JP2018142822A - Acoustic signal processing device, method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make it possible to determine (classify) which signal zone among target sound, disturbing sound and background noise an input signal belongs to, with a low processing load and good accuracy in real time.SOLUTION: An acoustic signal processing device according to the present invention, comprises: a front suppression signal generation unit which generates front suppression signals with a blind spot in front thereof based on differences among a plurality of frequency domain input signals resulting from conversion of input signals from a plurality of microphones from a time domain to a frequency domain; a coherence calculating unit which calculates a coherence based on signals obtained from the plurality of input signals; and a classification unit which classifies signal zones of the input signals based on a feature quantity value indicative of the relationship between the front suppression signal and the coherence, and a variable obtained in the course of calculation of the feature quantity.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an acoustic signal processing device, method, and program, and can be applied to, for example, a communication device or communication software that uses voice, such as a telephone or a video conference system, or acoustic signal processing that is used in preprocessing of voice recognition processing. is there.

  In recent years, voice assistants are rapidly spreading due to rapid performance improvement of voice recognition technology. The form of the voice assistant is not limited to a smartphone application, and for example, there are an increasing number of cases where it is spread as a dedicated terminal (speaker having a voice recognition function) such as Amazon Echo (registered trademark).

  Needless to say, speech recognition is an important technology for realizing voice assistants regardless of the form, but especially in the case of dedicated terminals, the positional relationship between the speaker and the terminal varies. One feature is that voice recognition processing must be performed. This is because, prior to speech recognition, pre-processing such as sound source localization technology such as the MUSIC method is used, and target sound enhancement or noise suppression is performed using beam former technology using the estimated orientation information. This is necessary. In order to realize such preprocessing, a sound source localization technique that can accurately estimate the sound source direction even under noise is indispensable. Moreover, in order to increase the user's satisfaction with the voice assistant, it is important that the time from when the user speaks until the response from the assistant is obtained (response time) is short. Therefore, at the same time as improving the accuracy of sound source localization, it is also required that the sound source localization processing is lightweight and real-time performance is guaranteed.

JP 2013-182044 A JP2015-189919A

Asano Tadashi, Acoustical Society of Japan “Sound Technology Series 16 Sound Array Signal Processing Sound Source Localization / Tracking and Separation”, Corona Publishing Co., Ltd., February 25, 2011, p. 115

  A typical method as a sound source localization technique is the MUSIC method. The MUSIC method estimates the peak of the spatial spectrum based on the property that the signal subspace and the noise subspace are orthogonal in the eigenspace spanned by the eigenvectors of the correlation matrix between the observation signals obtained from multiple microphones. (See Non-Patent Document 1).

  The MUSIC method is useful in that a higher spatial resolution can be obtained than a sound source localization technique using a conventional beam former. However, since the accuracy of sound source localization is a problem under noise, the noise correlation matrix is calculated in the pause period of the target sound, and eigenvalue decomposition is used to reduce the influence of noise. Measures are taken.

  However, noise is broadly divided into background noise (spatial white) and interference sound (for example, speech other than the intended speaker, spatially colored). The influence on the estimation accuracy of the MUSIC method is greatly changed between the case of being colored and the case of being colored. In particular, when the noise is colored, the signal subspace and the noise subspace are less likely to be orthogonal, so noise whitening is required, and control according to the spatial color of the noise is required. is there. Further, from the viewpoint of obtaining the characteristics of the background noise correctly, the target sound component becomes a disturbance, so it is necessary to distinguish the background noise from the target sound. For this reason, it is not sufficient to detect the pause period of the target sound as described above, and the input signal is classified according to whether it belongs to the target sound period, the interference sound period, or the background noise period. There is a need for an input signal classification method that can be used.

