JP3381838B2 - Echo canceller for loudspeakers in the hall - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an in-field loudspeaker echo that eliminates or suppresses an echo signal that is a cause of howling and an auditory obstacle when performing voice communication in a relatively large hall such as an auditorium. The present invention relates to an erasing device.

[0002]

2. Description of the Related Art FIG. 2 is a schematic diagram of a voice communication system. In the figure, 1 and 3 are microphones (sound receivers) for transmission, 2 and 4 are reception speakers, 5 and 7 are transmission signal amplifiers, 6 and 8 are reception signal amplifiers, 9 is a transmission line, and 10 is transmission. And 11 are listeners, respectively. The transmission voice uttered by the transmitter 10 includes a transmission microphone 1, a transmission signal amplifier 5, a transmission line 9, a reception signal amplifier 8, and a reception speaker 4.
And is transmitted to the listener 11. Unlike the conventional telephone call system, this voice call system does not require a handset to be held in the hand, so it has the advantages of being able to make a call while working, and also to realize a natural face-to-face call. It is widely used for communication conferences, videophones, and loudspeakers.

On the other hand, as a drawback of this voice communication system, the existence of echo is a problem. That is, in FIG. 2, the voice transmitted from the reception speaker 4 to the reception side is received by the transmission microphone 3, and the transmission signal amplifier 7, the transmission path 9,
The signal is reproduced on the side of the transmitted sound 10 through the reception signal amplifier 6 and the reception speaker 2. For the speaker 10, this phenomenon is a reverberation phenomenon in which the voice uttered by the speaker 10 is reproduced from the receiving speaker 2, and is called an acoustic echo or the like. This reverberation phenomenon causes adverse effects such as call disturbance and discomfort in a loud voice communication system. Further, the sound reproduced from the receiving speaker 2 is received by the transmitting microphone 1 to form a closed loop of the signal. When the loop gain is larger than 1, the howling phenomenon occurs and the call becomes impossible.

In order to overcome such problems of the voice communication system, an echo canceller using an echo canceller and a loss controller is used.

FIG. 3 is a block diagram showing the structure of an example of a conventional echo canceller, and the same parts as those in FIG. 2 are given the same numbers. The amplifier is omitted for simplicity. In this figure, 12 is a loss control device,
A loss control circuit 13 and loss inserters 14 and 15 are provided. An echo canceller 21 includes a pseudo echo path 22, a pseudo echo path estimation circuit 23, and a subtractor 24. Also,
x (n) 25 is a received signal, h (n) 26 is an echo path transfer characteristic (impulse response) between the receiving speaker 4 and the transmitting microphone 3, y (n) 27 is an echo signal, and y∧ (n). Two
8 is a pseudo echo signal, h ∧ (n) 29 is an estimated value of the echo path impulse response, e (n) 30 is an error signal (transmitted signal),
s (n) 31 represents the speaker signal of the near-end speaker, and z (n) 32 represents the microphone output signal. Reference numeral 33 indicates the entire echo canceller. However, n is a parameter indicating time.

Next, the operation will be described. In the echo canceller 21, first, the pseudo echo path estimation circuit 23 estimates the impulse response of the echo path, and its estimated value h∧.
(N) 29 is transferred to the pseudo echo path 22. Next, in the pseudo echo path 22, h∧ (n) 29 and the received signal x (n)
The convolution operation with 25 is performed and the pseudo echo signal y∧
(N) 28 is synthesized. Then, in the subtractor 24,
The pseudo echo signal y∧ from the output signal of the transmitting microphone 3
(N) Subtract 28. If the echo path impulse response is well estimated, the echo signal y (n) 27 and the pseudo echo signal y∧ (n) 28 are substantially equal to each other.
As a result of this subtraction, the echo signal y (n) 27 included in the microphone output is erased.

