JP6128547B2 - Sound source separation method, apparatus and program - Google Patents

Description

  The present invention relates to a sound source separation method, apparatus, and program for forming directivity toward a sound source in an arbitrary direction based on a sound wave signal.

  In order to accurately separate the sound waves of the target sound source and suppress undesired sounds such as noise, it is generally necessary to use a directional microphone or to arrange a plurality of these microphones at regular intervals. However, in a small sound collecting device such as an IC recorder, it is difficult to apply a sound collecting technique using a directional microphone and a plurality of microphones with wide intervals. In addition, it is difficult to accurately separate sound sources by applying these sound collection techniques to recorded sound that has been artificially downmixed from a plurality of sound sources.

  Therefore, many techniques have been proposed for separating and extracting the target sound source by analyzing the amplitude difference and phase difference between the signals output from each microphone after sound wave recording and applying signal processing according to the analysis result. . In recent years, statistical analysis, frequency analysis, complex analysis, and the like have been brought out to detect differences in the waveform structure of input signals, and use the detection results for sound source separation processing.

  For example, the input signal is converted from the time axis to the frequency axis, the phase difference is calculated for each frequency, the sound wave frequency band input from the target sound source is identified based on this difference, and the sound wave in that frequency band is emphasized Signal processing is performed as described above (see Patent Document 1).

  Further, in the signal processing, it is determined whether or not the sound wave input is in the target direction based on the input signals of the two microphones arranged close to each other, the difference in phase difference between the two input signals is corrected, and the target direction is corrected. Is emphasized (see Patent Document 2). Two input signals are referred to each other, and the filter is sequentially updated using the obtained signals (see Patent Document 3).

JP 2007-318528 A Special table 2009-135593 JP 2009-027388 A

  The downsizing of sound collecting devices or devices equipped with sound collecting devices is accompanied by a further closer arrangement of microphones, and therefore the amplitude difference and phase difference between signals are minimized, and these amplitude differences and phase differences are clearly defined. A great deal of effort is required to identify these. In particular, in a signal from a distant sound source, the amplitude difference is minimal, so it is difficult to clearly identify the amplitude difference and the phase difference.

  In recent years, as disclosed in Patent Documents 1 to 3, the frequency analysis, complex analysis, or statistical analysis of the waveform structure is complicated and advanced to cope with the proximity of microphones. However, these complicated sophistication results in an increase in the frame length when converting to the frequency domain, a large number of delay devices, a long filter length, a long filter coefficient, etc. It has become difficult to form directivity in real time. The number of microphones may be increased in order to reduce the processing load, but the distance between the microphones is further shortened due to the limited space of the device.

  The present invention has been made to solve the above-described problems of the prior art, and its purpose is to reduce the amount of computation of a sound coming from an arbitrary direction without requiring complex and sophisticated analysis. It is an object to provide a sound source separation method, apparatus, and program that can be emphasized or suppressed with a sound source.

In order to achieve the above object, the sound source separation method of the embodiment is a sound source separation method that forms directivity in a specific direction with respect to a pair of sound wave signals, and is provided in an arbitrary filter and subsequent stage. Ru used and any number of adders and changes the amplitude ratio of said pair of acoustic signals, and the amplitude ratio setting step of setting a specific amplitude ratio for each direction, so that the amplitude in the specific direction is the same to include an amplitude changing step of changing the amplitude ratio of the specific direction of the pair of acoustic signals, the gain decision to determine the gain according to the difference between the waveforms of the pair of ultrasonic signals having passed through the amplitude change step They include a step, an output step of outputting by multiplying the gain in sound signal, and characterized.

  In the amplitude ratio setting step, the amplitude of the pair of sound wave signals may be changed at a specific change rate depending on the direction, and the amplitude derived from the same direction may be changed at a different change rate.

  The polar pattern indicating the ratio of the change rates may be different for each pair of sound wave signals.