  One conventional method for classifying signal sections is Patent Document 1. This is a method of detecting a target sound section based on the magnitude of a feature quantity called coherence that is directly connected to the direction of arrival of the sound. This is effective in achieving distinction between the target sound and colored noise (interfering sound), which was difficult in the past by focusing on the direction of arrival, but when the target sound and colored noise (interfering sound) are superimposed, The problem is that the target sound section is determined regardless of the presence or absence of colored noise (interfering sound).

  The inventor of the present application has proposed a technique described in Japanese Patent Application No. 2015-189919 (referred to as a reference) in order to improve the above-described problems. This is a technique for determining the presence of colored noise (interfering sound) based on the sign of the correlation between coherence (formula (7) of Patent Document 1) and the front suppression signal. By this method, even if the colored noise superimposed on the target sound is detected, the presence of the colored noise can be detected only by monitoring whether or not the correlation is negative.

  However, in a section where there is no colored noise, the correlation is positive regardless of whether it is the target sound section or the background noise section, so that the background noise section has not been detected. For example, it is possible to classify target sound sections, interfering sound sections, and background noise sections by using a plurality of signal section detection methods such as combining Patent Document 1 and a reference document. However, as described above, it is applied to a voice assistant. In order to achieve this, real-time characteristics are required, and therefore, a method is required in which a plurality of detection methods are not executed simultaneously but can be classified by a single detection method.

  Therefore, in view of the above problems, an acoustic signal that can accurately determine (classify) in real time that the processing load is reduced and the input signal is a signal section of any of the target sound, interference sound, and background noise. There is a need for a processing device, method and program.

  In order to solve such a problem, the acoustic signal processing apparatus according to the first aspect of the present invention is (1) a plurality of frequencies obtained by converting each input signal from each of a plurality of microphones from a time domain to a frequency domain. A front suppression signal generation unit that generates a frontal suppression signal having a blind spot on the front surface based on the difference between the region input signals; and (2) a coherence calculation unit that calculates coherence based on signals obtained from a plurality of input signals; (3) A classification unit that classifies the signal section of the input signal based on a feature value indicating the relationship between the front suppression signal and coherence and a variable obtained in the feature value calculation process. And

  The acoustic signal processing program according to the second aspect of the present invention provides a computer, (1) a difference between a plurality of frequency domain input signals obtained by converting each input signal from each of a plurality of microphones from a time domain to a frequency domain. A front suppression signal generation unit that generates a frontal suppression signal having a blind spot on the front side, (2) a coherence calculation unit that calculates coherence based on signals obtained from a plurality of input signals, and (3) frontal suppression. It is characterized by functioning as a classification unit that classifies signal sections of an input signal based on a feature value indicating a relationship between a signal and coherence and a variable obtained in a feature value calculation process.

  The acoustic signal processing method according to the third aspect of the present invention includes: (1) a plurality of frequency domains obtained by the front suppression signal generation unit converting each input signal from each of the plurality of microphones from the time domain to the frequency domain; A front suppression signal having a blind spot in front is generated based on the difference between the input signals, (2) a coherence calculation unit calculates coherence based on signals obtained from a plurality of input signals, and (3) a classification unit The signal section of the input signal is classified based on the feature value indicating the relationship between the front suppression signal and the coherence and the variable obtained in the feature value calculation process.

  According to the present invention, it is possible to accurately determine (classify) in real time with a low processing load whether an input signal is a signal section of target sound, interference sound, or background noise.

1 is a block diagram showing an overall configuration of an acoustic signal processing device according to an embodiment. It is explanatory drawing explaining the example of arrangement | positioning of the microphone which concerns on embodiment. It is a figure which shows the characteristic of the directional signal applied with the acoustic signal processing apparatus which concerns on embodiment. It is a block diagram which shows the internal structure of the signal classification | category part which concerns on embodiment. It is a flowchart which shows the signal classification | category process in the classification | category part which concerns on embodiment.

(A) Main Embodiments Hereinafter, main embodiments of an acoustic signal processing device, method, and program according to the present invention will be described in detail with reference to the drawings.

