JP3384540B2 - Receiving method, apparatus and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To sufficiently suppress occurrence of howling with a comparatively simple configuration and insignificant sound quality deterioration. SOLUTION: A microphone 1 is placed to a taker side and a microphone 2 is placed as a speaker driven by a received signal from a subscriber, where output channel signals of the microphones 1, 2 are split into a plurality of bands so that a major component of one frequency band results from one sound source signal (4). An inter-channel level difference/arrival time difference for each identical frequency band are detected and compared with respective threshold levels for each band, and it is determined whether the frequency band is a voice signal component of the talker or other signal component (601). Then the frequency band component of the output of the microphone 1 is selected only for the frequency band discriminated as the voice signal (602) and they are synthesized and the result is transmitted to the subscriber (7A).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention radiates a received signal from the ground as an acoustic signal with a loudspeaker or the like, and collects a voice signal of a speaker by a microphone and transmits the received signal to the ground. It converted acoustic signal is picked up in the microphone, for suppressing the howling occurs, that is, receiving method acoustic signal of the received signal is suppressed from flowing into the signal to be transmitted to the ground, the device, and the program Recording medium

[0002]

2. Description of the Related Art There are roughly three types of conventional methods that can suppress unwanted wraparound sound and suppress howling. The first method detects a frequency at which howling occurs and introduces a notch filter of that frequency into a transmission signal (speaker voice signal) or a reception signal. In this method, the sound quality is deteriorated because the band component with the notch filter is missing from the transmission signal.

The second method is to receive and reproduce a signal,
In this method, howling is suppressed by applying frequency modulation so that the frequency characteristics of a signal that is picked up and transmitted are different. The received and reproduced signal can be surely grasped as an electric signal. On the other hand, the picked-up signal is a mixture of a signal to be picked up and transmitted and a signal to be received and reproduced, and it is not possible to reliably grasp the signal to be picked up and sent. Therefore, even a signal to be collected and transmitted, which is not necessary for suppressing howling, is modulated, resulting in deterioration of sound quality.

A third method is a method of predicting a situation in which a received and reproduced signal is picked up and mixed in a transmitted signal by using an adaptive filter. Howling is suppressed by subtracting the component of the reproduction signal whose mixing is predicted to be picked up from the signal to be transmitted. However, the adaptive filter for prediction varies from time to time, and the prediction of the adaptive filter is possible only when there is no signal to be picked up and transmitted, that is, when there is only a reproduction signal. Similar to the second method, in many cases, the picked-up signals are mixed with the signals to be picked up and transmitted and the signals to be received and reproduced, and it is sure that there is no signal to be picked up and transmitted. It is necessary to do so.

[0005]

Therefore, the conventional technique has a problem that the deterioration of the sound quality is small and the howling cannot be suppressed.

[0006]

The sound collecting method of the present invention comprises:
By using a plurality of microphones provided apart from each other, each output channel signal of each microphone is divided into a plurality of frequency bands in a band division process, and each of the bands mainly has only one sound source signal component. , The parameters of the acoustic signal reaching the microphone, that is, the level (power) and the arrival time (phase), which change due to the positions of the plurality of microphones, for each of the same bands of the divided output channel signals. The difference in the value is detected as a parameter value difference between the band-by-band, using the parameter value difference between the band-by-band of each band, based on a preset threshold, the band-divided output channel signal Whether or not it is the voice signal component of the speaker is determined in the voice signal determination process, and based on the determination in the voice signal determination process, the above band division is performed. From the power channel signals, at least one voice signal input from the same speaker is selected in the voice signal selection process, and a plurality of band signals selected as signals from the same speaker in the voice signal selection process are voice signals. As a voice synthesis process, and the synthesized voice signal is transmitted to the ground.

According to the embodiment of the present invention, the received signal from the ground is divided into a plurality of bands which are so narrow that there is a band having a negligibly small level in one band and the band division. Detected level of the received signal, and for each of these divided bands,
If the detected level is less than or equal to a predetermined value, it is determined as a transmittable band in the transmittable band determination process, and only the band determined to be transmittable in the band signal selected in the voice signal selection process is transmittable selection process. Select with and send to the voice synthesis process.

The selection in the transmittable selection process may be performed by performing the determination in the voice signal determination process only for the band determined to be transmittable. According to another embodiment of the present invention, the received signal is divided into a plurality of frequency bands, and the band-divided received signal component corresponding to the band selected in the voice signal selection process is processed in the frequency component removal process. The band component of the remaining received signal that has been removed
The signal in the time domain is synthesized in the resynthesis process, and the synthesized signal is supplied to the electroacoustic conversion means.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows the basic configuration used in a receiver according to the present invention. In FIG. 1, a speaker 211 is provided as an electroacoustic conversion means in a room 210, and a transmission line 212 is provided.
Voice signal (received signal) of the other speaker sent via
Is reproduced by the speaker 211 and radiated into the room 210 as an acoustic signal. On the other hand, the voice signal uttered by the speaker 215 in the room 210 is received by the microphone 1 and transmitted as an electric signal to the other speaker side through the transmission line 216.
In this case, howling occurs when the voice signal emitted from the speaker 211 is captured by the microphone 1 and transmitted to the other speaker side.

Therefore, in this embodiment, the microphone 2 and the microphone 1 are provided side by side, for example, about 20 cm apart from each other, substantially parallel to the arrangement direction of the speaker 211 and the speaker 215, and the microphone 2 is on the speaker 211 side. These microphones 1 and 2 are connected to the sound collection processing unit 220. A specific example of the sound collection processing unit 220 will be described with reference to FIG. The output of the microphone 1 is called an L channel signal and the output of the microphone 2 is called an R channel signal. The L-channel signal and the R-channel signal are supplied to an inter-channel time difference / level difference detection unit 3 and a band division unit 4, which are each divided into a plurality of frequency band signals, and inter-channel time difference / level difference between bands. It is supplied to the detection unit 5 and the sound source determination signal selection unit 6. In accordance with each detection output of the detection units 3 and 5, the selection unit 6 selects any channel signal for each band as the voice component of the speaker or the acoustic component from the speaker, and the speaker for each selected band. The voice component signal is synthesized by the voice signal synthesizer 7A, and only the speaker voice signal is extracted.

Since the speaker 215 is closer to the microphone 1 than the microphone 2, the speaker's voice reaches the microphone 1 earlier than the microphone 2 and has a higher level, and the speaker 211 has the microphone 2 than the microphone 1.
, The acoustic signal from the speaker 211 reaches the microphone 2 earlier than the microphone 1, and the level is high. As described above, in the present invention, the change amount of the acoustic signal reaching the microphones 1 and 2 due to the positions of the speaker as the sound source and the speaker with respect to the microphones 1 and 2, in this example, the arrival time difference and the level difference between the signals are calculated. To use.

In the voice signal judging section 201, when the level difference and the arrival time difference are larger than 0 as compared with the threshold value for each band, for example, 0, it is judged that the component of the band is the speaker voice component, When it is smaller than 0, the component in that band is determined to be the speaker sound component. However, this is the case where the level and arrival time obtained from the output signal of the microphone 2 are subtracted from the level and arrival time obtained from the output signal of the microphone 1 by the difference detection unit 5.

Only for the band determined as the speaker's voice in this way, the voice signal selection unit 602 selects band components of the signal of the microphone 1, and the selected band voice components are selected by the voice signal synthesis unit 7A. It is converted into a time domain signal, that is, a synthesized voice signal, and transmitted to the transmission line 216. An example of a method of separating the speaker voice signal from the speaker sound signal and extracting the speaker voice signal will be specifically described below. In the following, speaker 21
5 and the speaker 211 are referred to as sound sources A and B, respectively. For example, when there are a plurality of speakers, each voice signal of these speakers is separated, and it is also applicable to the case where only one or a plurality of them are transmitted. Because.

