JPS634973B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a noise removal device that can be applied to, for example, a preprocessing device in a speech recognition device, an input device for a hearing aid, a beat interference removal device for a receiver, etc. be.

Configuration of conventional example and its problems In general, as shown in FIG. The analog control low-pass filter 2 is controlled in accordance with the result of the energy measuring means 5, and a signal from which noise components have been removed is extracted from the output terminal 3. This means that the noise output is proportional to the frequency bandwidth of the system, and the signal-to-noise ratio (hereinafter referred to as the SN ratio)
It is based on two basic theories: if the input signal level is high enough, the noise will be masked by the signal, and the noise will not be audible (this is called the masking effect).As the input signal level increases, the average energy measurement method 5, that is, the control signal of the analog-controlled low-pass filter 2, increases, and the cut-off frequency of the analog-controlled low-pass filter 2 increases, thereby attempting to reduce noise. However, the amount of noise that can be reduced by the above-mentioned noise removal device is limited, and there is a problem that, for example, when a narrow band sound such as a sine wave is input as noise, it cannot be removed. .

Purpose of the Invention The purpose of the present invention is to accurately remove noise components, regardless of the frequency characteristics of the noise superimposed on speech, if the noise is quasi-stationary. The object of the present invention is to provide a noise removal device capable of reproducing clear and easy-to-hear voices.

Configuration of the Invention The noise removal device of the present invention includes a signal system that mainly performs band division of an audio signal on which noise is superimposed;
Each band signal is transmitted for a certain period of time (frame) at a value equivalent to the SN ratio of that band (this is called a channel).
It has a control system that controls the amount of attenuation for each
In the signal system, the input signal is delayed by the processing time required by the control means by the delay means, the signal is divided into n channels by the band division means, and each band signal is divided by the attenuation means based on the amount of attenuation determined by the control means. The signals are attenuated and added by an adding means to obtain an output signal.Meanwhile, in the control system, the input signal is divided into bands by a band dividing means having the same characteristics as the band dividing means, and the average energy of the signals in each band is calculated. is measured by an average energy measuring means, and a control means calculates an estimated value of noise energy and an estimated value of audio energy for each frame and each channel from the average energy of each band, and based on these, calculates the amount of attenuation given to the attenuating means. It is configured to determine.

DESCRIPTION OF THE EMBODIMENT FIG. 2 shows an embodiment of the noise removal device of the present invention. In Figure 2, 11 is a signal system, 12
is a control system. In this signal system 11, an input signal is applied from an input terminal 100, and a delay means 2
Microprocessor 1400 described below by 00
is delayed by the processing time for calculating the attenuation amount g _i , and the band dividing means 30i (i=1, 2,...n)
The band is divided into n channels by the attenuation means 4.
0i (i=1,2,...n) according to the attenuation amount g _i , and the band dividing means 50i (i=1,2,...n)
…n) performs filtering and damping means 4.
0i reduces the amplitude discontinuity that occurs due to updating the attenuation amount g _i every frame period, the adding means 600 adds the signals of all channels, and outputs an output signal from the output terminal 700. On the other hand, control system 1
2, an input signal is applied from the input terminal 100, and the band dividing means 80i (i=1, 2,...n)
The band is divided into n channels, and the average energy of each channel is obtained by an average energy measuring means 90i (i=1, 2,...n), and then passed through a multiplexer 1000 to an analog-to-digital converter (hereinafter referred to as an AD converter). The average energy is converted to a digital quantity at 1100 (referred to as
400 and then stored in the memory 1500. And in the microprocessor 1400, the memory 150
Calculate the attenuation amount g _i based on the information stored in 0,
The attenuation g _i is the input/output interface 1300
and the damping means 4 via the demultiplexer 1200.
0i. A flowchart of the program of the microprocessor 1400 is shown in FIG.

In this embodiment, for example, the following specific characteristics and circuits are used. Band division means 30i, 50i
And 80i is 200~ to almost include the audio band.
It consists of a group of 15 channel bandpass filters with center frequencies set every 1/3 octave from 300Hz to 5 to 6KHz, and a BBD (bucket brigade device) is used for the delay means 200.
A delay time of 20m _sec was obtained, and the attenuation means 40i had a 3dB
Using a digitally controlled attenuator that can provide attenuation of up to -40 to 50 dB with step resolution,
A full-wave rectifier circuit group and a smoothing circuit group are used for the average energy measuring means 90i. In addition, the time length of one frame is set to 10 to 20 m _sec , the number of observation frames M is set to 50 to 200 in order to set the observation interval for detecting a silent interval to 1 to 2 seconds, and the number of frames N to be considered as a silent interval. 1 to 10.