  The change rate of the amplitude may be larger in a specific direction in which emphasis or attenuation is performed than the amplitude rate in other directions.

  The gain determination step includes: an exchange step for generating a pair of exchange signals by alternately exchanging the pair of synthesized signals for each sample by an exchange circuit using the synthesized signal that has undergone the amplitude changing step; A generation step of generating an error signal of the exchange signal after multiplying one of them by a coefficient m, an update step of calculating a recurrence formula of the coefficient m including the error signal and updating the coefficient m for each sample; An output step of multiplying the pair of combined signals by the sequentially updated coefficient m and outputting the resultant signal.

  The pair of sound wave signals may be a pair of sound wave signals input from a pair of microphones.

  The pair of sound wave signals may be a convolution of a system represented by a transfer coefficient when a system from a sound source to a microphone is simulated.

  In order to achieve the above object, the sound source separation device according to the embodiment is a sound source separation device that forms directivity in a specific direction with respect to a pair of sound wave signals, and includes an arbitrary number of filters, and thereafter An arbitrary number of adders provided in a stage and a filtering process for changing the amplitude of the pair of sound wave signals at a certain ratio, and the certain ratio is an amplitude for changing a specific amplitude ratio to be the same. The gain is determined according to the difference between the changing unit, the amplitude changing unit that changes the specific amplitude ratio among the amplitude ratios of the pair of sound wave signals to the same amplitude, and the waveform of the pair of sound wave signals that have undergone the amplitude changing step. A gain determination unit; and an output unit that multiplies the sound wave signal by the gain and outputs the signal.

  Furthermore, in order to achieve the above-described object, the sound source separation program according to the embodiment is a sound source separation program that forms directivity in a specific direction with respect to a pair of sound wave signals. An amplitude ratio setting means for setting a computer having an arbitrary number of adders provided in a stage to a unique value for each direction by using the filter and the adder, And applying a filtering process for changing the amplitude of the pair of sound wave signals at a constant ratio, the fixed ratio being an amplitude changing means for changing a specific amplitude ratio to be the same, and a pair having undergone the amplitude changing step. And a gain determining means for determining a gain according to a difference between waveforms of the sound wave signal and an output means for multiplying the sound wave signal by the gain and outputting the gain.

  According to the present invention, the directivity in an arbitrary direction can be accurately emphasized while greatly reducing the number of operations without requiring a complicated analysis for specifying the amplitude and difference between signals.

It is a block diagram which shows the structure of the sound source separation apparatus which concerns on embodiment of this invention. It is a block diagram which shows an example of a coefficient determination circuit. It is a block diagram which shows an example of a coefficient update circuit. It is a model which shows the relationship between a sound source and microphones M1 and M2. It is a polar plot which shows the directivity of acoustic signal InM1 (k) and InM2 (k). It is a graph which shows the amplitude ratio between 270 deg-90 deg of sound wave signal InM1 (k) and InM2 (k). It is a polar plot of synthetic | combination signal InA (k) and InB (k). It is a graph which shows the amplitude ratio between 270deg-90deg of synthetic | combination signal InA (k) and InB (k). It is the graph which shifted the amplitude ratio of 30deg of synthetic | combination signal InA (k) and InB (k) to 1. FIG. It is a polar plot in which the amplitudes of the sound wave signals InM1 (k) and InM2 (k) are adjusted to eliminate the difference in the 50 deg direction. It is a polar plot in which the amplitudes of the combined signals InA (k) and InB (k) are adjusted to eliminate the difference in the 50 deg direction. It is a graph which shows coefficient m (k) according to 270deg-90deg when there is no amplitude ratio setting part. It is a graph which shows (m (k) +1) / 2 normalized so that it may become 0 to 1 about coefficient m (k) according to 270deg-90deg when there is an amplitude ratio setting part. It is a graph which shows the convergence speed of the coefficient m (k) according to the presence or absence of an exchange circuit. It is a block diagram which shows the structure of the sound source separation apparatus which concerns on other embodiment. It is a block diagram which shows the structure of the sound source separation apparatus which concerns on other embodiment.

