JPH0746791B2 - Howling suppression circuit - Google Patents

Description

TECHNICAL FIELD The present invention relates to a circuit for preventing howling that occurs in a communication system in which transmission and reception are separated. Here, "howling" means that the signal of the receiving system is circulated to the transmitting system for some reason and the entire system is oscillated.

〔Overview〕

The present invention, in a howling circuit that prevents howling that occurs in a communication system in which transmission and reception are separated, generates a pseudo echo signal of an echo signal that circulates into the transmission system from a signal that arrives at the reception system, and generates the pseudo echo signal. An echo canceling function that inverts the phase and adds it to the transmission system and a frequency shift circuit that shifts the frequency component of the signal passing through the signal path by a predetermined value are used together, and the echo canceling function and the frequency shifting function are provided in the loop. By functioning at least one or more functions, it is possible to eliminate the lack of volume of the speaker, which is included in the howling suppression circuit by the echo canceling method, and to obtain the optimum volume while suppressing howling.

[Conventional technology]

To simplify the description, the audio system of the video conference telephone device will be described below. FIG. 2 is a block diagram showing a voice system of the video conference telephone device. FIG. 2 (a) shows an example of opposite communication, and FIG. 2 (b) shows an example of 4-point communication.

Generally, in the audio system of a video conference telephone device, it is composed of a speaker (loudspeaker) that allows a plurality of people to hear the voice of the other party at the same time, a microphone that collects the sounds generated by a plurality of people, and an amplifier that drives the speaker and the microphone. It

In such a configuration, a reverberant sound generated when the sound from the speaker is reflected on the wall in the room may be circulated to the microphone and inserted as a reverberation signal. Further, a similar phenomenon occurs in the voice system of the video conference telephone device on the other side, and when this echo signal forms a loop, by increasing the gain of the amplifier so as to increase the sound level from the speaker, It is well known that the gain of a loop loop exceeds 1 to generate howling.

Further, in the form in which a plurality of grounds are connected at the same time as shown in FIG. 2 (b), even if the loop loop gain is 1 or less in the opposite communication, a plurality of grounds can be connected by connecting the plurality of grounds via the voice addition circuit. The total loop gain due to the signal loop exceeds 1, which also causes howling. The details are described in “Design conditions of a relay device in a voice conference communication system”, IEICE Communication Systems Research Group Material CS83-81 August 1983].

As described above, in the audio system of the video conference telephone device, a means for preventing howling is necessary regardless of the configuration of the communication system. As means for preventing howling in this communication system, there are (a) echo suppression (echo suppressor) system and (b) echo cancellation (echo canceller) system. The communication terminal device will be described.

(A) Echo blocking method An apparatus for realizing the echo blocking method has a voice switch circuit or an echo suppressor and has been widely used since ancient times.

FIG. 3 is a block diagram showing a howling suppression circuit using an echo suppressor.

Echo signal 3 from speaker 1 to microphone 2
And the voice signal 4 of the talker directly input to the microphone 2 can be discriminated from each other by an amplifier or attenuator 5, 6, 7, 8 for setting a level diagram so that the signal power of each can be discriminated. It is realized by inserting into the transmitting side and the receiving side respectively. An example of this is described in ["Conference Phone" Foreign Communication Technology No. 20 'January 1969].

The echo suppressor 9 shown in FIG.
11 and control circuit 12. Received signal input terminal Rin13
And the signal power at the transmission signal input terminal Sin15 is measured respectively, and when the reception level LRin at the reception signal input terminal Rin13 and the transmission level LSin at the transmission signal input terminal Sin15 are LRin> LSin, the attenuation of the variable attenuation circuit 10 is reduced. Set the amount of attenuation to the minimum and the amount of attenuation of the variable attenuation circuit 11 to the maximum. When LRin = LSin, set the attenuation amount of the variable attenuation circuit 10 and the attenuation amount of the variable attenuation circuit 11 to 6 dB, respectively, and set LRin <LSin At this time, the loop gain is controlled to be 1 or less by the control circuit 12 that sets the maximum attenuation amount of the variable attenuation circuit 10 and the minimum attenuation amount of the variable attenuation circuit 11 to prevent howling.

(B) Echo canceling method An echo canceller is famous as a device that realizes the echo canceling method.

FIG. 4 is a block diagram showing a howling suppression circuit using an echo canceller.