Then, the amount of attenuation g _i given to the attenuation means 40i
In order to determine this, it is necessary to calculate an estimated value of noise energy and an estimated value of voice energy for each frame and each channel from the average energy of each band.

Next, a method of obtaining an estimated value of noise energy will be explained. Figure 4b shows the voice in normal conversation.
As shown in , there is always a silent section in which no voice is heard. When such a voice is superimposed with stationary noise as shown in Figure 2a, its energy will be observed as shown in Figure 4c, and the section with the lowest energy will be the silent section, and the In this case, it can be assumed that only noise exists. If a noise section can be detected in this way, the characteristics of the noise can be determined by analyzing that section. Specifically, the average energy for each frame in the entire frequency band of the input signal is determined, and by observation for a period of M frames (M is constant), the value is determined from the smallest to N frames (N is constant, N < M) is regarded as a silent period, and the average energy W _i,k for each band in these N frames is the average 1/N _N 〓 ^k=1 W _i,k (where i is the band number i=1, 2 , ... n, k are the ranks k = 1, 2, ... M) when the average energy in all frequency bands for each frame is arranged in ascending order,
The average value of the already obtained noise energy estimate N^ _i ' is set as the new noise energy estimate N^ _i as shown in Equation 5.

Note that 1/N _N 〓 ^k=1 W _i,k may be used as the estimated value of the noise energy N^ _i , but as shown in Equation 5, the estimated value of the noise energy that has already been obtained can be used as is.
By averaging with N _i ', discontinuities in the output signal caused by sudden changes in the noise energy estimate can be reduced.

In addition, there are two ways to move the observation interval of M frames in order to find W _i,k , but this is the method of moving M frames at a time as shown in Figure 5, and the method of moving one frame at a time as shown in Figure 6. There is a way to do this. In the method shown in FIG. 5, the estimated value N^ _i of the noise energy is constant during M frames, and is suitable when the superimposed noise changes relatively slowly.
In the method shown in the figure, the estimated value N^ _i of the noise energy changes every frame, so it is suitable when the superimposed noise changes relatively rapidly.

When the estimated value S^ _i of the speech energy is calculated from the estimated value N^ _i of the noise energy obtained as above, S^ _i = W _i −N^ _i W _i ＞N _i ...(6) Become. Furthermore, in the case of W _i ≦N _i , the estimated value of the voice energy takes a negative value, which is unreasonable, so it is set as S^ _i =0 W _i ≦N^ _i (7). However, W _i indicates the average energy for each band measured in an arbitrary frame.

Based on W _i and the estimated value S^ _i of the sound energy, the amount of attenuation g _i given to the attenuation means is set as g _i =S^ _i /W _i (8). In other words, if the input signal to the attenuation means is S _io and the output signal of the attenuation means is S _put , then this attenuation amount g _i
By applying to the attenuation means, S _put = g _i · S _io , and the energy of S _put is the estimated value of the audio energy.
is equal to S^ _i .

In this way, noise superimposed on speech can be suppressed by making the energy of each band signal equal to the estimated amount S _i of speech energy using attenuation means. On the other hand, depending on the noise, if g _i is simply applied to the attenuation means, the noise may not be suppressed perceptually, so as shown in equation 10, the nonlinear characteristic E(g _i ) can be expressed as the attenuation amount g _i multiplied by g _i ′=E(g _i )·g _i (10) By applying this new attenuation amount g _i ′ to the attenuation means, the noise suppression characteristics can be changed. Furthermore, this E
If a plurality of (g _i ) are prepared and E(g _i ) can be selected depending on the noise, it becomes possible to obtain the optimum noise suppression characteristic in terms of hearing. An example of this E(g _i )·g _i is shown in FIG. Characteristic A is the case where E(g _i )=1 and there is no nonlinear characteristic, and other characteristics B, C,
Characteristic D is a case where a nonlinear characteristic is provided.

The role of the band division means provided between the attenuation means and the addition means is because the amount of attenuation g _i changes with each frame period, so the output signal of the attenuation means is as shown in Figure 8a. The waveform becomes a tone burst, and a harsh sound is produced at amplitude discontinuities, so the output signal of the attenuation means is band-divided again using a new band-dividing means that has the same characteristics as the band-dividing means corresponding to each band. By doing so (this is called refiltering), as shown in FIG. 8b, it is possible to reduce the amplitude discontinuity that occurs from frame to frame and to reduce the harsh sound.