  Hereinafter, embodiments of a sound source separation method and apparatus according to the present invention will be described in detail with reference to the drawings.

(First embodiment)
(Constitution)

  FIG. 1 is a block diagram illustrating a configuration of a sound source separation device. As shown in FIG. 1, the sound source separation device is connected to a pair of microphones M1 and M2 that are spaced apart. The sound source separation device receives the sound wave signal InM1 (k) and the sound wave signal InM2 (k) from the microphones M1 and M2, and performs signal processing on the sound wave signal InM1 (k) and the sound wave signal InM2 (k) to be specific to each direction. Are combined into a pair of combined signals having an amplitude ratio of The sound source separation device emphasizes the relative amplitude of the sound wave in the specific direction as compared with the sound wave coming from the other direction by enhancing the amplitude ratio of the pair of synthesized signals in the specific direction. The direction is the arrival direction of the sound wave toward the midpoint with respect to the midpoint of the microphones M1 and M2, in other words, the direction in which the sound source is arranged. The specific direction is the direction of the target sound source.

In the sound source separation device, an amplitude ratio setting unit 1 is provided at the subsequent stage of the microphones M1 and M2. The amplitude ratio setting unit 1 generates a combined signal InA (k) and a combined signal InB (k) from the sound wave signal InM1 (k) and the sound wave signal InM2 (k). The synthesized signal InA (k) is a signal obtained by changing the amplitude of the sound wave signal InM1 (k), and the synthesized signal InB (k) is a signal obtained by changing the amplitude of the sound wave signal InM2 (k). The amplitude ratio setting unit 1 outputs the sound wave signal InM1 (k) and the sound wave signal InM2 (k) so that the combined signal InA (k) and the combined signal InB (k) have specific amplitude ratios depending on directions. To process.
That is, the amplitude ratio setting unit 1 changes the amplitudes of the sound wave signal InM1 (k) and the sound wave signal InM2 (k) at a specific change rate depending on the direction. Furthermore, the amplitude ratio setting unit 1 changes the amplitudes of the sound wave signal InM1 (k) and the sound wave signal InM2 (k) derived from the same direction at different change rates. In other words, the amplitude ratio setting unit 1 has a polar pattern that defines the amplitude change rate of the sound wave signal InM1 (k) and a polar pattern that defines the amplitude change rate of the sound wave signal InM2 (k). Both polar patterns differ in amplitude change rate defined in the same direction. Furthermore, the ratio of the amplitude change rate defined by both polar patterns is unique for each direction.
This polar pattern is realized by passing the sound wave signal InM1 (k) and the sound wave signal InM2 (k) through the filters H1 to H4 and the adders 1a and 1b.

  The amplitude ratio setting unit 1 includes filters H1 to H4 and adders 1a and 1b in order to synthesize the synthesized signal InA (k) and the synthesized signal InB (k). The filters H1 to H4 are filters that filter the sound wave signal InM1 (k) and the sound wave signal InM2 (k) with the transfer functions H1 to H4. The filters H1 to H4 are, for example, FIR filters and IIR filters.

The transfer functions of the filters H1 to H4 are represented by the following expressions (1) to (4), for example.
H1 = X (1)
H2 =-(X * C195_2) / C195_1 (2)
H3 =-(X * C105_1) / C105_2 (3)
H4 = X (4)
In the equation, X is a signal having an arbitrary constant delay and amplitude characteristic, and is a transfer function serving as a reference for conversion.
C195_1 is a transfer function from the sound source in the 195 [deg] direction to M1.
C195_2 is the transfer function from the sound source in the 195 [deg] direction to M2.
C105_1 is a transfer function from the sound source in the 105 [deg] direction to M1.
C105_2 is the transfer function from the sound source in the 105 [deg] direction to M2.

