JP2001275182A - Howling suppression device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To surely suppress howling in a loudspeaking system in a simple configuration. SOLUTION: A delay unit 2 and an adaptive filter 3 delay a microphone input signal y(n) to generate a sine component b(n)=sin(ω0*t+θ'+θ") delayed from the howling component sin(ω0*t+θ) by 2nπ and a subtractor 4 is used to subtract the component bn from the microphone input signal y(n) to suppress the howling.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a howling suppression apparatus for suppressing howling in a loudspeaker system by adaptive processing.

[0002]

2. Description of the Related Art In general, howling in a loudspeaker system is performed as follows.
This is a kind of oscillation phenomenon that occurs when the closed loop gain of the loudspeaker exceeds "1", and has the following features. (1) Start from the highest gain frequency in the closed loop. (2) It changes more slowly than sound / voice.

The following methods (1) to (3) are used to suppress howling. (1) The loop gain is acoustically reduced by using a unidirectional microphone and controlling the directivity of a sound source using a loudspeaker array or the like. (2) By using a narrow-band filter, the loop gain near the frequency at which howling occurs is electrically lowered. (3) The feedback component of the closed loop is suppressed by electrical processing. As a conventional example, as shown in the Proceedings of the Acoustical Society of Japan, March 1991, p. A method has been proposed in which an error signal is used as a loudspeaker signal, and a loudspeaker signal is input to an adaptive filter via a delay circuit so as not to directly cancel the sound, and the adaptive filter is adapted to cancel only the feedback component.

However, the above methods (1) to (3)
Have the following problems. The above method (1)
Is a simple method, but we cannot expect great suppression. The above method (2) requires an operation to analyze and adjust the feedback component in advance, and cannot cope with a change in the feedback component. The above method (3) requires a filter corresponding to the length of the feedback component due to the nature of the processing, and increases the scale of hardware. As another conventional method, for example, Japanese Patent Application Laid-Open No. Hei 5-137191 discloses a method of detecting a howling frequency based on a convolution operation result of an adaptive filter and suppressing a howling frequency component based on the detection result. Although proposed, this method complicates the circuit, and if the howling frequency is erroneously detected, the howling frequency component cannot be suppressed.

The present invention has been made in consideration of the above-described problems of the related art, and has as its object to provide a howling suppressing device that can reliably suppress howling with a simple configuration.

[0006]

According to the present invention, in order to achieve the above object, a phase of a convolution operation result by an adaptive filter is delayed by an integral multiple of a period of a howling component included in a microphone input signal, By subtracting the result of the convolution operation from the microphone input signal, the howling component is subtracted from the microphone input signal including the original sound and the howling component.

That is, according to the present invention, a delay means for delaying a microphone input signal containing an original sound and a howling component so that the original sound is not canceled, and a signal delayed by the delay means, which is sequentially delayed for each sample, to Convolution means for multiplying the sample by a convolution coefficient and adding each multiplication result to perform a convolution operation; means for subtracting the operation result of the convolution operation means from the microphone input signal to generate an error signal; A howling suppression device comprising: a coefficient updating unit that updates each convolution coefficient based on the error signal so that a phase of a calculation result of the calculation unit is delayed by an integral multiple of a period of a howling component included in the microphone input signal. Is provided.

According to another aspect of the present invention, the first convolution coefficient is fixed to 1 by an adaptive filter, and the phase of the convolution operation result is an integer multiple of the period of the howling component included in the microphone input signal. Is updated so that the delay of
By subtracting the result of the convolution operation from the signal obtained by adding the microphone input signal and the delay signal thereof, the howling component and the delay signal of the microphone input signal are subtracted from the microphone input signal including the original sound and the howling component.

That is, according to the present invention, a delay means for delaying a microphone input signal containing an original sound and a howling component so that the original sound is not canceled, and a signal delayed by the delay means, which is sequentially delayed for each sample, to Convolution operation means for performing convolution operation by multiplying a sample by a convolution coefficient and adding each multiplication result; adding the microphone input signal and the signal delayed by the delay means; and calculating the operation result of the convolution operation means. Means for generating an error signal by subtraction, the first convolution coefficient of the convolution operation means being fixed to 1, and a phase of a howling component included in the microphone input signal being added to the phase of the operation result of the convolution operation means. The second and subsequent convolution coefficients are updated based on the error signal so that a delay of an integral multiple of And coefficient updating means, howling suppression apparatus comprising is provided.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS <First Embodiment> An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram modeling a first embodiment of a howling suppression device according to the present invention, and FIG. 2 is a waveform diagram for explaining processing of the howling suppression device of FIG.