The echo canceller 17 is generally composed of an adaptive control filter 18, a subtraction circuit 19, and a detection circuit 20 for instructing the start or stop of the adaptive control operation. The detection circuit 20 compares the reception level LRin at the reception signal input terminal Rin13 with the transmission level LSin at the transmission signal input terminal Sin15, and the reception signal input terminal Rin13 has an audio signal input, and sends it to the transmission signal input terminal Sin15. When it is possible to identify that there is no speaker voice signal input, the transfer function of the adaptive control filter 18 is controlled so that the output (residual signal) power of the subtraction circuit 19 is always close to zero.

If it can be identified that there is a voice signal input from the talker,
The operation of the adaptive control filter 18 is stopped. At this time, the received signal becomes the reverberant signal 3 through the reverberant path and is mixed with the voice signal 4 of the talker, but the pseudo reverberant signal generated from the received signal is output to the output of the adaptive control filter 18. From the output of the subtraction circuit 19, the voice signal of the speaker is extracted.

In such echo suppression method and echo cancellation method, it is necessary to identify whether the communication system at present is always in the transmitting state or in the transmitting state.

On the other hand, in addition to preventing howling in a communication system, there is a frequency shift method as a means for preventing howling in a sound field system caused by a microphone and a speaker in a performance hall.

An apparatus that realizes this method is called a frequency shifter, a pitch controller, or a harmonizer.

Harmonizers are used in concerts and lectures to enhance acoustic effects.The operating principle is that the input signal is stored for a preset time, and the stored information is slower than the time speed at which the input signal is captured. Alternatively, the frequency is shifted by speeding up the output.
Generally, when the pitch ratio, which is the ratio of the frequency after the change to the original frequency, is increased, the characteristics of the voice change, but if the pitch ratio is set to a level that does not change the voice quality, there is no problem, and the echo signal The wavy frequency characteristic can be improved.

[Problems to be solved by the invention]

However, such a conventional howling suppression method has the following problems.

In the echo suppression method, the echo suppressor 9 inserts / extracts the attenuation amount depending on the level of each of the transmission signal and the reception signal in order to suppress the echo, and the operation time of about several msec is required to insert / extract the attenuation amount. Therefore, there is a drawback that it takes time to recover, and therefore the beginning and end of words or gaps such as gaps occur and it becomes difficult to speak.

On the other hand, by setting the maximum insertion attenuation amount to a small value, some improvement can be made, but on the contrary, there is a drawback that the volume of the speaker cannot be increased and the other party's voice cannot be heard. .

In the echo canceling method, the echo canceller 17 is different from the echo suppressor 9 in that the amount of attenuation is not inserted and removed, a pseudo echo signal is created, and the pseudo echo signal is subtracted from the echo signal to cancel the echo signal. Therefore, unlike the echo suppressor 9, there is no difficulty in talking due to insertion and removal of the attenuation amount,
It has the following drawbacks.

It is generally known that the frequency spectrum of voice or music has a high power density in the low range and a low distribution in the high range. Details are described in [Sound] edited by The Acoustical Society of Japan, published by Corona Publishing Company in 1978. Echo canceller 17
Has a property that, even if a signal having any frequency characteristic is input to the reception signal input terminal Rin13, the frequency characteristic of the echo canceled signal becomes flat at the transmission signal output terminal Sout16.

FIG. 5 is a waveform diagram observing the frequency characteristics of an echo canceller having white noise as an input.

FIG. 6 is a waveform diagram observing the frequency characteristics of an echo canceller having Phos noise as an input.

Transmission when the echo canceller 17 is disabled or enabled when white noise is input to the reception signal input terminal Rin13 and a resistance attenuator (7 dB attenuation) is inserted from the reception signal output terminal Rout14 to the transmission signal input terminal Sin15 The frequency characteristic in the signal output terminal Sout16 is shown.

In the waveform diagrams shown in FIGS. 5 and 6, the vertical axis has a scale of 10
dB, one horizontal scale is 0.5 kHz, the upper waveform is disabled, and the lower waveform is enabled.

In this way, the characteristics that the frequency characteristics become flat after echo cancellation even if the received input noises with different frequency characteristics are observed in the experiment.

Furthermore, the frequency characteristic (echo path characteristic) of the transfer function of the echo signal 3 that circulates from the speaker 1 to the microphone 2 is:
As shown in FIG. 7, since the absorptivity of the wall in the room is lower than 1, it exhibits a low-pass filter characteristic, but the frequency characteristic has a large change in the echo path loss with respect to a slight change in each frequency. This is a phenomenon caused by the sound radiated from the speaker 1 being reflected by the wall and being inserted into the microphone 2.