As shown in the equation (1), the sound wave signal InM1 (k) is filtered by the function X serving as a reference for conversion through the filter H1.
Further, as shown in Expression (3), the sound wave signal InM2 (k) passes through the filter H3, so that the sound wave signal InM1 (k) has a relationship in which the positive and negative amplitudes are inverted in the same phase. The filter H3 multiplies the value obtained by multiplying the transfer function from the 105 deg sound source to the microphone M1 and the reciprocal of the transfer function from the 105 deg sound source to the microphone M2 by the function X serving as a conversion reference. It is. That is, this filter H3 represents the ratio of the difference in the signal generated when the sound wave transmitted from the sound source in the 105 deg direction reaches the microphones M1 and M2, based on the microphone M2. This filter H3 multiplies this ratio to the sound wave signal InM2 (k) derived from any direction.
Then, the sound wave signal InM2 (k) passed through the filter H3 has the same phase and the same amplitude as the sound wave signal InM1 (k) when the sound wave signal InM2 (k) is derived from a sound source in the 105 deg direction. However, the sign is reversed. When the sound source is derived from another direction, the difference between the signals generated when the microphones M1 and M2 are reached is multiplied by the ratio of deviation, so that the amplitude of the sound wave signal InM1 (k) and the sound wave signal InM2 (k) Are not matched, and the deviation increases as the distance from 105 deg increases.

In the amplitude ratio setting unit 1, the signal obtained by filtering the sound wave signal InM1 (k) with the filter H1 and the signal obtained by filtering the sound wave signal InM2 (k) with the filter H3 are combined by the adder 1a, and the combined signal InA ( k). The combined signal InA (k) is expressed by the following (5).
Composite signal InA (k) = M1 * H1 + M2 * H3 (5)
The sound wave signal InM1 and the sound wave signal InM2 have a relationship in which the positive and negative amplitudes are inverted in the same phase by the filter H1 and the filter H3. In Expression (5), the sound wave signal InM2 (k) whose amplitude is changed is subtracted from the sound wave signal InM1 (k) that has passed through the filter H1 at a rate corresponding to the amount of deviation from 105 degrees. For this reason, the closer to 105 deg, the greater the degree of cancellation, and the further away from the 105 deg direction, the smaller the degree of cancellation. In other words, a signal having a larger amplitude is output when the distance from 105 deg is 105 deg as the minimum amplitude.

  Similarly, as shown in Expression (4), the sound wave signal InM2 (k) passes through the filter H4 and is filtered by the function X serving as a reference for conversion. Further, as shown in Expression (2), the sound wave signal InM1 (k) passes through the filter H2, so that the sound wave signal InM2 (k) has a relationship in which the positive and negative amplitudes are inverted in the same phase. The filter H2 further multiplies the value obtained by multiplying the transfer function from the 195 deg sound source to the microphone M2 and the reciprocal of the transfer function from the 195 deg sound source to the microphone M1 by the function X serving as a conversion reference. It is. That is, this filter H2 represents the ratio of the difference in the signal generated when the sound wave transmitted from the sound source in the 195 deg direction reaches the microphones M1 and M2, based on the microphone M1. The filter H2 multiplies this ratio to the sound wave signal InM1 (k) derived from any direction.

In the amplitude ratio setting unit 1, the signal obtained by filtering the sound wave signal InM1 (k) with the filter H2 and the signal obtained by filtering the sound wave signal InM2 (k) with the filter H4 are synthesized by the adder 1b, and the synthesized signal InB ( k). The combined signal InB (k) is expressed by the following (6).
Composite signal InB (k) = M1 * H2 + M2 * H4 (6)
The sound wave signal InM1 (k) and the sound wave signal InM2 (k) have a relationship in which the positive and negative amplitudes are inverted in the same phase by the filter H2 and the filter H4. In Expression (6), the sound wave signal InM1 (k) whose amplitude is changed is subtracted from the sound wave signal InM2 (k) that has passed through the filter H4 at a rate corresponding to the amount of deviation from 195 deg. For this reason, the closer to 195 deg, the greater the degree of cancellation, and the further away from the 195 deg direction, the smaller the degree of cancellation. In other words, a signal with an increased amplitude is output when the amplitude is 195 deg and the distance is away from 195 deg.