(A-1) Configuration of Embodiment FIG. 1 is a block diagram showing an overall configuration of an acoustic signal processing device 1 according to the embodiment.

  As shown in FIG. 1, the acoustic signal processing apparatus 1 acquires input signals s1 (n) and s2 (n) from a plurality of microphones m_1 and m_2 (two cases are shown in FIG. 1). Note that n is an index indicating the input order of samples and is expressed by a positive integer. In the following, it is assumed that the smaller n is the older input sample, and the larger n is the new input sample.

  The acoustic signal processing device 1 determines (classifies) whether the input signal acquired from the microphones m_1 and m_2 is a signal section of the target sound, the disturbing sound, or the background noise, and the determination result is displayed in the subsequent stage. To the audio processing apparatus 2 of FIG.

  The sound processing device 2 performs predetermined processing on the input signals acquired from the microphones m_1 and m_2 using the determination result from the acoustic signal processing device 1. For example, the audio processing device 2 can use the determination result of the audio signal processing device 1 for a communication device or communication software such as a video conference system or a telephone terminal, or pre-processing for voice recognition. The content of processing performed by the audio processing device 2 is not particularly limited, and various types of processing can be applied. For example, it may be used for other arbitrary processing such as control of noise correlation matrix calculation of sound source localization processing and control of interference noise suppression processing. Note that the acoustic signal processing device 1 and the sound processing device 2 are only required to be able to exchange signals, and may be connected to circuits by wiring, or for example, a network via a wired line or a wireless line It may be one that can send and receive signals by communication.

  FIG. 2 is an explanatory diagram for explaining an arrangement example of the microphones m_1 and m_2.

  As shown in FIG. 2, the microphones m_1 and m_2 are arranged so that the plane including the two microphones m_1 and m_2 is perpendicular to the direction in which the target sound arrives (the direction of the target sound source). And In the following, as shown in FIG. 2, the arrival direction of the target sound is referred to as the front direction or the front direction when viewed from the position between the two microphones m_1 and m_2. In the following, as shown in FIG. 2, when referring to the right direction, the left direction, and the backward direction, each direction when viewing the arrival direction of the target sound from the position between the two microphones m_1 and m_2 is shown. It will be explained as a thing. In this embodiment, it is assumed that the target sound comes from the front direction of the microphones m_1 and m_2, and the non-target sound including the disturbing sound comes from the left-right direction (lateral direction).

  As illustrated in FIG. 1, the acoustic signal processing device 1 includes an FFT unit 11, a front suppression signal generation unit 12, a coherence calculation unit 13, and a signal classification unit 14.

  The acoustic signal processing apparatus 1 may be realized by installing a program (acoustic signal processing program) in a computer having a processor, a memory, and the like. In this case, the acoustic signal processing apparatus 1 is functionally shown in FIG. Can be used to show. Note that part or all of the acoustic signal processing device 1 may be realized by hardware.

  The FFT unit 11 receives the input signals s1 and s2 from the microphones m_1 and m_2 via an AD converter (not shown), and performs fast Fourier transform (or discrete Fourier transform) on the input signals s1 and s2. . As a result, the input signals s1 and s2 are expressed in the frequency domain.

  Note that, in performing the fast Fourier transform, the FFT unit 11 includes an analysis Fourier FRAME1 (K) and a predetermined N (N is an arbitrary integer) samples from the input signals s1 (n) and s2 (n). Assume that FRAME2 (K) is configured. An example of configuring FRAME1 from the input signal s1 is shown in the following equation (1).

In the equation (1), K is an index indicating the order of frames and is expressed by a positive integer. In the following, it is assumed that the smaller the K value, the older the analysis frame, and the larger the K value, the newer the analysis frame. In the following description, it is assumed that the index representing the latest analysis frame to be analyzed is K unless otherwise specified.