As shown in FIG.
The signals from the two sound sources A and B are captured (S0
1). The inter-channel time difference / level difference detection unit 3 uses the L channel
Time difference between channel and R channel signal or
Detect the level difference. Parameters used to detect the time difference
Is the mutual phase of the L channel signal and the R channel signal.
The case of using the function will be described. As shown in FIG.
Each sample L of the L channel signal and the R channel signal
(T) and R (t) are read (S02), and these sample
A cross-correlation function between the two modules is calculated (S03). This calculation
Cross-correlation of both channel signals at the same sample time
For one channel signal and the other channel
If the signal is staggered by one sampling time, only two
Cross-correlation in each case when shifted ...
To find the cross-correlation function. Obtain a large number of these cross-correlations,
Create a histogram that normalizes these with power
(S04). Next, the histogram cumulative frequency ranking first
Time difference Δα ₁, Δα₂Seeking
(S05). These time difference Δα₁, Δα₂Is
The time difference between channels Δτ₁, Δτ₂Conversion to
And output (S06).

Δτ ₁ = 1000 × Δα ₁ / F (1) Δτ ₂ = 1000 × Δα ₂ / F (2) where F is a sampling frequency, and the factor of 1000 is to increase the value to some extent for convenience of calculation. Is. The time differences Δτ ₁ and Δτ ₂ are the time differences between the L channel signal and the R channel signal of the signals of the sound sources A and B, respectively.

Returning to the description of FIG. 2 and FIG.
The channel signal and the R channel signal are divided into signals L (f1), L (f2), ..., (Fn) and signals R (f1), R (f2) ,. S
04). This division is performed, for example, by performing discrete Fourier transform on each channel signal to convert it into a frequency domain signal and then dividing it into each frequency band. This band division is performed to the extent that only the signal component of one sound source is mainly present in each band due to the difference in the frequency characteristics of the signals of the sound sources A and B, and in the case of an audio signal, for example, it is divided into 20 Hz bandwidths. The power spectrum of the sound source A is obtained, for example, as shown in FIG. 5A, and the power spectrum of the sound source B is obtained, as shown in FIG. 5B, and each spectrum is divided by a bandwidth Δf that is separable. At this time, for example, the spectrum of one sound source can be ignored with respect to the spectrum of the other sound source, as indicated by the corresponding spectrum with a broken line. See also this figure
As can be seen from A and 5B, they may be separated by a bandwidth 2Δf. In other words, each band does not have to include only one spectrum. The discrete Fourier transform is performed every 20 to 40 ms, for example.

Next, the inter-channel time difference / level difference detecting section 5 for each band, for example, L (f1) and R (f1), ... L (f
n) and R (fn), the time difference or level difference between channels for each band is detected between the channels of the corresponding band signals (S05). Here, the inter-channel time difference between bands is uniquely detected by using the inter-channel time differences Δτ ₁ and Δτ ₂ detected by the inter-channel time difference detection unit 3. The formula used for this detection is as follows.

Δτ ₁ − {(Δφi / (2πfi) ) + (ki1 / fi)} = ε _i 1 (3) Δτ ₂ − {(Δφi / (2πfi) ) + (ki2 / fi)} = ε _i 2 (4) i = 1, 2, ..., N, Δφi are signal L (fi) and signal R
It is the phase difference of (fi). In these expressions, the integers ki1 and ki2 are determined so that ε _i 1 and ε _i 2 are minimized. next,
The smaller channel time difference Δτ _j (j = 1, 2) between the minimum values ε _i 1 and ε _i 2 is defined as the inter-channel time difference Δτ _{ij of the} band i. In other words, it is the time difference between channels in one band of one sound source signal.

The sound source determination signal selecting section 6 uses the band-by-band time differences Δτ _{1j to} τ _nj detected by the band-by-band time difference / level difference detecting section 5 to obtain each band signal L (f1).
.About.L (fn) and R (f1) to R (fn) corresponding to each other, the audio signal determination unit 6 determines which is selected.
01 (S06). For example, of the time differences Δτ ₁ and Δτ ₂ calculated by the inter-channel time difference / level difference detection unit 3, Δτ ₁ is the inter-channel time difference of the signal from the sound source A near the L-side microphone, and Δτ ₂ is , R near the microphone, the case where there is a time difference between the channels of the signal from the sound source B will be described.

In this case, the band i in which the time difference Δτ _ij calculated by the inter-channel time difference / level difference detection unit 5 is Δτ ₁ is determined by the audio signal determination unit 601 by the gate 602L.
i is opened and the input signal L (fi) on the L side remains S
Input signal R of band i on the R side, output as A (fi)
(Fi) indicates the gate 602R by the audio signal determination unit 601.
Is closed and SB (fi) is output as 0. On the contrary, the band i in which the time difference Δτ _ij becomes Δτ ₂ is the signal L on the L side.
(Fi) is output as SA (fi) = 0, and the input signal R (fi) is directly output as SB (fi) on the R side. That is, as shown in FIG. 1, band signals L (f1) -L
(Fn) is supplied to the sound source signal synthesis unit 7A through the gates 602L1 to 602Ln, respectively, and the band signal R (f
1) to R (fn) are gates 602R1 to 602, respectively.
It is supplied to the sound source signal synthesis unit 7 through Rn. In the audio signal determination unit 601 in the sound source determination signal selection unit 6, Δτ _1j ~ Δ
For the band i in which τ _nj is input and Δτ _ij is determined to be Δτ ₁ , gate control signals CLi = 1 and CRi = 0 are generated, the corresponding gate 602Li is controlled to open and 602Ri is controlled to close, and Δτ _ij For the band i in which is determined to be Δτ ₂ , the gate control signal CLi = 0 and CRi = 1
Is generated, the corresponding gate 602Li is closed, 602R
i is controlled to be open. The above description is a functional configuration, and is actually processed by, for example, a digital signal processor.

In the sound source signal synthesizing unit 7A, the signals SA (fi) to
SA (fn) is combined, and in the example of the band division, inverse Fourier transform is performed, and the result is output to the output terminal t _A as the signal SA, and the source signal combiner 7B outputs the signal SB (fi).
To SB (fn) is output to the output terminal t _B as similarly synthesized by the signal SB. As is clear from the above explanation,
In the device of the present invention, it is determined which sound source each band component is, which is obtained by finely band-dividing each channel signal, and outputs all the determined components, that is, the signals of the sound sources A and B. If the frequency components do not overlap each other, the processing is performed without losing a specific frequency band, so that the signals of the sound sources A and B are separated while maintaining the sound quality higher than the conventional method of extracting only the harmonic structure. It is possible.

The above description uses only the inter-channel time difference and the inter-band time difference between channels detected by the inter-channel time difference / level difference detection section 3 and the band-specific channel time difference / level difference detection section 5. The determination signal unit 601 determines the determination condition. Next, an embodiment in which the determination of the determination condition is processed using the level difference between channels will be described. In this embodiment, as shown in FIG. 6, the L channel signal and the R channel signal are taken in from the microphones 1 and 2 (S02), and the channel level difference ΔL between the L channel signal and the R channel signal is detected. The detection is performed by the unit 3 (FIG. 2) (S03). Similar to step S04 in FIG. 3, n channel-specific channel signals L (f1) to L (fn), R (f
1) to R (fn) (S04), and band-specific channel signals L (f1) to L (fn) and R (f1) to R (fn).
Corresponding band, that is, L (f1) and R (f1), L (f
2) and R (f2), ..., L (fn) and R (fn), level differences between channels ΔL1, ΔL2 ,.
n is detected (S05).