  The amplitude ratio setting unit 1 uses the filters H1 to H4 and the adders 1a and 1b to generate a combined signal InA (k) and a combined signal InB (k) having a specific amplitude ratio for each direction.

  The sound source separation device includes an amplitude changing unit 2 that changes the signal of the amplitude ratio in a specific direction in the combined signal InA (k) and the combined signal InB (k) to the same amplitude, following the amplitude ratio setting unit 1. In the amplitude changing unit 2, the amplitude ratios in the specific direction in the combined signal InA (k) and the combined signal InB (k) are set to the same amplitude, while the amplitude ratio other than the specific direction is given an amplitude difference that deviates from the specific direction. To go. The amplitude changing unit 2 includes a filter D1 and a filter T1.

  The amplitude changing unit 2 filters the combined signal InA (k) and the combined signal InB (k), thereby making the amplitude ratio in a specific direction of the combined signal InA (k) and the combined signal InB (k) the same. To do. The amplitude changing unit 2 includes a filter D1 that performs filter processing on the combined signal InA (k), and a filter T1 that performs filter processing on the combined signal InB (k).

An example of the transfer functions D1 and T1 of the filters D1 and T1 is expressed by the following equations (7) and (8).
D1 = 1 (7)
T1 = (c11 * h2 +
c12 * h4) / (c11 * h1 + c12 * h3) (8)
c11 is a transfer function from the sound source S1 to the microphone M1.
c12 is a transfer function from the sound source S1 to the microphone M2.

  With the transfer functions D1 and T1 satisfying the equations (7) and (8), the amplitude ratio in a specific direction in the combined signal InA (k) and the combined signal InB (k) is changed to the same.

  The combined signal InA (k) and combined signal InB (k) that have passed through the amplitude changing unit 2 are distributed to the coefficient determination circuit 3 and the path to the combining circuit 4. In the present embodiment, the gain g corresponding to the coefficient m (k) determined by the coefficient determination circuit 3 is determined in the synthesis circuit 4. The synthesis circuit 4 outputs an output signal OutR (k) and an output signal OutL (k) based on the gain.

  As shown in FIG. 2, the coefficient determination circuit 3 is a circuit in which a characteristic correction circuit 3a, an exchange circuit 3b, and a coefficient update circuit 3c are connected in series.

  The characteristic correction circuit 3a includes a frequency characteristic correction filter and a phase characteristic correction circuit. The frequency characteristic correction filter extracts a sound wave signal in a desired frequency band. The phase characteristic correction circuit reduces the influence of the acoustic characteristics of the microphones M1 and M2 on the sound wave signal InM1 (k) and the sound wave signal InM2 (k).

The exchange circuit 3b alternately outputs the synthesized signal InA (k) and the synthesized signal InB (k) every other sample. That is, the data strings of the exchange signal InC (k) and the exchange signal InD (k) are as follows when k = 1, 2, 3, 4,.
InC (k) = {InA (1) InB (2) InA (3) InB (4) ...}
InD (k) = {InB (1) InA (2) InB (3) InA (4) ...}

  The exchange signal InC (k) and the exchange signal InD (k) are input to the coefficient update circuit 3c. The coefficient update circuit 3c calculates an error between the exchange signal InC (k) and the exchange signal InD (k) and determines a coefficient m (k) corresponding to the error. The coefficient updating circuit 3c sequentially updates the coefficient m (k) with reference to the past coefficient m (k-1).

An error signal e (k) between the exchange signal InA (k) and the exchange signal InB (k) is defined as the following equation (9).