  The FFT unit 11 performs a fast Fourier transform process for each analysis frame, thereby performing a frequency domain signal X1 (f, K) obtained by performing a Fourier transform on the analysis frame FRAME1 (K) configured from the input signal s1, and an input A frequency domain signal X2 (f, X) obtained by performing Fourier transform on the analysis frame FRAME2 (K) configured from the signal s2 is supplied to the front suppression signal generation unit 12 and the coherence calculation unit 13.

Here, f is an index representing a frequency. Further, the frequency domain signal X1 (f, K) is not a single value but is composed of m (m is an arbitrary integer) spectral components of a plurality of frequencies f1 to fm as shown in the equation (2). Suppose it is a thing.

  In the above equation (2), X1 (f, K) is a complex number and consists of a real part and an imaginary part. The same applies to X2 (f, K) and the front suppression signal N (f, K) described in the front suppression signal generation unit 12 described later.

  The front suppression signal generation unit 12 performs a process of suppressing the signal component in the front direction for each frequency with respect to the signal supplied from the FFT unit 11. In other words, the front suppression signal generation unit 12 functions as a directivity filter that suppresses a component in the front direction.

  For example, as shown in FIG. 3, the front suppression signal generation unit 12 uses an 8-shaped bi-directional filter having a blind spot in the front direction to generate a front direction component from the signal supplied from the FFT unit 11. A directional filter that suppresses the noise is formed.

Specifically, the front suppression signal generation unit 12 performs a calculation such as the following equation (3) based on the signals X1 (f, K) and X2 (f, K) supplied from the FFT unit 11. Thus, the front suppression signal N (f, K) for each frequency is generated. The calculation of the following equation (3) corresponds to a process of forming an 8-shaped bi-directional filter having a blind spot in the front direction as shown in FIG.
N (f, K) = X1 (f, K) -X2 (f, K) (3)

  As described above, the front suppression signal generation unit 12 acquires each frequency component of frequencies f1 to fm (power for one frame in each frequency band).

Further, the front suppression signal generator 12 calculates an average front suppression signal AVE_N (K) by averaging the front suppression signals N (f, K) over all frequencies f1 to fm according to the equation (4). To do.

  The coherence calculation unit 13 forms a highly directional signal in a specific direction included in the frequency domain signals X1 (f, K) and X2 (f, K) from the FFT unit 11 to calculate coherence COH (K). .

  Here, the calculation processing of coherence COH (K) in the coherence calculation unit 13 will be described.

The coherence calculator 13 processes the signal B1 (f, K) obtained by processing the frequency domain signals X1 (f, K) and X2 (f, K) with a filter having strong directivity in the first direction (for example, the left direction). The coherence calculation unit 13 forms the signal B2 (f) processed from the frequency domain signals X1 (f, K) and X2 (f, K) with a filter having strong directivity in the second direction (for example, the right direction). , K). An existing method can be applied to the formation method of the signals B1 (f) and B2 (f) having high directivity in a specific direction. Here, the following equation (5) is applied to the first direction. An example in which the signal B1 having high directivity is formed and the signal B2 having high directivity in the second direction is formed by applying the following equation (6) will be described.

  In the above formulas (5) and (6), S is the sampling frequency, N is the FFT analysis frame length, τ is the difference in arrival time of sound waves between the microphone m_1 and the microphone m_2, i is the imaginary unit, and f is the frequency. .

Next, the coherence calculation unit 13 performs the operations shown in the following expressions (7) and (8) on the signals B1 (f) and B2 (f) obtained as described above. Obtain coherence COH (K). Here, B2 (f, K) ^* in the equation (7) is a conjugate complex number of B2 (f, K).

  coef (f, K) is a coherence in a component of an index K having an arbitrary index K (an arbitrary frequency f (any one of frequencies f1 to fm) constituting the analysis frames FRAME1 (K) and FRAME2 (K)). .

  When obtaining coef (f, K), if the directionality of the signal B1 (f) is different from that of the signal B (f), the signals B1 (f) and B2 ( The directivity direction according to f) may be any direction other than the front direction. Further, the method for calculating coef (f, K) is not limited to the above calculation method.