In addition, it has been reported that the echo path loss changes greatly with a slight change in each frequency, and exhibits a wavelike frequency characteristic. Therefore, if the gain of the amplifier in the echo path is increased by 10 dB so that the volume from the speaker 1 is optimized by the echo cancellation amount obtained by the echo canceller 17, the echo path loss of 0 dB of FIG. The reference point moves to the level shown in A of FIG. 7, and even in a system in which the echo signal sneaking from the speaker 1 to the microphone 2 has a power loss, gain occurs at a certain frequency and howling occurs. Sometimes. Therefore, if an attempt is made to increase the gain of an amplifier having a flat frequency characteristic in the echo path, the gain can be increased only up to the B level point shown in FIG. Further, due to the characteristics of the echo canceller 17, it is not possible to take a larger amount of echo cancellation in the high frequency band of the audio frequency spectrum than in the low frequency band. Thus, the echo canceller
No. 17 had the drawback that the gain equivalent to the absolute value of the echo cancellation amount could not be increased by the amplifier, and the improvement effect for insufficient volume was reduced.

Further, in the frequency shift method, the wavelike frequency characteristic of the echo path characteristic is corrected, but the corrected smoothing amount is about 6 to 10 at most, which is not definitive. Therefore, in order to prevent the howling of the communication system that requires the gain of the amplifier above this value, the realization of each system alone is not sufficient.

The present invention has been made in view of such a conventional problem, and prevents howling of a communication system caused by insertion of an echo signal generated by a received signal through an echo path into a transmission signal. An object of the present invention is to provide a howling suppression circuit capable of increasing the gain of an amplifier to enable comfortable voice communication.

[Means for solving problems]

The present invention includes circuit means for predicting and calculating an echo signal that circulates into the transmitting system with a signal arriving at the receiving system as an input, and means for adding the output of this circuit means to the signal of the transmitting system in an inverted phase. In the howling suppression circuit, a frequency shift circuit for shifting the frequency component of the signal passing through the signal path by a predetermined value is provided in the signal paths of the receiving system and the transmitting system,
The predictive calculation circuit means is of an adaptive operation type and includes control means for controlling the frequency shift circuit into an operating state and a non-operational state. The control means corresponds to the adaptive operation of the predictive calculation circuit means. In the receiving state, the operation of the transmitting system frequency shift circuit is stopped and the operation of the receiving system frequency shift circuit is started, and in the transmitting state, the operation of the transmitting system frequency shift circuit is started and the receiving system frequency is started. The configuration is characterized in that the operation of the shift circuit is stopped, and a control signal for starting the operation of the frequency shift circuits of both systems is transmitted in the non-call state and the simultaneous call state.

[Action]

The present invention is to generate a pseudo echo signal of an echo signal that circulates from a signal arriving at a receiving system to a transmitting system, invert the phase of the pseudo echo signal, and add the echo canceling means to the transmitting system. When the frequency shift circuit for shifting the frequency component of the signal passing through the passage is used together and the frequency shift circuit is inserted in the signal path of the receiving system, the adaptive operation function of the echo canceling means has started. When the frequency shift circuit is inserted in the signal path of the transmission system, the frequency shift function is stopped when the adaptive operation function is stopped and the frequency shift function is started when the adaptive operation function is stopped. By starting the function, the howling suppression is controlled by appropriately controlling the adaptive operation and the frequency shift operation of the echo canceller in each of the non-talking state, the transmitting state, the receiving state and the simultaneous speaking state. Signal Rutotomoni call side can be prevented from being frequency shifted.

The frequency shift function corrects the wave-like frequency characteristic of the echo path characteristic, which can increase the gain of the amplifier.

Therefore, at least one of the echo canceling function and the frequency shift function functions in the loop, and the insufficient volume of the speaker, which is included in the howling suppressing circuit by the echo canceling method, can be eliminated.

〔Example〕

 Hereinafter, embodiments of the present invention will be described with reference to the drawings.

The howling suppression circuit of the present invention is characterized in that the echo canceller and the frequency shift circuit are used together. In this embodiment, an echo canceller will be used as the echo canceller.

