and the reverberating state and transmitting reverberating state located in the middle between them according to the presence or absence of voice. A control circuit for controlling the amount of circuit attenuation by referring to the output of a circuit that simulates the reverberation characteristics of a room, wherein the transmitting side is adjusted by referring to the output of a circuit that simulates the reverberation characteristics of a room. The level of the leaked echo is small and the level of the leaked signal is compared with the signal level of the receiving side. When the former is large, the voice switch is set to the transmitting state, and when the latter is large, the receiving state is controlled. A voice switch control circuit characterized by the following. 2. An audio switch used for a hands-free function for suppressing the occurrence of howling by alternately inserting an amount of attenuation into a communication circuit on a transmitting side and a receiving side to realize a telephone conversation with a microphone and a speaker. The amount of attenuation is controlled by transitioning between the receiving state, the transmitting state, and the reverberating state and the transmitting reverberating state located between them, depending on the presence or absence of voice. A control circuit for controlling the amount of circuit attenuation by referring to the output of a circuit that simulates the reverberation characteristics at the far end, wherein the control circuit is adjusted with reference to the output of the circuit that simulates the reverberation characteristics at the far end. The level of the echo that leaked due to a small attenuation on the receiving side is compared with the signal level on the transmitting side.If the former is large, the voice switch is set to the receiving state, and if the latter is large, the control is set to the transmitting state. Do A control circuit for an audio switch. 3. An attenuating circuit for inserting an amount of attenuation into a communication circuit on a transmitting side and a receiving side is provided with a delay circuit in front of a delay corresponding to a time required for voice detection by a voice switch. Item 3. A control circuit for an audio switch according to item 1 or 2.