  The signal classification unit 14 uses the front suppression signal N (f, N) (average front suppression signal AVE_N (K)) having directivity other than the front and the coherence COH (K), and the interference sound, the target sound, A signal interval of background noise is determined (classified), and a signal R (K) indicating the determination result is output.

  FIG. 4 is a block diagram illustrating a configuration of the signal classification unit 14 according to the embodiment.

  As shown in FIG. 4, the signal classification unit 14 includes a front suppression signal acquisition unit 21, a coherence acquisition unit 22, a correlation coefficient calculation unit 23, a classification unit 24, and a result output unit 25.

  The front suppression signal acquisition unit 21 acquires the average front suppression signal AVE_N (K) from the front suppression signal generation unit 12. Further, the coherence acquisition unit 22 acquires coherence COH (K).

  The correlation coefficient calculation unit 23 calculates a correlation coefficient cor (K), which is a feature amount indicating the relationship between the average front suppression signal AVE_N (K) and the coherence COH (K).

Here, the calculation method of the correlation coefficient cor (K) is not limited. For example, the correlation coefficient cor (K) is obtained for each frame using the following equation (9). In the following equation (9), cov [AVE_N (K), COH (K)] indicates the covariance between the average front suppression signal AVE_N (K) and the coherence COH (K). In the following equation (9), σAVE_N (K) represents the standard deviation of the average front suppression signal AVE_N (K), and σCOH (K) represents the standard deviation of the coherence COH (K). The correlation coefficient cor (K) obtained in this way takes a value of −1.0 to 1.0.

  The classification unit 24 determines whether or not a disturbing sound exists based on the sign of the correlation coefficient cor (K) between the average front suppression signal AVE_N (K) and the coherence COH (K). And when there is no interfering sound, based on the variable obtained in the process of calculating the correlation coefficient cor (K) between the average front suppression signal AVE_N (K) and the coherence COH (K), the background noise and the target sound And a second determination unit 242 for determining a signal section.

  First, the first determination unit 241 of the classification unit 24 determines whether or not an interference sound exists based on the correlation coefficient cor (K) between the average front suppression signal AVE_N (K) and the coherence COH (K). Processing will be described.

  Here, there is a sound source that emits a target sound in the front direction of the microphone m_1 and the microphone m_2, and the disturbing sound is generated from a direction other than the front direction (for example, the lateral direction of the microphone m_1 and the microphone m_2 (that is, the left direction and the right direction). Shall arrive.

  For example, when “no disturbing voice is present” and “the target sound is present”, the front suppression signal N (f, K) has a signal value proportional to the magnitude of the target sound component. However, as shown in FIG. 2, since the gain in the front direction is smaller than the gain in the horizontal direction, the gain is smaller than that in the case where an interfering sound is present.

  The coherence COH (K) is a feature quantity that has a deep relationship with the arrival direction of the input signal, and is rephrased as a correlation between two signal components. This is because the equation (6) is an equation for calculating the correlation for a certain frequency component, and the equation (7) is an equation for calculating the average of the correlation values of all frequency components, so that the coherence COH ( When K) is small, it can be said that the correlation between the two signal components is small, and conversely, when the coherence COH (K) is large, it can be said that the correlation between the two signal components is large. The input signal when the coherence COH (K) is small can be said to be a signal arriving from a direction other than the front direction because the arrival direction is greatly biased to either the right direction or the left direction. On the other hand, the input signal when the coherence COH (K) is large can be said to be a signal arriving from the front direction with little deviation in the arrival direction.

  Then, if “no interference sound exists” and “the target sound exists”, the coherence COH (K) becomes a large value, “the interference sound exists”, and “the target sound exists”. The coherence COH (K) is a small value.