Human voice can be regarded as a steady state for about 20 ms to 40 ms. Therefore, in the audio signal determination unit 601 (FIG. 2), 20 ms to 40 m
For each s, the sign of the value obtained by taking the logarithm of the inter-channel level difference ΔL and the sign of the value taking the logarithm of the band-by-band inter-channel level difference ΔLi are the same code in a percentage of all bands or more. It is calculated whether it becomes (+ or-), and a predetermined value, for example, 8
If both have the same code in the bandwidth of more than 50% (S06, S0
7) From there, for 20 ms to 40 ms, it is judged only by the inter-channel level difference ΔL (S08), and if the band having the same code is 80% or less, it is 20 ms to 40 m from there.
During s, determination is made for each band using the level difference ΔLi between bands for each band (S09). When the entire band is determined by the channel level difference ΔL, if ΔL is positive, the L channel signal L (t) is output as it is as the signal SA, and the R channel signal R (t) is the signal. It is output as SB = 0. If ΔL is 0 or less, the L channel signal L (t) is output as the signal SA = 0, and the R channel signal R (t) is output as the signal SB as it is. However, this is the explanation when the value obtained by subtracting the R side from the L side is used as the level difference between channels. When the band-by-band inter-channel level difference ΔLi is used to determine each band, if the band-by-band inter-channel level difference ΔLi is positive for each band fi, the L-side division signal L (fi) is the signal SA as it is. Is output as (fi) and the R-side division signal R
(Fi) is output as the signal SB (fi) = 0. On the contrary, if the level difference ΔLi is 0 or less, the divided signal L is on the L side.
(Fi) is output as the signal SA (fi) = 0, and the divided signal R (fi) is output as the signal SB (fi) on the R side. As described above, the gate control signals CL1 to CLn and CR1 to CRn are output from the audio signal determination unit 601, and the gates 602L1 to 602Ln and 602R1 to 60 are output.
2Rn are controlled respectively. Similarly to the former case, this is also an explanation in the case where a value obtained by subtracting the R side from the L side is used as the level difference between channels for each band. Signals SA (f1) -SA
(Fn) and signals SB (f1) to SB (fn) are output to output terminals t _A and t _B as synthesized signals SA and SB, respectively, as in the previous embodiment (S10).

In the above-described embodiment, only one of the arrival time difference and the level difference is used as the judgment condition used in the audio signal judgment unit 601. However, if only the level difference is used, L (fi) and R are used in the low frequency band.
There is a case where the level is in conflict with (fi), and in that case, it becomes difficult to accurately obtain the level difference. Also, when only the time difference is used, in the high frequency band,
Since phase rotation occurs, it may be difficult to calculate the time difference correctly. From these points, it is better to use the time difference in the low frequency band and the level difference in the high frequency band for the determination.
It may be advantageous over using a single parameter over the entire band.

Therefore, an embodiment in which both the band-based channel time difference and the band-based channel level difference are used in the audio signal determination section 601 will be described with reference to the drawings starting from FIG. The block of the functional configuration of this embodiment is the same as that of FIG. 2, but the inter-channel time difference / level difference detection part 3, the inter-channel time difference / level difference detection part 5 and the audio signal determination part 6 are included.
The processing in 01 differs as follows. The inter-channel time difference / level difference detection unit 3 detects the detected time differences Δτ ₁ , Δτ.
If the average of the absolute values of ₂ or Δτ ₁ and Δτ ₂ are relatively close values, only one of them is output as one time difference Δτ. The inter-channel time differences Δτ ₁ , Δτ ₂ , and Δτ are calculated before the channel signals L (t) and R (t) are band-divided on the frequency axis, but may be calculated after the band division.

As shown in FIG. 7 , the L channel signal L
(T), the R channel signal R (t) is converted into a frame (for example, 2
Every 0 to 40 ms) (S02), band division unit 4
The L channel signal and the R channel signal are each divided into a plurality of frequency bands. In this example, the L channel signal L
(T), the R channel signal R (t) is subjected to a Hanning window (S03), and Fourier transform is applied to the divided signals L (f1) to L (fn) and R (f1) to R.
(Fn) is obtained (S04). Next, the frequency difference f of the divided signals is detected by the time difference / level difference detection section 5 for each band.
It is checked whether i is a band (hereinafter referred to as a low band) equal to or smaller than 1 / (2Δτ) (Δτ is a channel time difference) (S05). If i is equal to or smaller than the band, the inter-channel phase difference Δφi for each band is output (S5).
08), it is checked whether or not the frequency f of the divided signal is in a band (hereinafter, referred to as a middle band) larger than 1 / (2Δτ) and less than 1 / Δτ (S06). The inter-channel phase difference Δφi and the level difference ΔLi are output (S09), and it is checked whether the frequency f of the divided signals is 1 / Δτ or more (hereinafter referred to as high band) (S0).
7) If it is in the high frequency band, the level difference ΔLi between channels for each band is output (S10).

The voice signal determination unit 601 uses the band-by-band phase difference between channels and the level difference detected by the band-by-band channel time difference / level difference detection unit 5 to determine L (f1) to L (f).
n) and R (f1) to R (fn) are determined. For the phase difference Δφi and the level difference ΔL, in this example, the values calculated by subtracting the values on the R side from the L side are used.

The signals L (fi), R (f
Regarding i), as shown in FIG. 8, first, the phase difference Δφi is π.
It is checked whether or not (S15). If π or more, Δφi is 2
The value obtained by subtracting π is Δφi (S17), and step S1
If Δφi is not π or more in 5, check whether it is −π or less (S
16), if the following is satisfied, the value obtained by adding 2π to Δφi is Δφ
i (S18), and Δφi is used as it is (S19) unless it is −π or less in step S16. Step S1
7. Phase difference Δ between channels obtained by S18 and S19
φi is converted into a time difference Δσi by the following equation (S20).

Δσi = 1000Δφi / 2πfi (5) When the divided signals L (fi) and R (fi) are determined to be in the middle band, as shown in FIG. ) Is used to uniquely determine the phase difference Δφi. That is, it is checked whether ΔL (fi) is positive (S23), and if positive, it is checked whether the band-by-band phase difference Δφi between channels is positive (S24). If positive, the Δφi is output as it is (S26). ), If not positive in step S24, Δφi
The value obtained by adding 2π to is output as Δφi (S2
7). If ΔL (fi) is not positive in step S23,
It is checked whether the phase difference Δφi between channels for each band is negative (S25), and if negative, the Δφi is unchanged as Δφi.
Is output (S28), and if not negative in step S25, the value obtained by subtracting 2π from Δφi is output as Δφi (S29). The Δφi of any of these steps S26 to S29 is calculated by the following equation, and the time difference ΔΣi between channels for each band is calculated.
Is calculated as (S30).

Δσi = 1000 · Δφi / 2πfi (6) As described above, the band-based inter-channel time difference Δσi in the low band and the mid band and the band-based inter-channel level difference ΔL (fi) in the high band are obtained. The sound source signal is discriminated according to the following. As shown in FIG. 10, the phase difference Δφi is used in the low band and the middle band, and the level difference ΔLi is used in the high band to discriminate each frequency component of both channels as a signal of either sound source. Specifically, in the low band and the middle band, it is checked whether the band-by-band time difference Δσi obtained in FIGS. 8 and 9 is positive (S3).
4) If positive, L-side channel signal L of band i
(Fi) is output as the signal SA (fi), and the R-side band channel signal R (fi) is output as the signal SB (fi) of 0 (S36). If the band-based channel time difference Δσi is not positive in step S34, conversely, SA (fi) is set to 0.
And outputs R-side channel signal R (f
i) is output (S37).

Further, in the high frequency range, the level difference ΔL (f between channels for each band detected in step S10 in FIG.
It is checked whether i) is positive (S35), and if positive, the signal S
The L-side channel signal L (fi) is output as A (fi), and 0 is output as SB (fi) (S38). If the level difference ΔLi is not positive in step S35, SA (f
0 is output as i) and the R-side band channel signal R (fi) is output as SB (fi) (S39).

As described above, the L side or the R side is output for each band, and the respective frequency components discriminated by the sound source signal synthesis units 7A and 7B are added over the entire band (S4).
0) and inverse Fourier transform of each added signal (S4
1) and outputs the converted signals SA and SB (S4)
2). As described above, in this embodiment, by using the parameter advantageous for the sound source separation for each frequency band, the sound source separation having the higher separation performance is realized as compared with the case where the single parameter is used over the entire band. It is possible to

The present invention can be applied even if the number of sound sources is three or more.
Wear. As an example, the number of sound sources is 3 and the number of microphones is 2.
If the sound source is
The case of separating will be described. In this case, the time difference between channels /
In the level difference detection unit 3, L channel signals, R
When calculating the inter-channel time difference of the channel signals, FIG.
Histogram normalized by cross-correlation power as shown in
Lamb's cumulative frequency (peak value) from 1st to 3rd
For each source signal by finding each time point taken
Channel time difference Δτ₁, Δτ₂, Δτ ₃To calculate.
Then, the band time difference between channels / level difference detection unit 5
In addition, the time difference between channels in each band is Δτ₁Or
Et Δτ₃To decide which one. This decision is based on the above
This is the same as the calculation formulas (3) and (4) described in the embodiment. sound
In the voice signal determination unit 601, for example, Δτ₁> 0, Δτ
₂> 0, Δτ₃The case where <0 is described. Where Δ
τ₁, Δτ₂, Δτ₃Are sound sources A, B, and C, respectively.
Assuming the inter-channel time difference of the signal, these values are
It is assumed that the value is calculated by subtracting the value on the R side from the value on the L side. this
In this case, the sound source A is close to the microphone 1 on the L side, and the sound source B is
It is near the microphone 2 on the R side. Therefore, L channel
, The time difference between channels for each band is Δτ₁Becomes
The signals of the sound source A are added by adding the band signals, and Δτ₂Tona
, The sound source B signal is separated.
And are possible. In addition, from the R channel signal,
The time difference between channels is Δτ₃Signals in the band
By applying the force, the signal of the sound source C is separated.

