JPS59103496A - Howling suppressing device - Google Patents

Abstract

PURPOSE:To correct a virtual impulse response and to eliminate the feedback of a signal from a speaker to a microphone by using the virtual impulse response formed by an impulse response approximating circuit, subtracting a signal simulating a signal fed back from the speaker output to the microphone and using a residual signal only. CONSTITUTION:The impulse response approximation circuit 25 approximating the impulse response between a speaker and a microphone in a speaker system comprising the amplifier 24 and the speaker 22 is provided. The impulse response formed by the impulse response approximation circuit is used from the input signal of microphone, a signal simulating a signal fed back to the microphone from the speaker output is subtracted, only the said residual signal is used so as to minimize the residual signal, and the approximated impulse response is corrected sequentially. A signal fed back from the speaker to the microphone is eliminated without using a special detecting signal, allowing to suppress howling.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a howling suppression device capable of suppressing howling in a public address system composed of a microphone, an amplifier, and a speaker.

Configuration of conventional example and its problems Figure 1 shows block 3 of a conventional howling suppression device.
-7 Show the diagram. In FIG. 1, 1 is an amplifier, 2 is a speaker, 3 is a microphone, 4 is an adaptive filter, and 5 is an adder.

This system uses a detection signal in advance, and uses the above-mentioned adaptive filter 4 to detect when the feedback loop between microphone 3, amplifier 1, and speaker 2 is cut.
By setting the transfer function Hz between the two, the feedback loop of the speaker output signal to the microphone 3 is eliminated. In such a device, since the transfer function Hz must be set in advance using the detection signal, the microphone 3. There is a drawback that the transfer function H2 in that state must be reset each time depending on the set position of the speaker 2.

FIG. 2 is a block diagram of another conventional lining suppression device. In Figure 2, 2 is a speaker, 1 is an amplifier,
6 is an adder, and 7 is a microphone for the main signal.

8 is the microphone for inputting the speaker output signal, 9 is the speaker 2
This circuit creates successive approximations of the impulse responses between the microphones 7 and 8.

Here, hj is the impulse response created there and at the time of tacho, Sj is the signal at time j, Xj is the speaker output signal at time j, h is the impulse response 1g between speaker 2 and microphone 7 is the signal Sj and microphone 8 Impulse response between - yjFij input signal to microphone 7, y
j is the output signal of the impulse response approximation circuit 9 at time j, e
j is a signal of the difference between yj at time j and Δy,l. This device uses the above-mentioned signal a
By modifying hj so as to minimize j, unblchj is approximated, and from hK, the feedback signal from the speaker 2 to the microphone 7 is removed, and no-ring is suppressed. However, these signals represent temporally discrete quantities.

Here, the relationship between the respective signals is as shown in the following equation.

△ yj - (Sj Og1Xko) Rice hj...
・1,2elj=7j−yj=(Sj+Xj−*h)−(Sj｜g−1−Xj)
Zhu hj-Sko rice (1-g*hj month-Xj well (h'-hj
)......1,352, -Z From equation 1.3, it can be seen that even if hj is approximated to h and the term of Xj is removed, the influence of Sj on microphone 8 still appears. . In order to eliminate this effect, the input Sj and XjO ratio at microphone 7 should be increased, and the input Sj at microphone 8 should be increased.
It is necessary to reduce the ratio between j and Xj, which limits the positional relationship between the microphone 7.8 and the Sj speaker 2, resulting in a disadvantage that the range of use becomes narrow.

Is Figure 3 a block diagram of a conventional echo canceller (considered to be a type of howling suppression device)? be. In FIG. 3, 12 is an adder. 1o is a reverberation path including a 2-wire-to-4-wire transition section that produces an echo between the sending path 13 and the receiving path 14 in a private telephone line.

h" is the impulse response of the echo path 10, 11 is an impulse response approximation circuit that approximates the impulse response in the echo path 1o, hj is the pseudo impulse response created there, xj is the transmission signal tyj is the echo signal s"lj is X
A pseudo echo signal resulting from the convolution operation of jj and hj, ej indicates the difference signal between yj and yjO. However, each signal 6 vane indicates a temporally discrete quantity, J represents an arbitrary time point, and the relationship between the respective signals is as shown in the following equation.

yj = Xj rice h ・・・・・・・・・・・・1,4 △.

y co=xj rice hj・・・・・・・・・・・・
・1,5ej=yko yj −Xj rice (h −hj)・・・・・・・・・・・・
1.5 Therefore, if hj is modified to approximate hVc, the echo signal yj will be completely removed. However, in this case, a two-person method requiring a difference signal ej and a transmission signal Xj had to be used to correct %hj.