When the above behavior is organized by focusing on the presence or absence of interfering sounds, the following relationship is obtained.
When “no interference sound exists” and “target sound exists”, the coherence COH (K) becomes a large value, and the front suppression signal N (f, K) (average front suppression signal AVE_N (K)) Is a value proportional to the magnitude of the target sound component.
When “disturbance sound exists”, the coherence COH (K) has a small value, and the front suppression signal N (f, K) (average front suppression signal AVE_N (K)) has a large value.

By the way, in the case of the above behavior, if the correlation coefficient cor (K) between the front suppression signal N (f, K) (average front suppression signal AVE_N (K)) and coherence COH (K) is introduced, The same can be said.
When “no disturbing sound exists”, the correlation coefficient cor (K) is a positive value (cor (K)> 0).
When “interference sound exists”, the correlation coefficient cor (K) is a negative value (cor (K) ≦ 0).

  Therefore, the first determination unit 241 observes the sign of the correlation coefficient cor (K) between the average front suppression signal AVE_N (K) and the coherence COH (K), and the correlation coefficient cor (K) is positive. It can be determined that there is no interfering sound, and it can be determined that there is an interfering sound when the correlation coefficient cor (K) is negative.

  Next, the second determination unit 242 of the classification unit 24 obtains the correlation coefficient cor (K) between the average front suppression signal AVE_N (K) and the coherence COH (K) when no disturbing sound exists. A process of determining the background noise and the signal section of the target sound based on the obtained variables will be described.

  As described above, the correlation coefficient cor (K) can be calculated using the equation (9).

  In equation (9), the numerator is the covariance of the mean front suppression signal AVE_N (K) and the coherence COH (K), and the denominator is the standard deviation σ (COH (K)) of the coherence and the mean front suppression signal. Of the standard deviation σ (AVE_N (K)).

  Assuming that the target sound comes from the front direction, the variable that changes most greatly between the target sound section and the background noise section in the equation (9) is the standard deviation σ (COH (K)) of coherence. This is because the coherence COH (K) has a large value if there is a signal component coming from the front direction, so that the fluctuation range of the coherence COH (K) increases in the target sound section, whereas in the background noise section. This is because the fluctuation range of the coherence COH (K) is small.

  By using this property, it is possible to further classify into a target sound section and a background noise section in a section where no disturbing sound exists.

In other words, the second determination unit 242 uses the value of the standard deviation σ (COH (K)) of the coherence, which is a variable obtained in the process of calculating the correlation coefficient cor (K), when there is no interfering sound as the threshold Θ. Based on the above, it is determined whether it is the target sound section or the background noise section as follows.
The signal interval in which the correlation coefficient cor (K) is positive and the standard deviation σ ((COH (K)) of the coherence is greater than or equal to the threshold Θ is the target sound interval.
A signal interval in which the correlation coefficient cor (K) is positive and the standard deviation σ (COH (K)) of the coherence is smaller than the threshold Θ is a background noise interval.

  As described above, the second determination unit 242 of the classification unit 24 is a variable obtained in the process of calculating the correlation coefficient cor (K) between the coherence COH (K) and the average front suppression signal AVE_N (K). By adding the value of the standard deviation σ (COH (K)) of coherence, it is possible to classify signal sections in detail by a single method. As a result, the determination result of the signal section can be notified promptly even when real-time performance is required in the subsequent audio processing device 2.

  In addition, the result output unit 25 outputs a signal R (K) indicating the determination results by the first determination unit 241 and the second determination unit 242 to the subsequent audio processing device 2. The format of the signal R (K) is not limited. For example, when the interference sound exists, the classification unit 24 indicates a value indicating the presence of the interference sound (for example, the signal R (K) = “0”). ), In the case of the target sound section, a value indicating the target sound section (for example, signal R (K) = “1”) is output, and in the case of the background noise section, a value indicating the background noise section (for example, signal R (K) = "2") may be output. In this embodiment, in order to facilitate the description, the result output unit 25 exemplifies a case where the signal R (K) is output to the sound processing device 2, but the present invention is not limited to this.