FIG. 1 is a block diagram showing an embodiment of a howling suppression circuit according to the present invention. In FIG. 1, an echo canceller 17 includes a detection circuit 20 and a detection circuit 2 for inputting signals from a reception signal input terminal Rin13 and a transmission signal input terminal Sin15.
The control output of 0, the signal from the reception signal input terminal Rin13 and the output of the subtraction circuit 19 are input, and the adaptive control filter 18 that connects the output signal to the subtraction circuit 19 and the transmission signal input terminal Sin15
Is added to the output signal of the adaptive control filter 18 in the inverted phase and is sent to the transmission signal output terminal Sout16.

The circuit of this embodiment shown in FIG. 1 includes two frequency shift circuits.
21 and 22 are the reception path between the reception signal input terminal Rin13 and the reception signal output terminal Rout14, and the transmission signal input terminal Si.
It is characterized in that it is inserted in the transmission path between n15 and the transmission signal output terminal Sout16.

In FIG. 1, the first frequency shift circuit 21 is inserted between the reception signal input terminal Rin13 and the echo canceller 17,
The second frequency shift circuit 22 is inserted between the echo canceller 17 and the transmission signal output terminal Sout16, and the signal from the reception signal input terminal Rin13 input to the detection circuit 20 is taken from the input side of the first frequency shift circuit 21. The signal from the reception signal input terminal Rin13 input to the adaptive control filter 18 is taken from the output side of the first frequency shift circuit 21. The control output from the detection circuit 20 is connected to the first and second frequency shift circuits 21 and 22.

The table is a table showing operating conditions of the detection circuit 20 of the present embodiment.

The operation of the circuit of this embodiment will be described below with reference to FIG. 1 and the table.

First, the detection circuit 20 determines whether the reception level LRin and the transmission level LSin of the signals respectively input to the reception signal input terminal Rin13 and the transmission signal input terminal Sin15 exceed the set level (identification level) Lth that can be recognized as voice power. Identify. When both the transmission level LSin and the reception level LRin do not exceed the identification level Lth (no call)
First, the operation of the adaptive control filter 18 is stopped and the operations of the first and second frequency shift circuits 21 and 22 are started. When the transmission level LSin exceeds the identification level Lth and the reception level LRin does not exceed the identification level Lth (sending state),
The operation of the adaptive control filter 18 is stopped, the operation of the first frequency shift circuit 21 is started, and the operation of the second frequency shift circuit 22 is stopped. When the reception level LRin exceeds the discrimination level Lth and the transmission level Lsin does not exceed the discrimination level Lth (listening state), the operation of the adaptive control filter 18 is started and the operation of the first frequency shift circuit 21 is stopped. , Starts the operation of the second frequency shift circuit 22.

Further, the detection circuit 20 has a transmission level LSin and a reception level.
When both LRin exceed the discrimination level Lth, if the difference between the reception level LRin and the transmission level LSin is larger than the minimum loss amount ERL, it means that the receiving state is established, and the operation of the adaptive control filter 18 is started. The operation of the frequency shift circuit 21 is stopped, and the operation of the second frequency shift circuit 22 is started. Reception level
If the difference between LRin and the transmission level LSin is smaller than the minimum loss amount ERL, it means that the call is in the simultaneous call state, the operation of the adaptive control filter 18 is stopped, and the operations of the first and second frequency shift circuits 21 and 22 are started. To do.

In this way, the detection circuit 20 sends a control operation command to the adaptive control filter 18, the first frequency shift circuit 21, and the second frequency shift circuit 22.

A control method called a learning identification method is currently adopted as an example of the adaptive control filter 18.
Received signal input terminal ERin from the received signal input terminal Rin13 input to 18, the input signal of the subtraction circuit 19, that is, the transmission input signal ESin from the transmission signal input terminal Sin15, the output signal of the subtraction circuit 19, that is, the transmission signal output terminal Sout16 By using the three signals of the transmission output signal ESout, the transmission output signal ESout is output in the operation start state of the adaptive control filter 18 shown in the table.
The reverberation signal is erased by automatically controlling so that the power of 0 is always zero. The stopped state of the adaptive control filter 18 means that the adaptive operation is stopped, and the pseudo echo signal is sent to the subtraction circuit 19 and the echo signal is erased.