Here, the pseudo echo path 22 has an echo path transfer characteristic h.
(N) It is necessary to follow the temporal variation of 26. Therefore, the pseudo echo path estimation circuit 23 estimates the echo path impulse response using an adaptive algorithm. This estimation operation is a receiving state, that is, s (n) ≈0, and z
It is executed only when (n) ≈y (n) can be considered. The reason is that the adaptive algorithm uses the error signal e (n) =
The estimated value h∧ of the echo path impulse response of the pseudo echo path 22 so that the square error of z (n) ... y∧ (n) is minimized.
(N) 29 is sequentially estimated. Therefore, in order to obtain a good estimate of h∧ (n), e (n) = y (n) −y∧
(N), that is, e (n) = h (n) * x (n) -h∧
This is because it is necessary to be (n) * x (n) (where * is a symbol indicating a convolution operation). From another perspective, if there is a talker signal, e (n) =
s (n) + h (n) * x (n) -h∧ (n) * x (n)
Therefore, h ∧ (n) that minimizes e (n) is h
Stay away from (n). Normally, it is possible to execute and stop the adaptive operation of the pseudo echo path estimation circuit 23 by determining whether or not e (n) becomes small.

Further, once the estimated value h∧ (n) 29 of the echo path impulse response of the pseudo echo path 22 is the echo path transfer characteristic h.
After converging to (n) 26, the talker signal s (n) 31
Error signal e (n) 30 is e (n) =
s (n) + h (n) * x (n) = h∧ (n) * x (n)
= S (n), which is a transmission signal.

In the echo canceller 33, the loss controller 12 is also used in order to prevent howling even when the echo canceller 21 is in the initial state or when the echo path changes abruptly.

The loss control circuit 13 receives the received signal x (n) 2.
5, and the error signal e (n) 30 which is the output from the echo canceller 21 is input, and the transmission / reception state is determined based on their magnitudes. As a method of determining the transmission / reception state, for example, the short-time power Px of x (n) and e (n) is used.
(N) and Pe (n) are calculated and their sizes are compared. Here, the echo signal y (n) 27 is the received signal x (n)
Assuming that it is smaller than 25, Px (n)> Pe (n) is satisfied when there is no talker signal s (n), and at this time, it is determined to be the receiving state. On the other hand, if there is a talker signal s (n) above a certain level, Px (n) <Pe (n)
Then, at this time, it is determined to be in the transmitting state. Then, if it is determined to be the receiving state, the loss is inserted into the loss inserting unit 15 on the transmitting side. As a result, the reverberant signal y (n) 27 that has sneaked from the receiving speaker 4 and is received by the microphone is attenuated by the loss inserter 15 and the reverberant phenomenon is reduced. On the other hand, if it is determined to be in the transmitting state, the loss of the transmitting side loss inserter 15 is set to 0 (dB), and the loss is inserted into the receiving side loss inserter 14. As a result, the transmission signal received by the transmission microphone 3 is transmitted without being attenuated. Also, by inserting the loss on the receiving side, the closed loop gain can be kept at 1 or less and the howling phenomenon can be prevented.

[0011] Now, in a normal loudspeaker communication conference in a conference room, the talker signal s picked up by the transmitting microphone 3 is used.
(N) is not amplified in the conference room where the talker is. However, when a voice communication conference is held in an auditorium or the like where several tens of people can be accommodated in the same room, it is necessary to increase the voice of the speaker on this side by a speaker in the same room.

In order to perform the voice communication while performing the voice amplification in the room in which the speaker's voice is also amplified, the following two configurations can be briefly considered.

The first structure will be described with reference to FIG. Figure 4
Shows an example of the configuration when a normal echo canceller is used as an in-field loudspeaker, and the same parts as those in FIG. 3 are assigned the same numbers. 41 is the transmission signal e (n)
30 is a terminal for branching, 42 is a branched outgoing signal e
(N) 30 is an adder that adds the received signal x (n) 25.

Now, the estimated value h∧ (n) 29 of the echo path impulse response of the pseudo echo path 22 is the echo path transfer characteristic h (n) 2
Consider a state close to 6, that is, a converged state. In this case,
As described above, the transmission signal e (n) 30 is s
Since it can be regarded as (n) 31, the transmission signal will be referred to as s (n) hereinafter.