In the above sound source separation, the speaker 215
And the speaker 211 is fixed, the time difference Δτ ₁ (or Δτ ₂ ) at which the acoustic signal from the speaker 215 (or the speaker 211) reaches the microphones 1 and 2 is constant, and it may be known in advance. Similarly, the level difference ΔL between channels can be known in advance. Therefore, the detection of the inter-channel time differences Δτ ₁ and Δτ ₂ in step S03 in FIG. 3 and the detection of the inter-channel level difference ΔL in step S03 in FIG. 6 can be omitted, and the inter-channel time difference / level in FIG. 2 can be omitted. The difference detector 3 can be omitted. Further, when using the inter-channel level difference ΔL (fi), steps S03, S06, S07, and S08 are omitted in FIG. 6, and the band-based inter-channel level difference is always used for each divided band. Sound sources may be separated. That is, it is not necessary to detect the level difference between channels. However, if p / n ≧ 0.8 is satisfied by performing the processing shown in FIG. 6, the processing becomes simple.

In the above description, the sound source signals are separated, and the separated sound source signals SA and SB are output separately. However, for example, one sound source A is the voice of the speaker and the other sound source B
In the case of noise, the present invention can be applied to suppress the noise by separating and extracting the signal sound of the sound source A mixed with the noise. When one sound source A, for example, a speaker has a wider frequency band than the other sound source B, that is, a speaker, and each frequency band is known in advance, as shown in FIG. , Separate the frequency bands where both sound source signals do not overlap. For example, the frequency band of the signal A (t) of the sound source A is f1 to fn, but the frequency band of the signal B (t) of the sound source B is f1 to fn (f
n> fm), the non-overlapping bands fm + 1 to fm
Is separated from the outputs of the microphones 1 and 2, and the signals in the bands fm + 1 to fn are separated by the audio signal determination unit 6
01, and in some cases, the band-by-band inter-channel time difference / level difference detection unit 5 is not performed, and the audio signal determination unit 601 selects the channel signal S selected as the signal of the sound source B.
Each of the R divided band channel signals R (fm + 1) to R (fn) selected as B (t) is SB (fm +).
1) to SB (fn), and SA (fm + 1) to
SA (fn) outputs the audio signal selection unit 6 so that 0 is output.
Control 02. That is, the gate 602Lm + 1 to 602L
n is normally closed, and gates 602Rm + 1 to 602Rn are normally open.

In the above description, the time difference Δσi between channels for each band
Is positive or negative, and the level difference between channels for each band Δ
Depending on whether Li is positive or negative, that is, with 0 as the threshold value, it was determined which microphone the band signal was closer to. This is the case where the sound source A and the sound source B are located symmetrically with respect to the bisector of the line connecting the microphones 1. If this relationship is not satisfied, the discrimination threshold may be determined as follows.

ΔL _{A is} the level difference between the channels for each band in which the signal of the sound source A reaches the microphone 1 and the microphone 2.
The arriving band-based channel time difference is Δτ _A , the band-based channel level difference that the signal of the sound source B reaches the microphone 1 and the microphone 2 is ΔL _B , and the arriving band-based channel time difference is Δτ _B. At this time, if the threshold value ΔLth of the level difference between channels by band is ΔLth = (ΔL _A + ΔLi) / 2, and the threshold Δτth of the time difference between channels by band is Δτth = (Δτ _A + Δτ _B ) / 2. Good. In the embodiment described above, ΔL _B = −Δ
When L _A , Δτ _B = −Δτ _A , ΔLth = 0, Δτth =
It becomes 0. In order to separate the sound sources A and B, the microphones 1 and 2 are positioned so that the two sound sources are on different sides with respect to the microphones 1 and 2, and the distance and direction to the microphones 1 and 2 are not always correct. In some cases, the thresholds ΔLth and Δτth may be variable, and ΔLth and Δτth may be adjustable so that separation is often performed.

FIG. 12 shows a further improvement of this howling suppression method. The branching unit 231 is inserted into the transmission line 212 from the other party connected to the speaker 211, the voice signal from the other party's speaker branched by this is delayed by the delaying unit 232 as necessary, and then the band dividing unit At 233, the frequency band is divided into a plurality of frequency bands. This division is equal to the number of divisions performed by the band division unit 4 and may be performed by a similar method. The component of each band obtained by dividing the band of the voice signal from the other party is analyzed by the transmittable band determination unit 234, and it is determined whether or not the frequency band of the component is a transmittable frequency band. That is, a band having no frequency component of the voice signal from the other side or a band having a sufficiently low level is determined as a transmittable band. Further, the division unit 4 divides the received signal from the other side into narrow bands so that a band in which the components of the received signal can be ignored is obtained in the divided band.

A transmittable component selection unit 235 is inserted between the audio signal selection unit 602L and the sound source signal synthesis unit 7A.
The voice signal selection unit 602L determines and selects the output signal S1 of the microphone 1 as the voice signal of the speaker 215,
Further, of these band components selected and selected, the transmissible component selection unit 235 selects only those band components determined by the transmissible band determination unit 234 as transmissible bands and sends them to the sound source signal synthesis unit 7A. Therefore, the frequency component that is emitted from the speaker 211 and may cause howling is not transmitted to the transmission line 216, so that howling can be suppressed more reliably. As the transmittable component selection unit 235, the audio signal determination unit 601 may determine only the band that is determined to be transmittable by the transmittable band determination unit 234 and may not transmit other bands.

The delay unit 232 determines the delay amount in consideration of the propagation time of the acoustic signal between the speaker 211 and the microphones 1 and 2. The branching unit 231 and the transmittable component selecting unit 2 are means for obtaining the delaying action performed by the delaying unit 232.
It may be inserted after any processing stage between 35 and 35. When inserting as shown by the dotted line frame 237 in the subsequent stage of the transmittable band determination unit 234, a readable / writable recording unit that stores data is used, and after a time corresponding to the required delay amount,
It is also possible to read out and supply to the transmittable component selection unit 235. In short, in consideration of the propagation time, a means for delaying the control of the transmittable component selection unit 235 may be provided. In some cases, these delay means can be omitted.

In the embodiment shown in FIG. 12, components that may cause howling are blocked on the transmitting side (output side), but they may be blocked on the receiving side (input side). The essential part of the embodiment is shown in FIG. The received signal from the transmission line 212 is a band division unit 241.
Is divided into multiple frequency bands. This division is the same as the division of the band division unit 4 (FIG. 2) and can be performed by the same method. The band-divided reception signal is input to the frequency component removing unit 242. The control signal, which is obtained from the voice signal determination unit 601, for selecting the voice component of the speaker 215 from the microphone 1 by the voice signal selection unit 602L is input to the frequency component removal unit 242, and the voice signal selection unit 602.
A band component that is not selected by L, that is, a band component that is not transmitted to the transmission line 216 is selected from the reception signals band-divided by the frequency component removal unit 242 and supplied to the acoustic signal synthesis unit 243.
The two are combined into an audio signal and supplied to the speaker 211. The acoustic signal synthesizer 243 has the same function as the sound source signal synthesizer 7A. According to this configuration, the speaker 2
Since the frequency component transmitted to the transmission line 216 is excluded from the acoustic signal emitted from 11, the occurrence of howling is suppressed.

