JPS63108822A - Noise suppressor for voice speech signal - Google Patents

Abstract

PURPOSE:To suppress the influence of environmental noise, by dividing a signal component in frequencywise, and attenuating the component of a frequency band area with a large level and without having time fluctuation. CONSTITUTION:A signal inputted to a signal input terminal 1 is divided into the component by every frequency band area at a filter circuit 2, and is rectified at a fullwave rectifier circuit 3, and is smoothed at a smoothing circuit 4, then it is inputted to a signal delay circuit 5. And a subtractor 6 detects a difference between the signal on which delay is given at the delay circuit 5 and the signal before the delay is given, and outputs the signal corresponding to the time fluctuation. An output from the subtractor 6 is detected at the fullwave rectifier circuit 7, and the output is compared at a comparator 8, and is compared whether it is larger than a prescribed level E. Also, the component of each frequency band area is compared whether it is larger than a prescribed level E at a comparator 9. And a variable attenuator 11 is controlled by an AND output of the two comparators 8 and 9, then, attenuation by every frequency band area is performed.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention is used to reduce noise in voice communication lines. It is particularly suitable for suppressing the influence of environmental noise present in conference rooms and other places during remote conference calls.

[Conventional technology]

When holding a conference call using a communication line, environmental noise is a major cause of hindering communication. There are various types of environmental noise, but one that cannot be avoided even if a conference room is acoustically shielded from the outside is caused by air conditioning equipment such as heating and cooling. One cause of environmental noise, such as wind noise generated by air conditioning equipment, that impedes conversation is that it simply degrades the signal-to-noise ratio.

This is because when having a face-to-face conversation, the signal-to-noise ratio is often psychologically improved due to the selectivity and adaptability of the human sensory system, but on the other side via a communication line, the signal-to-noise ratio is often improved. Adaptability is limited, and the effects of noise are more pronounced than in face-to-face conversations.

This problem becomes even more serious when the spectral distributions of environmental noise in a plurality of conference rooms connected by communication lines are different from each other.

Another cause is the adverse effect of communication lines on howling prevention circuits used in conference systems. The howling prevention circuit, which operates with reference to the speech of conference participants, that is, the audio band signal level, may operate unstablely due to environmental noise.

Conventionally, as a means to reduce environmental noise that has such a negative impact, studies have been made to improve external conditions, such as the acoustic design of conference rooms and the design of air conditioning equipment.

As a technology related to the technology of the present invention, there is an audio switch for preventing howling. This detects the speech of participants in a video conference, that is, the audio signal level, and when there is a signal input above a predetermined threshold level, the two-way communication line, microphone, speaker, transmitter, etc. This method attenuates the gain of the closed loop formed by the receiving device by a certain amount. Howling can be prevented by appropriately setting the threshold level and attenuation amount. However, the signal-to-noise ratio of speech cannot be improved.

There is also a method of using an echo canceller to prevent howling. This is a two-way communication line that electrically cancels out the feedback of the received signal to the transmitting side, and there is little risk of problems such as loss of intelligibility at the beginning of speech or instability due to environmental noise, but Similar to the voice switch, it does not improve the signal-to-noise ratio.

[Problem that the invention seeks to solve]

Improving external conditions, such as improving the acoustic settings of conference rooms and air conditioning equipment, is effective and desirable, but it is not always possible to create an environment with less noise, and it is not possible to reduce noise by improving external conditions. There are also limits. Rather, we should provide a communication device that can make good calls even in poor environments.

The present invention provides a communication device that suppresses the effects of environmental noise from air conditioning equipment and other sources during calls using conference telephones and other communication lines, improves conversation communication, and does not destabilize the operation of the howlin prevention circuit. The purpose is to provide

[Means for solving problems]

The present invention provides a filter circuit that divides a human-powered audio band signal into a plurality of frequency wave number components, and a level changing means that individually changes the signal level of each frequency component signal that has passed through the filter circuit in accordance with a control signal. and a synthesis circuit for synthesizing the outputs of the level changing means, a circuit for determining levels corresponding to each of the frequency components, and a circuit for determining temporal fluctuations in the levels corresponding to each of the frequency components. and a control means for sending a control signal that causes the level changing means to suppress the amplitude of a frequency component whose level is determined to be greater than a predetermined value and whose temporal fluctuation is smaller than a predetermined value by the two determining circuits. It is characterized by having the following.

[For production]

Environmental noise caused by air conditioning equipment 4 Mainly, both in terms of level and frequency distribution, it is almost constant over time.

Therefore, the most important feature is that signal components are divided in frequency and captured, and components in frequency bands that have a large level and do not vary over time are attenuated.

All conventional howling prevention techniques are designed to prevent howling in bidirectional lines.

The present invention basically performs noise suppression on a unidirectional line. It is also possible to use the above conventional howling prevention circuit together.

〔Example〕

Embodiments of the present invention will be specifically described below with reference to the drawings.

