JPS6318796A - Howling suppressing method - Google Patents

Abstract

PURPOSE:To correct an amplitude characteristic in a transmission function between a speaker and a microphone and to suitably suppress a howling by considering a frequency in which an increasing ratio higher than the average increasing ratio of all spectra is continued as a howling generating frequency and lowering the gain of the frequency. CONSTITUTION:The sound receiving signal of the microphone 10 is converted into the scale of the respective spectra Fi(i = 1-N) by an FFT circuit 11. In a howling detection circuit 12, the increasing ratio from the preceding spectrum of the respective spectra Fi is examined to obtain the average increasing ratio of all the spectra. Then, the increasing ratio of the respective spectra is compared with the average increasing ratio to detect the spectrum in which the average conciderably exceeds a certain level, if the state continues M times, it is decided that the howling is produced in the frequency and an equalizer is automatically adjusted so as to lower the gain in the frequency. Thereby, the change in the transfer function circulating from the speaker to the microphone can be followed and the howling can be suitably suppressed.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a howling suppression method used in a public address system or the like.

(Prior Art) FIG. 2 shows a block configuration of a device that implements a conventional howling suppression method. In FIG. 2), the microphone 20 is connected to an echo canceller 21, and the echo canceller 21 is connected to an amplifier 22. 2
3 is a white noise generator. 24 is a switch, which is connected to both the amplifier 22 and the white noise generator 23. 25 is a speaker, which is connected to both the echo canceller 21 and the switch 24.

Next, the operation of the above conventional example will be explained. In FIG. 2, the white noise generator 23 is activated by the switch 24.
When you select , white noise is output from the speaker 25, received by the microphone 20, and transmitted to the two-canceller 21.
is operated to estimate the transfer characteristics from the speaker 25 to the microphone 20 in advance.

Thereafter, by selecting the switch 24, the output signal of the i-multiplier 22 is outputted from the speaker.

In this way, howling can be suppressed even in the conventional method for suppressing howling by estimating the transfer characteristics from the speaker to the microphone in advance.

(Problem that the invention attempts to solve) However, in the conventional howling method described above.

Since the transfer characteristics from the speaker to the microphone are estimated in advance, there is a problem in that if the transfer characteristics change after the estimation, howling cannot be suppressed.

The present invention is intended to solve such conventional problems, and aims to provide an excellent howling suppression method that can adaptively suppress howling.

(Means for Solving the Problems) In order to achieve the above object, the present invention performs fast Fourier transform (FFT) on a sound signal received by a microphone, and uses the amplitude information of the spectrum obtained thereby to transform each spectrum. and the average increase rate of all spectra,
It is characterized in that it is assumed that howling occurs at a frequency that exhibits a larger rate of increase than the average rate of increase and continues in that state, and that the components of that frequency are reduced.

(Function) The present invention has the above-described configuration, and according to the present invention, the amplification η rate of the spectrum of the sound signal received by the microphone is checked, and the frequencies at which the increase rate continues to be larger than the average increase rate of the entire spectrum are determined. Detect. and.

By lowering the gain of that frequency, it has the effect of adaptively suppressing howling.

(Example) FIG. 1 shows an example of the configuration of an apparatus for carrying out the method of the present invention. This apparatus includes a microphone 10, an FFT circuit 11 that spectrally decomposes a sound signal received by the microphone 10, The system includes a howling detection circuit 12 that detects howling from each spectrum, an equalizer circuit 13 that reduces the gain of the frequency at which howling occurs, an amplifier 14, and a speaker 15.

Next, the operation of the above embodiment will be explained. In FIG. 1, the sound reception signal of the microphone IO is converted into
1 to convert the magnitude of each spectrum Fi (i=l to N). The howling detection circuit 12 examines the rate of increase of each spectrum Fi from the previous spectrum. Next, the average increase rate of all spectra is calculated. Next, the increase rate of each spectrum is compared with the average increase rate, and a spectrum whose average increase rate is much higher than a certain level is detected, and its state is If this occurs M times in a row, it is determined that howling is occurring at that frequency.

Automatically adjusts the equalizer to reduce the gain at that frequency. To explain this using Figure 3, Figure 3 (a
) to Fig. 3(b) or Fig. 3(c), do not consider this to be howling, but consider that the input signal has changed or the amplification factor of the amplifier has changed. However, when changing from Figure 3(a) to Figure 3(d'),
Since the third spectrum from the left in the figure shows an increase rate larger than the average increase rate, it is considered howling when this state occurs continuously.

According to the above embodiment, the spectrum of the sound signal received by the microphone is constantly observed, and in order to attenuate the spectrum in which the increase rate continues to be larger than the average increase rate of all spectra, the feedback from the speaker to the microphone is reduced. It can also follow changes in the transfer function and has the effect of adaptively suppressing howling.

(Effects of the Invention) The present invention regards a frequency where the increase rate continues to be larger than the average increase rate of the entire spectrum as the howling generation frequency, and lowers the gain of that frequency, thereby increasing the frequency between the speaker and the microphone. Howling can be adaptively suppressed by correcting the amplitude characteristics of the transfer function.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a howling suppression method according to an embodiment of the present invention, and FIG. 2 is a block diagram of a conventional howling suppression method. FIG. 3 shows the spectrum of the sound signal received by the microphone. lO...Microphone, 11...FFT circuit, 12
...Howling detection circuit, 13...Equalizer circuit, 14...Amplifier, 15...Speaker. 20...Microphone, 21...Echo canceller, 22...Amplifier, 23...White noise generator, 24...Switch, 25...Subey power. Patent applicant: Matsushita Electric Industrial Co., Ltd. Figure 1, No. 3 (C) (b) (d)

Claims (1)

[Claims] Fast Fourier transform is performed on the signal received by the microphone, and the increase rate of each spectrum and the average increase rate of all spectra are investigated using the amplitude information of the spectrum obtained. 1. A method for suppressing howling, which assumes that howling occurs at a frequency where the state is continuous, and reduces the component of that frequency.