OBJECTS OF THE INVENTION The present invention provides a howling suppression device that eliminates the above-mentioned drawbacks of conventional devices, does not require a detection signal, uses only one residual signal, and can be used in any situation. The purpose is to

Composition of the Invention The howling suppression cylinder of the present invention has a microphone, an 7 l
- It has an impulse response approximation circuit that approximates an impulse response between a speaker and a microphone in a loudspeaker system consisting of a knob and a speaker, from a microphone input signal.

Using the impulse response created by this impulse response approximation circuit, subtract a signal that simulates the signal fed back from the microphone from the speaker output, and use only the residual signal to minimize the residual signal. By successively modifying the impulse response, the signal fed back from the speaker to the microphone is removed without using a special detection signal, and the Howlig is suppressed.

Description of examples The present invention will be described below with reference to drawings of embodiments.

FIG. 4 shows an embodiment of the present invention.

In FIG. 4, 21 is an adder and 22 is a speaker.

23 is a microphone, 24 is an amplifier, 26 is an impulse response approximation circuit, hj is the impulse response created here, xj
is the speaker output signal, Sj is the main signal, h is the impulse response between the speaker 22 and the microphone 23,
yj is the input signal of the mask 230, yj is the output signal of the impulse response approximation circuit 26, ej is the difference signal b between yj△ and yj, hamp is the impulse response of the amplifier 24, sh Bp n, and the speaker 22.

Xj is an input signal to the speaker 22. Each signal represents a temporally discrete sequence, and j represents an arbitrary time point j.

The relationship between these signals is expressed by the following equation.

Half yj = Sj + Xj rice h ・・・ ・・・・・・
2,1ej=yj−yj °゛°゛°°゛°“−°−2,2 Also, Xj=ej−1*hamp*hsp−・−・・−2,3
Because it is.

7] = ej −1y\hj −−−−−−−
−2.4, then from equations 2 and 2, ej = sj +
j-4 rice hamp-J' hsp rice h elj
* + hj.

97, -ノ (However, the operation symbol thread indicates a convolution operation.) Therefore, to approximate hj to hamp"sp rice, h
By modifying j, the microphone 23 from the speaker 22
-\The feedback signal can be removed, and howling can be suppressed from this.

Here, in order to approximate hj to hamp rice hs, rice h, as the correction amount Δhj of hj, △hj=((ej-sign(a-)/1el-4-2
+a ko 1 Here, sign(ei 1 ) : 61-1 sign (N is the length of impulse response hj) (2> 20) as.

hj−z=hj+△hj °°゛°°°”°“−°−2
- By correcting γ, it is possible to converge in the direction of approximating hjehamp*hsp*h by gradually decreasing ej.

Effects of the Invention As detailed above, according to the present invention, from a microphone input signal, a pseudo impulse response created by an impulse response approximation circuit that approximates an impulse 1o Hoeno response between a speaker and a microphone is used to By subtracting a signal that simulates the signal fed back to the microphone from the output and using only the residual signal, the pseudo impulse response is made to gradually approach the impulse response between the speaker and the microphone. It is possible to modify the impulse response and eliminate the feedback of the signal from the speaker to the microphone, and it is possible to provide a howling suppression device that can adapt to any situation with only one residual signal without using a detection signal. has advantages.

[Brief explanation of the drawing]

1, 2, and 3 are block diagrams showing conventional howling suppression devices, and FIG. 4 is a block diagram showing an embodiment of the howling suppression device of the present invention. 22°°゛°°° speaker, 23°°°...microphone,
24... Amplifier, 25... Impulse response approximation creation circuit. Name of agent: Patent attorney Toshio Nakao and one other person. Figure 1 Figure 2 Figure 3 Figure 1.3 Figure 4

Claims (1)

[Claims] (1) In the public address system consisting of microphones, amplifiers, and speakers,
An impulse response approximation circuit is provided that sequentially simulates the impulse response between the speaker and the microphone in order to remove the signal fed back from the speaker output to the microphone from the microphone input signal, and the impulse response is calculated from the microphone input signal. Using the impulse response created from the approximation circuit, subtract the pseudo speaker output signal that is a copy of the signal output from the speaker and fed back to the microphone, and use only the remaining residual signal to calculate the residual signal in the impulse response approximation circuit. A Heuring suppression device characterized in that the impulse response is successively modified by using a learning fixation method to minimize the signal. (For the residual signal, input the microphone input signal to an adder circuit, and branch the output signal of this adder circuit into two. Page 2) Input one branched signal to an amplifier, and input the other branched signal. In order to create a simulated impulse response between the speaker and the microphone using only this other branch signal, a learning identification method is used to successively modify the pseudo impulse response so that the other branch signal is minimized. Then, this other branch signal is input to an impulse response approximation circuit having a function of performing a convolution operation with the pseudo impulse response, and the output signal of the impulse response approximation circuit (the result of the convolution operation) is input to the addition circuit. The now-ring suppression device according to claim 1, characterized in that the device is configured to be created by inputting in reverse phase.