FIG. 1 is a block diagram of a main part of an apparatus according to a first embodiment of the present invention. A filter circuit 2 divides a signal input to a signal input terminal 1 into components for each frequency band. The components for each frequency band are input to the double wave rectifier circuit 3. Here, each double-wave rectified signal is smoothed by a smoothing circuit 4 and input to a signal delay circuit 5. The input/output of the delay circuit 5 is provided with a subtracter 6 for detecting the difference between the delayed signal and the signal before being applied to obtain an output according to time fluctuations. The output of the subtracter 6 is detected by a double-wave rectifier circuit 7, and the output is input to a comparator 8, which compares whether the detected output is constant and greater than a bell E+. Based on this, it is determined whether or not there is a time variation. Additionally, a comparator 9 determines whether the components of each frequency band are greater than the threshold level E2.
Equipped with

The outputs of the two comparators 8 and 9 are input to an AND circuit 10. These circuits 6 to 10 are provided for each frequency band. The output of the AND circuit 10 is given to the control input of the variable attenuator 1) through which the components of each frequency band pass. The signals that have passed through the variable attenuator 1) for each frequency band are added together in an adder 12 for all bands, and sent to a signal output terminal 13. Subtractor 6 to variable attenuator 1
) are also provided for each frequency band, but are omitted from the diagram.

As an example of the filter circuit 2, it is configured as a bandpass filter bank with a center frequency set at one octave step. The entire filter bank is then made to cover the entire audio band.

FIG. 2 shows a characteristic diagram of one embodiment of the frequency characteristics of a filter bank. In this example, the center frequency of the band-pass filter in the lowest frequency band is 60 Hz, and the band-pass filter is composed of nine band-pass filters in octave steps.

The delay circuit 5 is constituted by a capacitor memory. This can also be configured using an analog memory such as COD.

With the above configuration, it is possible to determine the time fluctuation and the magnitude of energy for each divided frequency band. Therefore, components with small time fluctuations and continuous high levels are determined to be environmental noise, and the energy in the corresponding band is attenuated by the variable attenuator 1) to suppress the noise. In other words, since attenuation is applied only to the band in which the noise component is determined to be dominant among the energy of each divided subband, it is possible to effectively reduce noise without damaging the information to be transmitted. .

This also suppresses the component within the corresponding band of the audio signal, but since this component was originally masked by noise, even if this component is suppressed, it will not further deteriorate the naturalness of the conversation. On the contrary, the improved signal-to-noise ratio improves conversational communication.It has been confirmed that there is no malfunction when the present invention is used in combination with a voice switch for preventing howling.

FIG. 3 is a block diagram of an apparatus according to a second embodiment of the present invention.

This embodiment realizes a similar function using a digital circuit.

A signal input to the input terminal IN passes through one digital filter J and is sent to the output terminal OUT. The digital filter J includes a filter means for dividing an input audio band signal into a plurality of frequency components, and a level changing means for individually changing the signal level of each frequency component signal passed through the filter means according to a control signal. and,
Each function, including the synthesizing means for synthesizing the outputs of the level changing means, is constituted by one digital circuit.

Input signal IN is input to correlator A. This correlator A calculates autocorrelation at fixed time intervals (for example, every 5 seconds). The calculation result is input to the fast Fourier calculation circuit B. The fast Fourier arithmetic circuit B calculates absolute amplitude values for points (for example, 32 frequencies) among predetermined audio frequencies.

This calculation is performed every 5 seconds. The results of this calculation are stored in memory C and updated for each calculation. The result of this calculation is also given to the circuit, and the difference between it and the immediately previous calculation result is calculated. The calculated difference is compared with a predetermined threshold by circuit F.

Furthermore, the calculation result of the fast Fourier calculation circuit B is compared with a predetermined threshold for each frequency point by the circuit E. The result of this comparison is applied to the gate circuit G together with the output of the circuit F described above. This gate circuit G selects a frequency point with a large time fluctuation and a level smaller than a predetermined value. The result of this selection is given to the digital filter J with parameters set by the circuit H.

Circuits A to H of this second embodiment can be realized by one arithmetic circuit. Therefore, the circuit can be implemented as an LSI, and it can be made economically efficient.

〔Effect of the invention〕

As described above, according to the present invention, it is possible to remove noise components that do not need to be transmitted in a conference call or the like over a communication line, that is, components that do not vary over time and whose energy exceeds a predetermined level. It has the effect of suppressing the effects of noise and improving conversation transmission. This also makes it possible to relax the conditions regarding where the terminal equipment is installed.
Costs for this can be saved.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of main parts of a device according to a first embodiment of the present invention. FIG. 2 is a diagram showing an example of characteristics of a filter bank circuit. FIG. 3 is a configuration diagram of an apparatus according to a second embodiment of the present invention. DESCRIPTION OF SYMBOLS 1...Signal input terminal, 2...Band pass filter circuit that divides the input signal into frequency components, 3...Double wave rectifier circuit, 4...Smoothing circuit, 5...Delay circuit, 6・・・
・Subtractor, 7...Double wave rectifier circuit, 8.9...Comparator, 10...AND circuit, 1)...Variable attenuator, 12
...Adder, 13...Signal output terminal. Atsushi 2 figure

Claims (3)

[Claims] (1) a filter circuit that divides an input audio band signal into a plurality of frequency components; a level changing means that individually changes the signal level of each frequency component signal that has passed through the filter circuit according to a control signal; A device comprising: a synthesis circuit for synthesizing the outputs of the level changing means; , control means for sending a control signal to cause the level changing means to suppress the amplitude of a frequency component whose level is determined to be greater than a predetermined value and whose temporal fluctuation is smaller than a predetermined value by the two determining circuits; What is claimed is: 1. A noise suppression device for a voice call signal, comprising: (2) The noise suppression device for a voice communication signal according to claim (1), wherein the filter circuit, the level changing means, the two determining means, and the control means are analog circuits. (3) The filter circuit, the level changing means, and the combining means are constituted by one digital filter, and the two determining means and the control means are constituted by digital arithmetic circuits. Noise suppression device for voice call signals.