When only the reception signal x (n) exists, the reception signal is amplified by the reception speaker 4, and the listener in the hall can hear it. The echo signal in which x (n) passes through the echo path and is picked up by the transmission microphone 3 is suppressed by the echo canceller 21 and is not sent to the other party. On the other hand, the transmission signal s (n) of the speaker who speaks in the room is picked up by the microphone 3, passes through the subtractor 24, and is input to the loss inserter 15 and the loss control circuit 13. At this time, if the reception signal x (n) does not exist, the loss control circuit 13 satisfies Px (n) <Pz (n),
It is determined to be in the transmitting state and the loss is inserted into the loss inserting unit 14 on the receiving side. The transmission signal s (n) is sent to the other side through the transmission line without being lost by the loss insertion unit 15.
Here, in order to perform loudspeaking in the field, a transmission signal which is branched from the branch terminal 41 and added to the reception signal by the adder 42 is input to the reception speaker 4 after passing through the loss inserter 14.
For this reason, the transmission signal to be amplified by the speaker side becomes a signal with a loss inserted, which causes a problem that the speaker cannot be amplified with a loud sound.

As another problem, since the signal obtained by adding the transmitting signal and the receiving signal by the adder 42 becomes the input signal to the echo canceller 21, the echo path transfer characteristic h∧.
(N) The convergence state of 29 is deteriorated. That is, in this configuration example, the signal x (n) + s (n) becomes the input signal to the echo canceller 21, and the echo generated at that time is regarded as h (n) * x (n) + s (n). In some cases, the error signal e (n) 30 becomes small even though the talker signal is mixed, and the execution of the adaptive operation is instructed. When the adaptive operation is executed in such a state, the estimated value h∧ (n) of the echo path impulse response of the pseudo echo path 22 does not become h (n), and the echo cancellation performance deteriorates. Also, s (n), which should not be erased originally, is erased.

Next, the second structure will be described with reference to FIG. FIG. 5 is a schematic diagram showing a second example of the configuration when a normal echo canceller is used as an in-field loudspeaker, and the same numbers are assigned to the same parts as in FIG.
Reference numeral 43 is a terminal for branching the microphone output signal z (n) 32, and 44 is an adder for adding the branched output signal z (n) 32 and the reception signal x (n) 25.

In this configuration, the transmission signal is branched at the branch terminal 43, added to the reception signal by the adder 44, and then amplified into the hall through the reception speaker 4. At this time, since the transmitted signal is not affected by the lossy circuit, it can be amplified in the hall with the same quality as that of the normal public hall. On the other hand, the reception signal x (n) 25 is amplified by the reception speaker 4, passes through the echo path, and is picked up by the transmission microphone 3. The echo signal y (n) 27 is also branched at the branch terminal 43. The received signal is added by the adder 44, and the voice is again amplified by the receiving speaker 4. Therefore, the estimated value of the echo path to be estimated by the echo canceller 21 is the receiving speaker 4
And the estimated value h (n) 26 of the echo path transfer characteristic between the microphone 3 and the transmitting microphone 3 is not h '(n) = h (n) / (1-
h (n) z ⁻¹ ). This response is called an infinite impulse response and has a very long response time. Therefore, the estimated value h of the echo path impulse response corresponding to all the responses
In order to simulate ∧ (n) 29, a very large amount of hardware is required, which causes a problem that the device cannot be actually mounted.

[0019]

As described above,
Using the echo canceling device used for conducting a normal loudspeaking conference, as a configuration for conducting a loudspeaker communication conference while performing public loudspeaking, the transmission signal after passing through the echo canceling device,
If the received signal is added to the received signal and then the received signal is input to the echo canceller, there arises a problem that a loss is inserted in the transmitted signal to be amplified in the field. In addition, as a configuration different from this, when the microphone output signal is added to the received signal immediately before being amplified by the speaker and then amplified by the speaker, the echo path impulse response to be simulated by the echo canceller in the echo canceller is The estimated value becomes an infinite impulse response, which causes a problem that a pseudo echo path with a very large hardware scale must be used.

It is an object of the present invention to provide an echo canceller for in-field loudspeaking, which enables a communication conference while comfortably amplifying both the talker signal and the received signal in the hall.