FIG. 1 shows a state in which howling is occurring. In the case where a signal including an original sound x (n) and a howling component is input to a microphone, the output signal of the adder 1 is y (n), and the output signal of a speaker is Is a diagram in which an error signal output from an adder (subtractor) 4 is modeled as e (n). At this time, the output signal y (n) of the microphone is, as shown in detail in FIG. 2A, the input sound x (n) which is the original sound and the error signal e (n).
(Output of the subtractor 4) with respect to the impulse response h (n),
Time waveform sin of the frequency that caused howling at a certain gain
(ω0 * t + θ) can be obtained as a signal. y (n) = x (n) + sin (ω0 * t + θ) where n indicates a sample, and ω0 indicates an angular velocity of a frequency at which howling occurs because the closed-loop gain exceeds “1”.

Therefore, a component delayed by 2nπ from the howling component sin (ω0 * t + θ) included in the microphone input signal y (n) is generated, and this is howled component sin (ω0 * t +
By subtracting from (θ), the howling component sin (ω0 * t + θ) can be removed from the microphone input signal y (n). Therefore, in the first embodiment, the microphone input signal y (n) is delayed by the delay unit 2 and the adaptive filter 3, and the component sin (ω0) delayed by 2nπ from the howling component sin (ω0 * t + θ).
* t + θ + θ ″) = b (n) is generated, and howling is suppressed by subtracting this from the microphone input signal y (n) by the subtractor 4.

More specifically, first, the output y (n) = x (n) + sin (ω0 * t + θ) of the microphone (adder 1) is applied to the subtracter 4 as a desired response of the adaptive filter 3. At the same time, the delay is delayed by the delay unit 2 so that the subtractor 4 does not cancel the voice x (n). Output a (n) of delay unit 2
Is the following signal as shown in FIG. a (n) = y (nm) = x (nm) + sin (ω0 * t + θ + θ ′) where m is the number of delay samples

The output a (n) of the delay unit 2 is
Is applied to the convolution operation unit 3a, which performs the convolution operation. Here, the convolution arithmetic unit 3a generates samples a (0), a (1), a (2) to which the signal a (n) is sequentially delayed for each sample, and Coefficients w0, w for (0), a (1), a (2)
1, multiply w2 ~ and add each multiplication result (convolution operation)
I do. This addition result Σwk · a (nk) = b (n) is obtained by subtracter 4
Is subtracted from the microphone output y (n), which is the desired response of the adaptive filter 3, and the error signal e (n) = y (n) -b
(n) is generated.

In the first embodiment, the coefficient updater 3b of the adaptive filter 3 is (ω0 * t + θ + θ ")-(ω0 * t + θ) = 2nπ, where θ": adaptive The phase delay by the filter 3 is delayed so that n = 1, 2,..., And a signal b (n) = x (nm) + sin (ω0 * t + θ + θ) as shown in FIG. "). The howling component sin (ω0 * t +
Adaptive processing is performed so that the phase of θ) rotates by 2nπ, and the power of the error signal e (n) is minimized.

Next, when the output b (n) = sin (ω0 * t + θ + θ ′) of the adaptive filter 3 is subtracted from the microphone input y (n) by the subtractor 4, the output of the subtractor 4 after the adaptation is obtained. e '(n)
Is e '(n) = y (n) -b (n) = x (n) -x (nm) as shown in FIG. 2 (4). Therefore, the howling component sin is obtained from the microphone input signal y (n). (ω0 * t + θ) can be subtracted.

<Second Embodiment> The first embodiment
In the embodiment, the howling component sin (ω0 * t + θ) can be subtracted from the microphone input signal y (n), but the output e ′ (n) of the subtractor 4 after the adaptation is as shown in FIG. As shown in (1), there is a problem that the output e (n) includes the original sound x (n) and the echo component which is a delay component −x (nm), and the output e (n) has the frequency characteristic of the comb filter. Therefore, the second embodiment is configured to remove both the howling component sin (ω0 * t + θ) and the echo component x (nm).