As described in the embodiment of FIG. 2, the thresholds ΔLth and Δτth for determining which sound source signal the band component belongs to based on the time difference between the bands and the level difference between the bands are the sound sources. The preferred value varies depending on the relative position between the and microphones. Therefore, as shown in FIG. 12, a threshold value setting unit 251 is provided, and the audio signal determination unit 6
01, that is, thresholds ΔLth and Δτth
Is preferably changed and set according to the situation.

Further, in order to improve the noise resistance, a reference value setting unit 252 is provided, and a silencing standard for silencing frequency components having a level below a certain value is set and sent to the audio signal selection unit 602L. . As a result, in the audio signal selection unit 602L, among the frequency components of the sound pickup signal of the microphone 1 selected by the level difference threshold value and the phase difference (time difference) threshold value, the frequency component whose level is equal to or lower than the fixed value is dark. Noise components such as noise and air-conditioning noise are considered to be removed, and noise resistance is improved.

By the way, in order to prevent the occurrence of howling, a howling prevention standard for holding a frequency component having a level of a certain value or more at a certain value or less is added to the reference value setting unit 252, and the audio signal selection unit 602L is added. send. As a result,
In the audio signal selection unit 602L, the level is equal to or more than a certain value among the frequency components of the sound pickup signal of the microphone 1 selected by the level difference threshold value and the phase difference threshold value, or the above-mentioned silence standard added thereto. The frequency component is corrected to a level below the fixed value. This correction is performed by clipping to a certain value level instantaneously and occasionally when it becomes above a certain value level, and by compressing the dynamic range when it becomes relatively frequent above the certain value level. By doing so, it is possible to suppress an increase in the acoustic coupling amount that causes howling and prevent howling.

As shown in FIG. 13, a structure for suppressing reverberant sound can be added. That is, the wraparound signal estimator 26 that estimates the delayed wraparound signal is output to the output terminal t _A.
1 is connected to the estimated wraparound signal subtraction unit 262 that reduces the estimated delayed wraparound signal, and the wraparound signal estimation unit 261 is utilized by utilizing the property of the transfer characteristics of the direct sound and the reverberation sound.
The delayed wraparound signal is estimated and extracted. For this estimation processing, for example, the complex cepstrum method considering the minimum phase characteristic of the transfer characteristic is used. If necessary, the transfer characteristics of the direct sound and the reverberant sound can be measured by the impulse response method. The delayed wraparound signal estimated by the estimation unit 261 is removed by the wraparound signal removal unit 262 from the separated sound source signal (voice signal of the speaker 215) from the output terminal t _A and is sent to the transmission line 216. Regarding the suppression of the sneak signal by the sneak signal estimation unit 261 and the sneak signal removal unit 262, see, for example, literature, 1987.
It is shown in "Digital Signal Processing", translated by Gen Gen Date, published by Corona Publishing Co., Ltd. on November 25, 2013. The sneak signal estimation unit 261 and the sneak signal removal unit 262 can be processed by, for example, one DSP (digital signal processor).

When the speaker 215 moves only within a certain range, the microphone 1 installed on the speaker 215 side.
The level difference and the phase difference / arrival time difference between the frequency component of the sound collected by (1) and the frequency component of the sound collected by the microphone 2 installed on the side of the speaker 211 are limited within a certain range. . Therefore, the judgment reference range is set in the threshold value setting unit 251, signal processing is performed only for those within the phase difference range of the level difference range, and those outside the range are excluded from the processing target. In this way, the speech of the speaker 215 can be selected from the sound pickup signals of the microphone 1 with higher accuracy.

From the viewpoint different from the above case, since the speaker 211 is fixed, the frequency component of the voice of the speaker 211 picked up by the microphone 1 on the speaker 215 side and the speaker 211 side. The level difference, phase difference, or arrival time difference from the frequency component of the sound of the speaker 211 picked up by the microphone 2 is limited to a certain range.
The range of the level difference and the phase difference / arrival time difference is also a criterion for discarding by the audio signal selection unit 602L, and the threshold value is used as a criterion for the selection by the audio signal selection unit 602L based on these. It can also be set in the setting unit 251.

Even in this howling suppression, if three or more microphones are used, the function of selecting a necessary frequency component can be achieved with higher accuracy. Further, although the present invention is applied to the wraparound sound suppressing type sound collecting device of the sound system of the loudspeaking system, the present invention can also be applied to a general telephone transmitting / receiving device. Also, the audio signal selection unit 6
The frequency component to be selected in 02L is the microphone 1
It is not limited to a specific frequency component (voice of the speaker 215) in the frequency components of the voice signal collected in step S1, and depending on the situation, for example, when the speaker 215 has an outlet of an air conditioner, The frequency component determined to be the voice of the speaker 215 out of the frequency components picked up by the microphone 2 is selected, or the speaker 215 out of the frequency components picked up by both the microphones 1 and 2 in an environment with a large amount of noise. It is also possible to select the frequency component determined to be the voice.

It has been described above that the present invention can be applied to the case where a plurality of speakers are separated, and one or two voice signals are transmitted while suppressing the wraparound sound. In this case, the synthesized speech signals of a plurality of speakers are obtained separately from each other, but if the synthesized speech signals corresponding to the speakers who are not speaking are suppressed or cut off, the quality of the transmission speech signal is further improved. For this purpose, it is detected whether or not the speaker is speaking, but which sound source is not sounding is detected, and a suppression signal for the corresponding synthesized speech signal is created. A method of creating this suppression signal will be briefly described.

As shown in FIG. 14, the microphones M1,
M2 and M3 are arranged at the positions of the vertices of an equilateral triangle having a side of 20 cm, for example. A space is divided and set based on the directional characteristics of the microphones M1 to M3, and each divided space is called a sound source zone. All microphones M
When 1 to M3 are omnidirectional and have the same characteristic, they are divided into six zones Z1 to Z6 as shown in FIG. 12, for example. That is, each microphone M1, M2, M3
And six zones Z1 divided into six at equal angular intervals by the straight line passing through the center point Cp.
~ Z6 is formed. The sound source A is located in the zone Z3 and the sound source B is located in the zone Z4. That is, the microphones M1 to M3 are arranged so that one sound source belongs to one sound source zone.
Each sound source zone is determined based on the arrangement and characteristics of.

In FIG. 14, the band dividing unit 41 divides the acoustic signal S1 of the first channel picked up by the microphone M1 into n frequency band signals S1 (f1) to S1 (fn), and the dividing unit 42. The sound signal S2 of the second channel picked up by the microphone M2 is converted into n frequency band signals S2 (f
1) to S2 (fn), and the band division unit 43 divides the acoustic signal S3 of the third channel picked up by the microphone M3 into n frequency band signals S3 (f1) to S3 (fn). Each of these bands f1 to fn is divided into band dividing units 41 to 4
3 is common, and such band division can utilize a discrete Fourier transformer.

The sound source separation unit 80 separates the sound source signal by using the method described with reference to FIGS. However, in FIG. 14, since there are three microphones,
Similar processing is performed for each two combinations of the signals of the three channels. Therefore, the band dividing unit and the band dividing units 41 to 43 in the sound source separating unit 80 can be used together. The level (power) detection section S1 for each band detects the level (power) signals P (S1f1) to P (S1fn) of the signals S1 (f1) to S1 (fn) of each band obtained by the band division section 41, Similarly, the band-by-band level detection units 52 and 53 respectively obtain the band signals S2 (f) obtained by the band division units 42 and 43, respectively.
1) to S2 (fn), S3 (f1) to S3 (fn), P (S2f1) to P (S2fn), P (S3f1) to P
(S3fn) is detected. The level detection for each band can also be realized by the Fourier transformer. That is, each channel signal may be decomposed into spectra by discrete Fourier transform, and the power of each spectrum may be obtained. Therefore,
A power spectrum may be obtained for each channel signal and the power spectrum may be band-divided. Each channel signal of each of the microphones M1 to M3 is divided into each band by the band-specific level detection unit 400 and the level (power) is output.

On the other hand, the level (power) P (S1) of all frequency components of the sound signal S1 of the first channel picked up by the microphone M1 is detected by the all band level detecting section 61, and all band level detecting sections 62, 63 are detected. Each with a microphone M
2nd and 3rd channels 2, 2
Level P of all frequency components of each acoustic signal S2, S3 of 3
(S2) and P (S3) are detected.