Further, the frequency shift circuits 21 and 22 have a memory approximately equal to the pitch time interval of voice, sample the input signal at a constant time interval, store the sample value in the memory, and then store the stored sample value. Is read out at an interval longer than the sampling interval at the time of inputting, and an output signal equivalent to the frequency shifted to the low frequency side is generated. Here, the stop operation of the frequency shift circuits 21 and 22 means stopping the frequency shift function, and is equivalent to passing the input signal as it is.

Generally, in the audio system of a video conference telephone device, the devices shown in FIG. 1 are arranged to face each other via a communication path, and when one of them is in a call (during one-way call), an adaptive control filter is used.
Although the operation of 18 is stopped and the operation of the frequency shift circuits 21 and 22 is started or stopped, the device on the opposite side performs the opposite operation and the frequency distortion does not occur on the call side. Further, the howling margin in the four-wire loop system via the communication path is the sum of the echo cancellation amount and the smoothing amount due to the frequency shift, which makes it possible to further increase the gain of the amplifier.

In the above specific example, a system having a speaker and a microphone for four-line audio of the video conference telephone device has been described as an example, but the present invention is also applicable to the case where the cause of howling is reflection which is not necessarily an acoustic system. Therefore, the present invention can be similarly applied to a four-line section system of a telephone network having a two-wire / four-wire conversion circuit and connecting one or more grounds.

[Effects of the Invention] As described above, the present invention outputs the transmission signal without inserting the frequency shift by stopping the operation of the adaptive filter in the transmission state. Therefore, the communication quality of the communication signal from the communication side is improved. Since there is no deterioration of, when considering in a four-wire loop system,
At least one of the echo canceling function and the frequency shift function functions in the loop. Therefore, the sum of the echo canceling amount and the smoothing amount by the frequency shift becomes a howling margin, so that the gain of the amplifier can be increased. Has the effect that can be done. That is, there is an effect that an insufficient volume of the speaker, which is included in the howling suppression circuit by the echo canceling method, which has been used up to now, can be solved and an optimum volume can be obtained.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention. FIG. 2 is a block diagram showing a configuration example of a voice system of a conventional video conference telephone device. FIG. 3 is a block diagram of a howling suppression circuit using an echo suppressor. FIG. 4 is a block configuration diagram of a howling suppression circuit using an echo canceller. FIG. 5 is a waveform diagram in which the frequency characteristics of an echo canceller with white noise as an input are observed. FIG. 6 is a waveform diagram observing the frequency characteristics of an echo canceller with Phos noise as input. FIG. 7 is a diagram showing an example of indoor echo characteristics. 1 ... speaker, 2 ... microphone, 3 ... echo signal, 4
… Speaker's voice signal, 5,6,7,8… Amplifier or attenuator, 9… Echo suppressor, 10,11… Variable attenuation circuit, 12… Control circuit, 13… Reception signal input terminal Rin, 14… Reception Signal output terminals Rout, 15 ... Transmission signal input terminals Sin, 16 ...
Transmission signal output terminals Sout, 17 ... Echo canceller, 18 ... Adaptive control filter, 19 ... Subtraction circuit, 20 ... Detection circuit, 21, 22
… Frequency shift circuit.

Continuation of the front page (72) Inventor Ryoji Kijima 2356 Take 1, Yokosuka City, Kanagawa Prefecture, Yokosuka Telecommunications Research Laboratories, Nippon Telegraph and Telephone Corporation (56) Reference JP-A-58-3430 (JP, A) JP-A-61 -202532 (JP, A)

Claims (1)

[Claims] 1. A circuit means for predicting and calculating an echo signal that circulates into a transmitting system with a signal arriving at the receiving system as an input, and means for adding the output of this circuit means to the signal of the transmitting system in an inverted phase. In the howling suppression circuit including, a frequency shift circuit for shifting the frequency component of the signal passing through the signal path by a predetermined value is provided in the signal path of the receiving system and the transmitting system, and the circuit means for performing the predictive calculation is an adaptive operation type. And a control means for controlling the frequency shift circuit into an operating state and a non-operating state, the control means corresponding to the adaptive operation of the circuit means for performing the predictive calculation, in the receiving state, of the transmitting system frequency shift circuit. The operation is stopped and the operation of the receiving system frequency shift circuit is started. In the transmitting state, the operation of the transmitting system frequency shift circuit is started and the operation of the receiving system frequency is started. The operation of the shift circuit is stopped, the non-calling state and simultaneous talking state, howling suppression circuit, characterized in that the arrangement for sending a control signal for starting the operation of the frequency shift circuit both systems.