(A-2) Operation | movement of embodiment Next, the processing operation | movement which classify | categorizes the signal area of the input signal in the acoustic signal processing apparatus 1 which concerns on embodiment is demonstrated in detail with reference to drawings.

  First, input signals s1 (n) and s2 (n) for one frame (for one processing unit) are supplied from each of the microphones m_1 and m_2 to the FFT unit 11 via an AD converter (not shown). The FFT unit 11 performs Fourier transform on the analysis frames FRAME1 (K) and FRAME2 (K) based on the input signals s1 (n) and s2 (n) for one frame, and a signal X1 (f, K) indicated in the frequency domain. , X2 (f, K). The signals X1 (f, K) and X2 (f, K) generated by the FFT unit 11 are given to the front suppression signal generation unit 12 and the coherence calculation unit 13.

  The front suppression signal generator 12 calculates a front suppression signal N (f, K) based on the signals X1 (f, K) and X2 (f, K) from the FFT unit 11. Then, the front suppression signal generation unit 12 calculates an average front suppression signal AVE_N (K) based on the front suppression signal N (f, K), and provides it to the signal classification unit 14.

  The coherence calculation unit 13 generates coherence COH (K) based on the signals X1 (f, K) and X2 (f, K) from the FFT unit 11 and gives them to the signal classification unit 14.

  In the signal classification unit 14, the correlation coefficient calculation unit 23 uses the equation (9), for example, to calculate a correlation coefficient that is a feature amount indicating the relationship between the average front suppression signal AVE_N (K) and the coherence COH (K). Calculate cor (K).

  Then, the classification unit 24 calculates the correlation coefficient cor (K) calculated by the correlation coefficient calculation unit 23 and the standard deviation σ (COH (K)) of the coherence obtained in the calculation process of the correlation coefficient cor (K). And determining the presence / absence of interfering sound, the target sound section or the background noise section, and outputting a signal R (K) indicating the determination result.

  FIG. 5 is a flowchart showing signal classification processing in the classification unit 24 according to the embodiment.

  First, in the classification unit 24, the first determination unit 241 determines whether or not the value of the correlation coefficient cor (K) between the average front suppression signal AVE_N (K) and the coherence COH (K) is positive ( S101). When the value of the correlation coefficient cor (K) is not positive (when cor (K) ≦ 0), the first determination unit 241 determines that there is an interfering sound (S102).

  On the other hand, when the value of the correlation coefficient cor (K) is positive (when cor (K)> 0), that is, in a signal section where no disturbing sound exists, the second determination unit 242 uses the correlation coefficient cor (K). It is determined whether or not the value of the standard deviation σ (COH (K)) of the coherence obtained in the calculation process is equal to or greater than a threshold value (S103).

  If the value of the standard deviation σ (COH (K)) of the coherence is equal to or greater than the threshold Θ, the second determination unit 242 determines that the target sound section is present (S104), and the standard deviation σ (COH (K) of the coherence ) Is less than the threshold value Θ, the second determination unit 242 determines that the background noise section (S105).

  As described above, the signal classification unit 14 calculates the correlation coefficient cor (K) between the average front suppression signal AVE_N (K) and the coherence COH (K), and based on the sign of the correlation coefficient cor (K). To determine whether there is any interfering sound. Then, when there is no interfering sound, the signal classification unit 14 determines the purpose based on the value of the standard deviation σ (COH (K)) of coherence, which is a variable obtained by calculating the correlation coefficient cor (K). In order to determine whether it is a sound section or a background noise section, it is possible to classify the presence of a disturbing sound, the target sound section, or the background noise section by a single method.

  Then, the result output unit 25 outputs a signal R (K) indicating the determination result by the classification unit 24.