[0021]

The echo canceller for in-field loudspeaker according to the present invention is to perform loudspeaking in the hall by loudspeaking the received signal and the talker signal received from the transmission line in the same room by the loudspeaker. In an echo canceller for in-field loudspeaker used for an in-field loudspeaker for transmitting a talker signal received by a sound receiver via a transmission path, a speech transmitted by the sound receiver A receiver signal composed of a human signal and an echo signal after the reception signal passes through the echo path, a pseudo echo path estimated from the reception signal, and the reception signal as an input of the pseudo echo path. An error signal is obtained by subtracting the pseudo echo signal obtained by the above from the sound receiver signal, and a subtracter that uses this error signal as a transmission signal to the transmission line, and further adds the transmission signal and the reception signal. And this The calculated signal is one having an adder to the input of the loudspeaker for Supikahe.

Further, the pseudo echo path is estimated only when it is judged that the speaker signal is not mixed with the sound receiver signal.

Further, a reception signal loss insertion device for inserting a loss into the reception signal, a transmission line transmission signal loss insertion device for inserting a loss into the transmission signal to the transmission line, and a transmission signal to be an input to the adder. It has a loudspeaker transmission signal loss inserter for inserting a loss.

Also, a loss control circuit for controlling the received signal loss inserter, the transmission line outgoing signal loss inserter, and the loudspeaking outgoing signal loss inserter is provided, and the loss control circuit comprises:
When only the reception signal exists, or when both the reception signal and the talker signal do not exist, a loss is given to the transmission path transmission signal loss inserter and the loudspeaking transmission signal loss inserter, and the talker In the case where only a signal is present, a loss is given to the reception signal loss inserter, and both the reception signal and the talker signal are present, and the reception signal is larger than the talker signal. Is applied to the transmission line transmission signal loss inserter, and when the reception signal is smaller than the sender signal, the reception signal loss insertion device is imparted with loss. A loudspeaker transmission signal loss inserter does not give a loss.

Further, a frequency modulator for performing frequency modulation processing on the speaker signal which is an input to the adder is provided.

[0026]

BEST MODE FOR CARRYING OUT THE INVENTION In order to solve the problem that arises when simultaneously performing loudspeaker communication while performing loudspeaking in the field, the present invention provides a speaker signal after passing through an echo canceller.
It is characterized in that it is branched into two, one of which is used as a transmission signal, and the other is added to a reception signal and amplified by a speaker.
However, as the input of the echo canceller, only the reception signal before addition is used. Therefore, the input of the echo canceller is only x (n), and since the echo canceller operates to cancel the echo of h (n) * x (n) using the received signal x (n), the correct h∧ (N) can be estimated. Further, among the talker signal s (n) and the echo signals h (n) * x (n) picked up by the microphone, the echo signals h (n) * x (n) are canceled by the echo canceller. , S (n) + x (n) are amplified and the echo signals h (n) * x (n) are not amplified again.

Further, the present invention is characterized in that the loss is independently inserted into each of the two branched transmission signals. As a result, the signal sent to the transmission line is suppressed and transmitted when the received signal is larger than the talker signal as in the case of a normal voice conference, but the signal transmitted to the speaker in the hall is transmitted. The speech signal can be amplified without loss insertion.

[0028]

EXAMPLE An example of the present invention will be described with reference to FIG. However, the same numbers are assigned to the same parts as in FIG.
45 is a branch terminal for branching the error signal e (n) 30;
46 is a loss inserter, 47 is an adder, and 48 is a frequency modulator.

Next, the operation will be described. The error signal e (n) 30 of the echo canceller 21 is obtained when the echo canceller 21 is operating correctly, that is, h∧ (n) = h.
In the state of (n), it coincides with the talker signal s (n) 31. The error signal e (n) 30 is 2 at the branch terminal 45.
One is input to the loss control device 12, and the other is input to the loss inserter 46. The signal passing through the loss inserter 46 is added to the reception signal x (n) 25 by the adder 47 via the frequency modulator 48, and becomes the input signal to the reception speaker 4.