The sound source state judging section 70 judges the sound source zone which does not emit sound by computer processing.
First, band-specific levels P (S1f1) to P (S1fn) and P (S2f1) obtained by the band-specific level detection unit 50.
~ P (S2fn), P (S3f1) ~ P (S3fn)
Are mutually compared for signals in the same band. Then, the channel of the highest level is specified for each of the bands f1 to fn.

By setting the number of band divisions n to be a predetermined number or more, it can be considered that one band includes only the sound signal of one sound source, as described above. fi levels P (S1fi), P (S
2fi) and P (S3fi) can be regarded as sound levels from the same sound source, and thus levels P (S1fi) and P (S2) in the same band for the first to third channels.
When there is a difference between fi) and P (S3fi), the level of the band of the channel of the microphone closest to the sound source becomes the highest.

As a result of the above process, the highest level channel is assigned to each of the bands f1 to fn. For each of the first to third channels in the n bands,
The total number χ1, χ2, χ3 of the band with the highest level is calculated. It can be considered that the microphone of the channel having the larger value of the total number is closer to the sound source. If the total numerical value is, for example, about 90n / 100 or more, it can be determined that the sound source is close to the microphone of the channel. However, the total number of bands with the highest level is 53n / 10.
When the value of 0 and the next largest total value is 49n / 100, it is not clear whether the sound source is close to each corresponding microphone. Therefore, the total number is a preset reference value ThP,
For example, when it exceeds about n / 3, it is determined that the sound source is closest to the microphone of the channel corresponding to the total number.

Further, the sound source state judging section 70 has a level P of each channel detected by the full band level detecting section 60.
If (S1) to P (S3) are also input and all the levels are equal to or lower than the preset reference value ThR, it is determined that there is no sound source in any zone. A control signal is generated based on the determination result by the sound source state determining unit 70, and the signal suppressing unit 90 suppresses the acoustic signals A and B divided by the sound source separating unit 80. That is, the control signal SAi
The acoustic signal SA is suppressed (attenuated or deleted) by the control signal SAi and the acoustic signal SB is suppressed by the control signal SBi.
Both acoustic signals SA and SB are suppressed by Bi. For example, normally closed switches 9A and 9B are provided in the signal suppressing unit 90,
The output terminals t _A and t _{B of the} sound source separation unit 80 are normally closed switches 9
The output terminals t _A ′ and t _B ′ are connected through A and 9B, the switch 9A is opened by the control signal SAi, the switch 9B is opened by the control signal SBi, and both the switches 9A and 9B are opened by the control signal SABi. Be opened.
As a matter of course, the same signal is used for the frame signal to be separated by the sound source separation unit 80 and the frame signal for obtaining the control signal used for suppression in the signal suppressing unit 90. The generation of the suppression (control) signals SAi, SBi, SABi will be described in an easy-to-understand manner.

Now, as shown in FIG. 15, when the sound sources A and B are located, the microphones M1 to M3 are arranged as shown in the figure, the zones Z1 to Z6 are determined, and the sound sources A and B are separated. It should be located in each of the zones Z3 and Z4.
At this time, the distances SA1, SA2 and SA3 of the sound source A to the microphones M1 to M3 are SA2 <SA3 <SA1. The distances SB1, SB2, SB3 of the sound source B to the microphones M1 to M3 are SB3 <SB2 <S.
It becomes B1.

The detection signal P (S
When all of 1) to P (S3) are smaller than the reference value ThR, it is considered that the sound sources A and B are not sounding, for example, not speaking, and both acoustic signals SA and SB are suppressed by the control signal SABi. At this time, the output acoustic signals SA and SB become silent signals (101 and 102 in FIG. 16). When only the sound source A is producing sound, the frequency components of all the bands of the acoustic signal reach the microphone M2 with the largest sound pressure level (power), so that the total number of bands χ2 of the channels of the microphone M2 is The most.

Further, when only the sound source B is sounding,
Since the frequency components of all the bands of the acoustic signal reach the microphone M3 with the highest sound pressure level, the total number of bands χ3 of the channels of this microphone M3 is the largest. Further, when the sound sources A and B are both sounding, the number of bands reached by the acoustic signal at the highest sound pressure level is balanced by the microphones M2 and M3.

Therefore, if the total number of bands that reach the microphone with the highest sound pressure level by the reference value ThP exceeds the reference value ThP, the sound source exists in the zone controlled by the microphone. By making a determination, the sound source zone that is producing a sound can be detected. In the above example, when only the sound source A is sounding, only χ2 exceeds the reference value ThP, and it is detected that the sounding sound source exists in the zone Z3 controlled by the microphone M2. The audio signal SB is suppressed by the control signal SBi, and only the acoustic signal SA is output (103 and 104 in FIG. 16).

Furthermore, sound sources A and B are both sounding,
When both χ2 and χ3 exceed the reference value ThP, it is possible to give priority to the sound source A, for example, and process only the sound source A is sounding. The processing procedure of FIG. 16 is as such. When both χ2 and χ3 have not reached the reference value ThP, it is determined that both sound sources A and B are sounding as long as the levels P (S1) to P (S3) exceed the reference value ThR. , The control signals SAi, SBi, and SABi are not output, and the voice suppression unit 90 outputs the combined signals SA and S.
B is not suppressed (107 in FIG. 16).

As described above, the sound source signals SA and SB separated by the sound source separation unit 80 correspond to the sound source determined by the sound source state determination unit 70 as not being sounded by the signal suppression unit 90. It is suppressed, and unnecessary sound comes to be suppressed. The generation of such a control signal can also be detected by utilizing the arrival time difference between bands. That is, in FIG. 14, the inter-band level difference detection unit 51 detects arrival time difference signals An (S1f1) to An (S1fn) instead of the level difference, and similarly, arrives time difference signals An (S2f1) to An.
(S2fn), An (S3f1) to An (S3fn) are detected, and the process of obtaining the arrival time difference signals is performed, for example, by calculating the phase (or group delay) of the signals in each band by Fourier transform and calculating the same band. signal S1 of fi (f
i), S2 (fi), S3 (fi) (i = 1, 2, ...,
By comparing the phases of n) with each other, it is possible to obtain a signal corresponding to the arrival time difference of the same sound source signal. In this case as well, the division by the band division unit 40 is made so small that only one sound source signal component is present in one band.

In this method of expressing the arrival time difference, for example, if one of the microphones M1 to M3 is set as a reference and the arrival time difference for the reference microphone is set to 0, the arrival time difference for the other microphones is set to the reference microphone. On the other hand, since it can be determined whether the arrival is fast or slow, it can be represented by a numerical value with a positive or negative polarity. In this case, assuming that the reference microphone is M1, for example, the arrival time difference signals An (S1f1) to An (S1
fn) is all 0.

In the sound source state judging section 70, the arrival time difference signals An (S1f1) to An (S1fn), An (S2f).
1) to An (S2fn), An (S3f1) to An (S
3fn) are compared with each other for signals in the same band. As a result, the channel through which the signal arrives fastest can be determined for each of the bands f1 to fn. Therefore, the total number of bands for which it is determined that the signal arrives fastest for each channel is calculated and compared between the channels. As a result, it can be considered that the microphone of the channel having the larger value of the total number of bands is closer to the sound source. Then, when the total number of bands for a channel exceeds a preset reference value ThP, it is determined that a sound source exists in a zone controlled by the microphone of the channel.

Now, suppose that the microphones M1 to M3 are arranged for the sound sources A and B as shown in FIG. The total number of bands for the channels of the microphone M1 is χ1, and the total number of bands for the channels of the microphones M2 and M3 is χ2, χ3, respectively. Also in this case, the processing procedure shown in FIG. 16 may be performed. That is, first, the detection signals P (S1) to P (S1)
When all of (S3) are smaller than the reference value ThR (1
01), it is considered that the sound sources A and B are not sounding, and the control signal SABi is generated (102) to generate both sound source signals SA and S.
Suppress B. At this time, the output signals SA 'and SB' are silent signals.

When only the sound source A is producing sound, the frequency components of all the bands of the sound source signal are microphone M2.
The total number of bands χ2 of the channels of this microphone M2 is the largest, since Further, when only the sound source B is sounding, the frequency components of all the bands of the sound source signal reach the microphone M3 fastest, so the total number of bands χ3 of the channels of this microphone M3.
Is the most.