(A-3) Effect of Embodiment As described above, according to this embodiment, when the interference sound exists, the correlation coefficient between the front suppression signal and the coherence is negative, and the interference sound does not exist. When there is no interfering sound, the input signal interval is obtained by using the characteristic behavior that if the standard deviation of the coherence, which is an intermediate variable, is large, it is the target sound interval, and if it is small, it is the background noise interval. Can be easily classified. By using this result, for example, the accuracy is improved by controlling the sound source localization processing based on whether the noise is a disturbance sound (spatial color) or background noise (spatial white), Appropriate processing can be realized. In addition, since the processing load is small, real-time performance can be guaranteed. As described above, the satisfaction of the voice assistant user can be improved by applying the method of this embodiment to the speech recognition preprocessing used for the voice assistant or the like.

(B) Other Embodiments Although various modified embodiments have been mentioned in the above-described embodiments, the present invention can also be applied to the following modified embodiments.

  (B-1) In the above-described embodiment, the case where the target sound section and the background noise section are classified based on the standard deviation of the coherence in the signal section where no disturbing sound exists is illustrated. However, the variable for determining the signal interval is not limited to the standard deviation of coherence. This is because when there is no interfering sound, the target sound section and the background noise section are classified by paying attention to the magnitude of the coherence fluctuation range. Therefore, the target sound section and the background noise section may be classified based on the coherence variance without being limited to the standard deviation of the coherence. Also in this case, since the value of the coherence variance is obtained in the process of calculating the standard deviation of the coherence, the processing load is small.

  DESCRIPTION OF SYMBOLS 1 ... Acoustic signal processing apparatus, 11 ... FFT part, 12 ... Front suppression signal generation part, 13 ... Coherence calculation part, 14 ... Signal classification part, 21 ... Front suppression signal acquisition part, 22 ... Coherence acquisition part, 23 ... Phase relationship Number calculation unit, 24 ... classification unit, 241 ... first determination unit, 242 ... second determination unit, 25 ... result output unit.

Claims (5)

Front-side suppression signal generation that generates a front-side suppression signal having a blind spot in front based on the difference between the plurality of frequency-domain input signals obtained by converting each input signal from each of the plurality of microphones from the time domain to the frequency domain And
A coherence calculator that calculates coherence based on signals obtained from the plurality of input signals;
A classification unit that classifies a signal section of the input signal based on a value of a feature amount indicating a relationship between the front suppression signal and the coherence and a variable obtained in the process of calculating the feature amount. An acoustic signal processing device. The classification unit is
A first determination unit that determines the presence / absence of an interfering sound based on the correlation coefficient between the front suppression signal and the coherence as the feature amount, and based on whether the correlation coefficient is positive or negative;
And a second determination unit that determines a target sound section and a background noise section based on a variable that affects a magnitude of a fluctuation range of the coherence in a signal section where the interfering sound does not exist. Item 2. The acoustic signal processing device according to Item 1.   The second determination unit compares the standard deviation value of the coherence with a threshold value. If the standard deviation value of the coherence is equal to or larger than the threshold value, the target sound period is set. If not, the background noise period is set. The acoustic signal processing apparatus according to claim 2. Computer
Front-side suppression signal generation that generates a front-side suppression signal having a blind spot in front based on the difference between the plurality of frequency-domain input signals obtained by converting each input signal from each of the plurality of microphones from the time domain to the frequency domain And
A coherence calculator that calculates coherence based on signals obtained from the plurality of input signals;
Functioning as a classification unit for classifying the signal section of the input signal based on the feature value indicating the relationship between the front suppression signal and the coherence and the variable obtained in the feature value calculation process. An acoustic signal processing program. A front suppression signal having a blind spot in front based on a difference between a plurality of frequency domain input signals obtained by converting each input signal from each of a plurality of microphones from a time domain to a frequency domain by a front suppression signal generation unit. Produces
A coherence calculator calculates coherence based on signals obtained from the plurality of input signals;
A classification unit classifying a signal section of an input signal based on a feature value indicating a relationship between the front suppression signal and the coherence, and a variable obtained in a process of calculating the feature value; Acoustic signal processing method.

Images