The second embodiment will be described in detail with reference to FIGS. In the second embodiment, as shown in FIG. 3, the adder 4 outputs the microphone input signal y as follows.
(n) and the output a (n) of the delay unit 2 are added, and the addition result y (n) +
It is configured to subtract the convolution operation result b (n) from a (n). e (n) = y (n) + a (n) −b (n) = x (n) + sin (ω0 * t + θ) + ｛x (nm) + sin (ω0 * t + θ +
θ ')｝ − b (n)

Here, the microphone input signal y (n) and the delay unit 2
The output a (n) has the following addition value as shown in FIG. y (n) + a (n) = x (n) + x (nm) + sin (ω0 * t + θ) + sin (ω0 * t + θ + θ ′)

Therefore, in the adaptive filter 3 according to the second embodiment, the coefficient of the first sample a (0) is fixed to w0 = 1 as shown in FIGS. Based on the error signal e (n), each convolution coefficient to be multiplied is updated based on the error signal e (n) so that the phase of the operation result b (n) is delayed by an integral multiple of the period of the howling component included in the microphone input signal. And perform the convolution operation, and the operation result b
(n) is output to the adder 4. By fixing the coefficient of the first sample a (0) to w0 = 1, b (n) = x (nm), so that the output e ′ (n) of the subtracter 4 after adaptation is e ′ (n ) = Y (n) + a (n) −b (n) = x (n). Therefore, the echo component x (nm) is cancelled, and the howling component sin (ω0 * t + θ) + sin (ω0 * t +
θ + θ ′) is delayed by 2nπ, and this is
It can be subtracted from (ω0 * t + θ) + sin (ω0 * t + θ + θ ′).

FIG. 5 shows an original sound x (n), FIG. 6 shows an error signal e (n) when howling occurs, and FIG. 7 shows an error signal e '(n) after adaptation. Although the error signal e '(n) after the adaptation is not completely the same as the original sound x (n), it shows that howling can be suppressed as compared with the error signal e (n) at the time of howling. .

[0022]

As described above, according to the present invention, the phase of the convolution operation result is delayed by an integral multiple of the period of the howling component included in the microphone input signal so that the convolution operation result is obtained from the microphone input signal. Is subtracted, the howling component is subtracted from the microphone input signal including the original sound and the howling component, so that the howling can be reliably suppressed with a simple configuration. According to the present invention, the first convolution coefficient is fixed to 1 by the adaptive filter, and the phase of the convolution operation result is delayed by an integer multiple of the period of the howling component included in the microphone input signal. By updating each convolution coefficient after the second and subtracting the result of the convolution operation from the signal obtained by adding the microphone input signal and its delay signal,
Since the howling component and the delay signal of the microphone input signal are subtracted from the microphone input signal including the original sound and the howling component, howling can be reliably suppressed with a simple configuration.

[Brief description of the drawings]

FIG. 1 is a block diagram modeling a first embodiment of a howling suppression device according to the present invention.

FIG. 2 is a waveform chart for explaining a process of the howling suppression device of FIG. 1;

FIG. 3 is a block diagram modeling a howling suppression device according to a second embodiment.

FIG. 4 is a waveform chart for explaining the process of the howling suppression device of FIG. 3;

FIG. 5 is a waveform diagram showing an example of an original sound.

FIG. 6 is a waveform chart showing an error signal when howling occurs.

FIG. 7 is a waveform chart showing an error signal after adaptation.

[Explanation of symbols]

 Reference Signs List 1 adder 2 delay unit (delay means) 3 adaptive filter 3a convolution operation unit (convolution operation means) 3b coefficient updater (coefficient update means) 4 addition (subtraction) unit (error signal generation means)

Claims (2)

[Claims] A delay means for delaying a microphone input signal including an original sound and a howling component so that the original sound is not canceled; and sequentially delaying the signal delayed by the delay means for each sample, thereby forming a convolution coefficient in each sample. Multiplication, convolution operation means for performing convolution operation by adding each multiplication result, means for generating an error signal by subtracting the operation result of the convolution operation means from the microphone input signal, operation result of the convolution operation means And a coefficient updating means for updating each convolution coefficient based on the error signal such that a delay of an integral multiple of the period of the howling component included in the microphone input signal occurs in the phase of the howling suppression apparatus. 2. A delay means for delaying a microphone input signal including an original sound and a howling component so that the original sound is not canceled, and sequentially delaying the signal delayed by the delay means for each sample, thereby forming a convolution coefficient in each sample. A convolution operation means for performing a convolution operation by multiplying and adding each multiplication result, and adding the microphone input signal and the signal delayed by the delay means, and subtracting the operation result of the convolution operation means from the result to obtain an error signal. And a first convolution coefficient of the convolution operation means is fixed to 1, and the phase of the operation result of the convolution operation means is delayed by an integral multiple of the period of the howling component included in the microphone input signal. Coefficient updating means for updating each of the second and subsequent convolution coefficients based on the error signal so that Howling suppression device that.