In the echo canceller 21, the estimated value h
Since the signal input to the pseudo echo path estimation circuit 23 for estimating ∧ (n) 29 is only the reception signal x (n) 25, the microphone reception output signal z (n) 32 is assumed to be the speaker signal s.
When (n) 31 is mixed, z (n) = h
The pseudo h∧ (n) 29 is estimated so that (n) * x (n) + s (n) is canceled by the received signal x (n) 25, but in this case, the echo cannot be canceled correctly. ,
That is, since the error signal e (n) 30 does not become small,
It can be easily determined that the transmission signal is mixed, and the adaptive operation of the pseudo echo path estimation value h (n) 29 is not performed. On the other hand, from the receiving speaker 4, the receiving signal x (n)
Both 25 and the talker signal s (n) 31 can be amplified.

Further, in the present invention, the loss control device 12 provided to prevent howling in the loudspeaker communication system has a structure that does not affect the loudspeaker in the field even if the conventional operation is performed. ing. Also, the loss inserter 46
Inserts a loss when only the reception signal x (n) 25 exists, and does not insert a loss when the talker signal s (n) 31 exists.

By carrying out such an operation, even when only the reception signal x (n) 25 is present and the operation of the echo canceller 21 is in an unstable state, the error signal e (n) 30 thereof is generated. It is possible to prevent the voice from being amplified by the receiving speaker 4 again. Further, when the talker signal s (n) 31 is present, no loss is inserted, so that it is possible to make a loud sound in the hall at a desired volume. However, with respect to the speaker signal s (n) 31, the signal added by the adder 47 and output from the receiving speaker 4 is input again to the microphone 3 for transmission, a closed loop is formed, and there is concern about howling in loudspeaking in the hall. To be done. Therefore, the frequency modulator 48 can be inserted in front of the adder 47 to increase the howling margin. Here, the frequency modulator 48
Can be replaced with various anti-howling devices such as howling suppressors and automatic equalization devices.

[0033]

The echo canceller for in-field loudspeaking of the present invention receives the received signal and the talker signal received from the transmission path in the same room by the loudspeaker for loudspeaking while receiving them by the sound receiver. In an echo canceller for use in an in-field loudspeaker for transmitting the sounded speaker signal through a transmission line to hold a communication conference, the speaker signal received by the sound receiver and the received speech A sound receiver signal composed of an echo signal after the signal has passed through the echo path, a pseudo echo path estimated from the received signal, and a pseudo echo obtained by inputting the received signal to the pseudo echo path. An error signal is obtained by subtracting the echo signal from the receiver signal,
Since there is a subtractor that uses this error signal as a signal to be sent to the transmission line, and an adder that adds the sent signal and the received signal and inputs this added signal to the speaker for loudspeaker, When a communication conference is conducted while performing, the echo signal generated by the reception signal does not affect the cancellation of the reception echo signal at all, even though the transmission signal is amplified in the room. It is possible to cancel the received echo as in a normal voice conference. Also, with respect to the talker signal, it is possible to transmit to the other party with the same quality as in a normal public address conference, and maintain the same quality as in the case of normal in-place public address. It becomes possible.

Further, since the pseudo echo path is estimated only when it is determined that the talker signal is not mixed in the sound receiver signal, the correct pseudo echo path is always estimated. You can

Further, a reception signal loss insertion device for inserting a loss into the reception signal, a transmission line transmission signal loss insertion device for inserting a loss into the transmission signal to the transmission line, and a transmission signal to be an input to the adder. Since there is a loudspeaker transmission signal loss inserter for inserting a loss, it is possible to selectively give a loss only to a necessary signal.

Also, a loss control circuit for controlling the reception signal loss inserter, the transmission line outgoing signal loss inserter, and the loudspeaking outgoing signal loss inserter is provided, and the loss control circuit comprises:
When only the reception signal exists, or when both the reception signal and the talker signal do not exist, a loss is given to the transmission path transmission signal loss inserter and the loudspeaking transmission signal loss inserter, and the talker In the case where only a signal is present, a loss is given to the reception signal loss inserter, and both the reception signal and the talker signal are present, and the reception signal is larger than the talker signal. Is applied to the transmission line transmission signal loss inserter, and if the reception signal is smaller than the sender signal, the reception signal loss insertion device is inflicted with loss. Since a loss is not given to the transmission signal loss inserter for use, it is possible to selectively give a loss only to a necessary signal.

Further, since a frequency modulator for performing frequency modulation processing is provided for the speaker signal which is input to the adder, a howling margin can be obtained.