Further, when the sound sources A and B are both sounding, the number of bands in which the sound source signal reaches the earliest is equal in the microphones M2 and M3. Therefore, when the total number of bands that reaches the microphone with the sound source signal fastest with the reference value ThP exceeds the set value ThP, the sound source exists in the zone controlled by the microphone, and the sound source emits sound. Determine that

In the above example, when only the sound source A is sounding, only χ2 exceeds the reference value ThP (1 in FIG. 16).
03), it is detected that the sound source generating the sound exists in the zone Z3 controlled by the microphone M2.
The control signal SBi is generated (104), the acoustic signal SB is suppressed, and only the signal SA is output. Further, when only the sound source B is producing sound, only χ3 exceeds the reference value ThP (105), and it is detected that the sound source producing sound exists in the zone Z4 controlled by the microphone M3. , The control signal SAi is generated (106), the signal SA is suppressed, and only the signal SB is output.

In this example, ThP is set to, for example, about n / 3, both sound sources A and B are sounding, and both χ2 and χ3 may exceed the reference value ThP. In this case,
As shown in the processing procedure of No. 3, one sound source, A in this example, can be prioritized and only the separated signal can be output to the sound source A. If both χ2 and χ3 do not reach the reference value ThP, the levels P (S1) to P (S3) are set to the reference value T.
As long as hR is exceeded, both sound sources A and B are determined to be sounding, and control signals SAi, SBi, and SABi are not output (107 in FIG. 16).
A and SB are not suppressed.

A functional block diagram of an example in which the method of suppressing or muting a synthetic sound signal which is not sounded is applied to the suppression sound collecting apparatus as described above corresponds to FIG. 17 and corresponds to FIG. 2, FIG. 12 and FIG. The same symbols are given to the parts to be shown. In other words, in this case, the channel signals from the microphones 1 and 2 are divided into a plurality of bands by the band division unit 4, and the audio signal selection unit 602L, the band time difference between channels / level difference detection unit 5, the band level / time difference. It is supplied to the detection unit 50. The outputs of both microphones 1 and 2 are also supplied to the inter-channel time difference / level difference detection section 3, and the inter-channel time difference or level difference is supplied to the band-based inter-channel time difference / level difference detection section 5 and the audio signal determination section 601. The levels of the outputs of the microphones 1 and 2 are supplied to the sound source state determination unit 70.

The output of the time difference / level difference detecting unit 5 for each band for each band is supplied to the audio signal judging unit 601, and as described above, which sound source component is judged for each band.
Based on this determination result, the sound signal selection unit 602L selects only the sound signal component of a specific sound source, in this example, the sound component of one speaker, and supplies it to the sound source signal synthesis unit 7. On the other hand, the band-by-band level / time difference detection unit 50 detects the level of each band or the arrival time difference, and these detection outputs are detected by the sound source state determination unit 70 as described above. The signal suppressing unit 90 suppresses the synthesized sound source signal that is not sounded.

Similarly, the method of suppressing the synthesized sound source signal which is not sounding can be applied to the wraparound suppression sound collecting device shown in FIGS. 13 and 14. The band division unit 4 in FIG.
Each band division unit 40 in FIG. 14 and band division unit 2 in FIG.
33, the division of each frequency band in the band division unit 241 in FIG. 13 does not necessarily have to be the same. These division numbers may be different from each other depending on the required accuracy. The band division unit 4 when using the level difference between bands in FIG. 2, the band division unit 233 in FIG. 12, and the band division unit 40 in FIG. 14 respectively obtain the power spectrum of the input signal for subsequent processing. It may be obtained first and then divided into a plurality of frequency bands.

[0074]

As described above, according to the present invention,
The output signals of multiple microphones are divided into multiple sufficiently narrow bands, the parameter values of the acoustic signal for each band are detected, and the differences between these bands are detected, and the parameter value difference is determined by the threshold value. Compared with the value, the voice signal of the speaker can be correctly separated from other acoustic signals, and the occurrence of howling can be sufficiently suppressed with a relatively simple configuration. Moreover, there is little deterioration of sound.

Further, the received signal is divided into a plurality of bands which are sufficiently narrow so that there is a band whose level (power) in the band is sufficiently negligible, and this signal is separated and extracted only in the negligible band. Component is extracted and then subjected to voice synthesis, or a band-divided reception signal is subjected to voice synthesis to remove a component in a band to be transmitted, and the removed divided band reception signal is subjected to voice synthesis to perform an electroacoustic transducer. By supplying to, it is possible to further reliably suppress the occurrence of howling.

[Brief description of drawings]

FIG. 1 is a block diagram showing the main configuration of the device of the present invention.

FIG. 2 is a block diagram showing a functional configuration of an embodiment of a sound source separation section used in the present invention.

FIG. 3 is a flowchart showing a processing procedure of an embodiment of a sound source separation method used in the present invention.

FIG. 4 is a flowchart showing an example of a processing procedure for obtaining time differences Δτ ₁ and Δτ ₂ between channels in FIG.

5A and 5B are diagrams showing examples of spectra of two sound source signals, respectively.

FIG. 6 is a flowchart showing a processing procedure of an embodiment in which sound source separation is performed by using a level difference between channels in the sound source separation method.

FIG. 7 is a flowchart showing a part of a processing procedure of an embodiment that uses a level difference between channels and a time difference between arrivals between channels in a sound source separation method.

FIG. 8 is a flowchart showing a continuation of step S08 in FIG. 7.

9 is a flowchart showing a continuation of step S09 in FIG. 7.

10 is a flowchart showing a continuation of step S10 in FIG. 7 and steps S20 and S30 in FIG. 7 and FIG.

FIG. 11 is a block diagram showing a functional configuration of an embodiment for separating sound source signals having different frequency bands.

FIG. 12 is a block diagram showing a functional configuration of an embodiment of a receiver according to the invention.

FIG. 13 is a block diagram showing a part of the functional configuration of another embodiment.

FIG. 14 is a block diagram showing a functional configuration of an embodiment of a sound source separation unit to which a configuration for suppressing unnecessary sound source signals by using a level difference is added.

FIG. 15 is a diagram showing an arrangement example of three microphones, zones which are responsible for the microphones, and two sound sources.

FIG. 16 is a flow chart showing an example of a sound source zone detection and a suppression control signal generation processing procedure in the case where there is one sound source that is sounding.

FIG. 17 is a block diagram showing a functional configuration of still another embodiment of the present invention.

─────────────────────────────────────────────────── ─── Continuation of the front page (72) Hiroyuki Matsui Inventor Hiroyuki Matsui 3-19-2 Nishishinjuku, Shinjuku-ku, Tokyo Nihon Telegraph and Telephone Corporation (56) Reference JP-A-8-84392 (JP, A) Kaihei 6-292292 (JP, A) JP 5-199590 (JP, A) JP 59-161995 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) H04R 3 / 04 G10K 15/00 G10L 13/00 H04S 7/00

Claims (10)