Effects of the Invention As detailed above, according to the present invention, an input signal is divided into bands, each band signal is added and combined via an attenuation means, and outputted, and each band signal obtained by band-dividing the input signal is output. Measure the average energy of
The structure is such that the attenuation means is controlled according to the attenuation amount determined based on the estimated values of the noise energy and voice energy calculated for each channel for each frame from the average energy of each band, so that the noise superimposition Regardless of the frequency characteristics of the noise component in the voice input information being used, if it is quasi-stationary, the noise component can be accurately removed in real time to create clear, easy-to-hear audio. It has the advantage of being able to output

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a conventional noise removal device, FIG. 2 is a block diagram showing an embodiment of the noise removal device of the present invention, FIG. 3 is a program flowchart of the microprocessor of the same device, and FIG. 4 Figure 5 shows the temporal changes in noise energy, voice energy, and voice + noise energy, and Figure 5 shows the case where the observation interval of M frames for detecting a non-speech interval is moved by M frames. Figure 6 shows the case where the observation interval of M frames for detecting a silent interval is moved one frame at a time, and Figure 7 shows the attenuation amount E(g _i )・g _i with non-linearity. FIG. 8 is a diagram showing a change in the output signal of the attenuation means due to refiltering. 100...Input terminal, 200...Delay means, 3
0i, 50i, 80i...band division means, 40i
... Attenuation means, 600 ... Addition means, 700 ...
Output terminal, 90i...Average energy measuring means, 1
000... Multiplexer, 1100... Analog-to-digital converter, 1200... Demultiplexer, 1300... Input/output interface, 1
400...Microprocessor, 1500...Memory.

Claims (1)

[Claims] 1. Delay means for delaying an input signal by an arbitrary amount of time; band division means for dividing the signal delayed by the delay means into a plurality of frequency bands; and each frequency band obtained by the band division means. Attenuation means for attenuating a signal by an arbitrary amount; addition means for adding up each band signal attenuated by the attenuation means to obtain a final output; and said band division means for dividing said input signal into a plurality of frequency bands. Based on a band dividing means having the same characteristics, an average energy measuring means for measuring the average energy of each band signal obtained by the band dividing means, and a measured value of the average energy of each band measured by the average energy measuring means. and control means for controlling the amount of attenuation of the attenuation means. 2. The control means measures the average energy from the input signal containing a mixture of voice and noise for each short period and for each band, and estimates the voice energy and noise energy based on the obtained average energy value. Noise removal according to claim 1, characterized in that the attenuation amount of the attenuation means is determined based on the calculated estimated value of voice energy and estimated value of noise energy. Device. 3 The amount of attenuation in the control means is determined by finding the average energy for each short time (this is called a frame) in the entire frequency band of the input signal, and then calculating the amount of attenuation from the smallest value in the observation of M frames (M is constant). Regarding N frames as a silent section where no voice exists, this N
Average energy for each band in frames
Average of W _i,k 1/N _N 〓 ^k=1 W _i,k (where i is the band number i=1, 2,...n, k is the average energy in all frequency bands for each frame arranged in descending order) The average value of the estimated value N^ _i of the noise energy that has already been obtained is determined as the new estimated value N^ _i of the noise energy as shown in the first equation, On the other hand, let W _i be the average energy for each band measured in an arbitrary frame, and calculate the estimated value S^ _i of the audio energy in that frame as shown in the second equation, S^ _i = W _i −N^ _i W _i ＞N _i =0 W _i ≦N _i ...(2) The attenuation amount gi of the attenuation means is g _i =S^ _i /W _i ...(3) As shown in the third equation, the average energy of each channel is Claim 2 characterized in that the calculation is performed from W _i and the estimated value S^ _i of the sound energy.
Noise removal device as described in section. 4 The attenuation amount by the control means is the attenuation amount g _i of the attenuation means
Let the new attenuation amount g _i ′ shown in the fourth equation obtained by multiplying by the nonlinear characteristic E(g _i ) be g _i ′=E(g _i )・g _i ……(4) and this nonlinear characteristic E( g _i ) from among several types, and the resulting attenuation amount g _i ' is applied to the attenuation means to adjust the degree of noise removal. A noise removal device according to claim 3. 5. Before inputting the band signals attenuated by the attenuation means to the addition means, each band signal is again band-divided using a new band division means having the same characteristics as the band division means corresponding to each band. 2. The noise removal device according to claim 1, wherein a new band signal obtained by the operation is input to said adding means.