As described above, by adopting the present invention, it is possible to solve various problems that occur when simultaneously performing in-field loudspeaking and loudspeaker communication, and it is possible to perform comfortable on-site loudspeaker type loudspeaker communication. .

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an echo canceller for in-field loudspeaking according to the present invention.

FIG. 2 is a block diagram showing a configuration of conventional voice communication.

FIG. 3 is a block diagram showing a configuration of a conventional echo canceller.

FIG. 4 is a block diagram showing a first configuration example in the case of performing in-field loudspeaking by using a conventional echo canceller.

FIG. 5 is a block diagram showing a second configuration example in the case of performing in-field loudspeaking using a conventional echo canceller.

[Explanation of symbols]

1,3 speaker microphone 2,4 earpiece speaker 12 Loss control device 13 Loss control circuit 14,15 Loss inserter 21 Echo canceller 22 Pseudo echo path 23 Pseudo echo path estimation circuit 24 Subtractor 25 Received signal 26 Echo Path Transfer Characteristics 27 Echo signal 28 Pseudo echo signal 29 Estimated value of echo path impulse response 30 Error signal 31 Near-end talker signal 32 microphone output signal 33 Echo canceller 45 branch terminals 46 loss inserter 47 adder 48 frequency modulator

─────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Junko Sasaki 3-19-2 Nishishinjuku, Shinjuku-ku, Tokyo Nihon Telegraph and Telephone Corporation (72) Inventor Masayuki Yamada 1-1-1, Gotenyama, Musashino-shi, Tokyo No. 3 in NTT Advanced Technology Co., Ltd. (56) Reference JP-A-7-226700 (JP, A) JP-A-11-18191 (JP, A) JP-A-3-114344 (JP, A) JP 58-205397 (JP, A) JP 60-239199 (JP, A) JP 1-174019 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) ) H04B 1/76 H04B 3/00 H04B 7/00 G10L 21/00 H04R 3/00 H04M 1/00 INSPEC (DIALOG) JISST file (JOIS)

Claims (5)

(57) [Claims] 1. A speaker signal received by a sound receiving device and a talker signal received from a transmission line in a same room by a speaker for sound amplification, and the speaker signal received by a sound receiver is transmitted through the transmission line. In an echo canceller for in-field loudspeaker used for in-field loudspeaker for transmitting via a communication conference, the speaker signal received by the sound receiver and the echo signal after the received signal passes through the echo path. And a pseudo echo path estimated from the received signal, and a pseudo echo signal obtained by inputting the received signal to the pseudo echo path is subtracted from the received signal. The error signal is obtained by the subtractor, and the subtractor that uses this error signal as a transmission signal to the transmission line, and the adder that adds the transmission signal and the reception signal and inputs this addition signal to the loud speaker With An echo canceller for in-field loudspeaker, which is characterized by: 2. The loudspeaker system according to claim 1, wherein the pseudo echo path is estimated only when it is determined that the speaker signal is not mixed with the speaker signal. Echo canceller. 3. A reception signal loss inserter for inserting a loss into the reception signal, a transmission line transmission signal loss inserter for inserting a loss into a transmission signal to the transmission line, and a transmission signal to be an input to the adder. The echo canceller for in-field loudspeaker according to claim 1, further comprising a loudspeaker transmission signal loss inserter for inserting a loss. 4. A loss control circuit for controlling the received signal loss inserter, the transmission line outgoing signal loss inserter, and the loudspeaking outgoing signal loss inserter, wherein the loss control circuit has only a received signal. When both the receiving signal and the talker signal do not exist, a loss is given to the transmission path sending signal loss inserter and the loudspeaking sending signal loss inserter, and only the talker signal exists. In the case where a loss is given to the reception signal loss inserter and both the reception signal and the talker signal are present, and the reception signal is larger than the talker signal,
A loss is given to the transmission line outgoing signal loss inserter, and when the receiving signal is smaller than the sender signal, a loss is given to the receiving signal loss inserting device. The echo canceller for in-field loudspeaker according to claim 3, wherein no loss is given to the loss inserter. 5. The echo canceller for in-field loudspeaker according to claim 1, further comprising a frequency modulator that performs frequency modulation processing on the speaker signal that is input to the adder.