(57) [Claims] 1. Output channel signals of a plurality of microphones separated from each other are divided into a plurality of frequency bands, respectively.
And, of these divider has been each channel signal for each band, a band specific parameter value detection step of detecting Rupa parameter value will change due to the position of the plurality of microphones, the detection for each same band A parameter value difference detection process for detecting the difference between the channel values of the detected parameter values, and using the detected parameter value differences, based on a preset threshold value , a sound signal is generated from the band-divided channel signal . A voice signal selection process of selecting the voice signal component in the band unit, a voice synthesis process of synthesizing the voice signal component of the selected band into a voice signal over the entire band, and a plurality of the channel signals described above. Divide into bands and divide these
Multiple microphones for each channel signal for each specified band
Detects parameter values that change due to the position of the Rohon
Second band-specific parameter value detection process and each band detected in the second band-specific parameter value detection process
Comparing parameter values by region between channels for the same band
Based on the result of the detection, a speaker who is not speaking is detected
The sound source state determination process and the speaker test that does not speak in the sound source state determination process
The voice signal synthesized in the voice synthesis process by the output signal
Of the above, the synthesized signal corresponding to the speaker not speaking
Of the speech signal synthesized in the speech synthesis step and the signal suppression step for suppressing
Receiving an uncompressed synthetic speech signal
Method. 2. Each output of two microphones separated from each other.
Obtaining the cross-correlation of the force channel signals L (f) and R (f)
Acoustic signals from each sound source A and B reach each microphone
Time difference (called time difference between channels) Δτ ₁ , Δ
The process of obtaining τ ₂ and the channel signals L (f) and R (f) are discretely divided.
Fourier transform, and transform each frequency band signal into
Bandwidth of the extent that only the signal component of the square of the sound source is present mainly
Divide by width to obtain band signals L (fi) and R (fi)
Process, i = 1, ..., N, n is the number of divided bands, and both signals L (f
i), R (fi) phase difference Δφ _i , and frequency fi
_{Te, Δτ 1 - {(Δφi /} (2πfi)) + (ki1 / fi)} = ε i 1 Δτ 2 - {(Δφi / (2πfi)) + (ki2 / fi)} = ε i 2 ε i 1, Integers ki1 and ki2 are set so that ε _i 2 is minimized.
Between the channels with the smaller minimum values ε _i 1 and ε _i 2
The time difference Δτ _j (j = 1, 2) is calculated as the channel of the band i.
The process of setting the inter-time difference Δτ _ij and the signal L (fi) of the band i in which the time difference Δτ _ij is Δτ ₁
The signal R (fi) is selected as a voice signal component in band units.
And the audio signal component L of the selected band i in the above bands 1 to n.
(Fi) is synthesized and inverse Fourier transformed to produce a voice signal
A process of synthesizing a receiving method and a step of outputting the synthesized speech signal. 3. Each output of a plurality of microphones separated from each other.
Level difference between input channel signals ΔL (level difference between channels
, And each output channel signal is divided into n frequency bands.
The process of dividing into different channel signals, and between the different channel signals for each divided same band
Level difference (called level difference for each band) ΔLi (i = 1,
, N, where n is the number of divided bands), the sign (+ or −) of the inter-channel level difference ΔL,
Find the number of bands that match the sign of the level difference ΔLi for each band.
And the process of determining whether the number of matching bands is greater than or equal to a predetermined value
And the level difference between channels ΔL
Is positive, the corresponding one channel signal
And the corresponding one channel if the above judgment is not more than a predetermined value.
All of the above-mentioned band-specific level differences ΔLi in the signal
Of the band-specific channel signal of
And the sound signal component of these selected bands over the entire band.
Receiving with the process of synthesizing the voice signal, and a step of outputting the synthesized speech signal
Method. 4. Each output of a plurality of microphones separated from each other.
The microphone of the acoustic signal from the sound source
Time difference to reach the channel (called time difference between channels)
And the step of detecting the output channel signal is divided into a plurality of frequency bands,
Time of each output channel signal for each of these divided bands
Difference (called channel time difference between channels) and level difference (bandwidth)
The process of obtaining the level difference between different channels)
And the divided time based on the time difference between the channels.
Output channel signal with three frequencies, low, mid and high
The process of dividing into regions and the inter-channel time for each band in the above low frequency region
The difference is used to set the band based on a preset threshold.
Bandwidth of audio signal component from band-divided channel signal
In the frequency range of the above mid-range,
It is judged whether the bell difference is positive and based on the judgment result.
Calculate the time difference between the channels for each band
Threshold preset using time difference between channels for each band
Depending on the value, the audio from that band-split channel signal
The signal components selected in band units, the frequency range of the high band, Les between the respective band-by-band channel
Bell difference is used based on a preset threshold.
The audio signal component from the band-divided channel signal of
The process of selecting in units and the audio signal components of these selected bands over the entire band
Receiving method comprising the steps of synthesizing a speech signal, and a step of outputting the synthesized speech signal. 5. The method according to claim 1, 3 or 4.
Method, dividing each output channel signal into multiple frequency bands
Is mainly the signal component of one sound source.
Receiving method characterized by having a band as narrow as possible
Law. 6. Each output of two microphones separated from each other.
Obtaining the cross-correlation of the force channel signals L (f) and R (f)
Acoustic signals from each sound source A and B reach each microphone
Time difference (called time difference between channels) Δτ ₁ , Δ
The inter-channel time difference detecting means for obtaining τ ₂ and the channel signals L (f) and R (f) are respectively separated.
Fourier transform, and transform each frequency band signal into
Band in which only the signal component of one sound source mainly exists
Divide by width to obtain band signals L (fi) and R (fi)
Band dividing means, i = 1, ..., N, n is the number of divided bands, and both signals L (f
i), R (fi) phase difference Δφ _i , and frequency fi
_{Te, Δτ 1 - {(Δφi /} (2πfi)) + (ki1 / fi)} = ε i 1 Δτ 2 - {(Δφi / (2πfi)) + (ki2 / fi)} = ε i 2 ε i 1, Integers ki1 and ki2 are set so that ε _i 2 is minimized.
Between the channels with the smaller minimum values ε _i 1 and ε _i 2
The time difference Δτ _j (j = 1, 2) is calculated as the channel of the band i.
Inter- channel time difference detection means for each band with inter-time difference Δτ _ij
And a signal L (fi) in the band i having a time difference Δτ _ij of Δτ ₁
The signal R (fi) is selected as a voice signal component in band units.
And a voice signal selecting means for blocking the voice signal component L of the band i selected from the above bands 1 to n.
(Fi) is synthesized and inverse Fourier transformed to produce a voice signal
Synthesizer for synthesizing speech and outputting the synthesized speech signal
A receiving device having means . 7. Each output of a plurality of microphones separated from each other.
Level difference between input channel signals ΔL (level difference between channels
Inter-channel level difference detection means for obtaining each of the output channel signals and n frequency bands for each of the output channel signals.
Band splitting means for splitting into separate channel signals, and between the separate channel signals for each of the same split bands
Level difference (called per-band level difference) ΔLi (i = 1,
, N, where n is the number of divided bands)
Output means, the sign (+ or −) of the level difference ΔL between the channels, and
Find the number of bands that match the sign of the level difference ΔLi for each band.
Means for determining whether or not the number of matching bands is greater than or equal to a predetermined value
The determining means and the inter-channel level difference ΔL if the above determination is a predetermined value or more
Is positive, the corresponding one channel signal
Means for outputting as a single channel
All of the above-mentioned band-specific level differences ΔLi in the signal
Sound that selects the channel signal for each band as the audio signal component
The voice signal selection means and the sound signal components of these selected bands
Receiving with a voice synthesizing means for synthesizing the voice signal, and means for outputting the synthesized speech signal
apparatus. 8. Each output of a plurality of microphones separated from each other.
The microphone of the acoustic signal from the sound source
Time difference to reach the channel (called time difference between channels)
Inter-channel time difference detection means for detecting, and dividing each output channel signal into a plurality of frequency bands,
Time of each output channel signal for each of these divided bands
Difference (called channel time difference between channels) and level difference (bandwidth)
For each band to obtain the level difference between different channels)
Inter-channel level difference detection means and each of the divided signals based on the inter-channel time difference.
Output channel signal with three frequencies, low, mid and high
The frequency domain dividing means for dividing into regions and the inter-channel time for each band in the above low frequency region
The difference is used to set the band based on a preset threshold.
Bandwidth of audio signal component from band-divided channel signal
In the frequency range of the above mid-range,
It is judged whether the bell difference is positive and based on the judgment result.
Calculate the time difference between the channels for each band
Threshold preset using time difference between channels for each band
Based on the value, the sound from the band-divided channel signals of their
The signal components selected in band units, the frequency range of the high band, Les between the respective band-by-band channel
Bell difference is used based on a preset threshold.
The audio signal component from the band-divided channel signal of
An audio signal selection means that selects in units and the audio signal components of these selected bands over the entire band
A receiving device having a voice synthesizing unit for synthesizing a voice signal and a unit for outputting the synthesized voice signal . 9. A device according to claim 1, 3 or 4.
, The band component of each output channel signal into multiple frequency bands.
In the splitting means, the signal of each band is mainly the signal of one sound source.
Being selected in a narrow band so that only the components are included
Characteristic receiver device. 10. The receiver according to any one of claims 1 to 5.
A professional for having a computer execute each step of the speaking method.
Computer-readable recording medium that records grams
body.