  The coefficient update circuit 3c uses the error signal e (k) as a function of the coefficient m (k-1) and calculates a recurrence formula between adjacent binomials of the coefficient m (k) including the error signal e (k). The coefficient m (k) that minimizes the error signal e (k) is searched. The coefficient update circuit 3c is updated in a direction to decrease the coefficient m (k) as the amplitude difference in a specific direction is generated between the combined signal InA (k) and the combined signal InB (k) by this arithmetic processing. If there is no amplitude difference, the coefficient m (k) is output close to 1.

  The coefficient m (k) is input to the synthesis circuit 4 together with the synthesis signal InA (k) and the synthesis signal InB (k). The synthesis circuit 4 multiplies the composite signal InA (k) and the composite signal InB (k) by a coefficient m (k) at an arbitrary ratio, adds them at an arbitrary ratio, and as a result, outputs the output signal OutL (k ) And an output signal OutR (k).

  An example of the coefficient update circuit 3c will be further described. FIG. 3 is a block diagram illustrating an example of the coefficient update circuit 3. As shown in FIG. 3, the coefficient update circuit 3c is composed of a plurality of accumulators and adders, and is a circuit that embodies a recurrence formula between adjacent binomials, and refers to a past coefficient m (k−1). The coefficient m (k) is gradually updated. Adaptive filters with long tap numbers are eliminated.

In the coefficient update circuit 3c, an error signal e (k) is generated using the exchange signal InD (k) as a reference signal. That is, the exchange signal InC (k) is input to the integrator 5. The accumulator 5 multiplies the exchange signal InC (k) by −1 times the coefficient m (k−1) one sample before. An adder 6 is connected to the output side of the integrator 5. The adder 6 receives the signal output from the integrator 5 and the exchange signal InD (k), and adds these signals to obtain an instantaneous error signal e (k). The error signal e (k) resulting from this arithmetic processing is as shown in the following equation (10).

The error signal e (k) is input to the integrator 7 that multiplies the sound wave signal by μ. The coefficient μ is a step size parameter of less than 1. An integrator 8 is connected to the output side of the integrator 7. The accumulator 8 receives the exchange signal InC (k) and the signal μe (k) that has passed through the accumulator. The accumulator 8 multiplies the exchange signal InC (k) and the signal μe (k) to obtain a differential signal ∂E (m) ² / ∂m of an instantaneous square error expressed by the following equation (11).

An adder 9 is connected to the integrator 8. The adder 9 completes the coefficient m (k) by calculating the following equation (12), and outputs the output signals OutL (k) and OutR (k) from the combined signal InA (k) and the combined signal InB (k). The coefficient m (k) is set in the synthesis circuit 4 that generates
That is, the adder 9 completes the coefficient m (k) by adding the signal β · m (k−1) to the differential signal ∂E (m) ² / ∂m.

  The signal β · m (k−1) is connected to the output side of the adder 9 by a delay unit 10 that delays the signal by one sample and an accumulator 11 that accumulates a constant β. The coefficient m (k−1) updated by the process is generated by multiplying the constant 11 by the integrator 11.

As a result, the coefficient update circuit 3 realizes the calculation process of the following recurrence formula (13), generates the coefficient m (k), and updates it gradually for each sampling.

(Function)
(Operation of amplitude ratio setting unit)
FIG. 4 shows the positional relationship between the sound source and the microphones M1 and M2. In the model indicating the positional relationship, microphones M1 and M2 separated by 4 cm from the origin are set on the x-axis. The sound source S1 is arranged at a distance of 1.0 m from the origin. The sound source S1 is specified by an angle with the y-axis positive direction being 0 deg and the x-axis positive direction being 90 deg. In this embodiment, the sound source S1 is arranged in the direction of 50 deg.

  The sound wave signal from the sound source S1 is input to the microphones M1 and M2. The microphones M1 and M2 are omnidirectional microphones. That is, as shown in FIG. 5, the microphones M1 and M2 have a circular polar pattern and do not have directivity with respect to a specific direction. For this reason, when a sound wave signal having the same sound pressure is input from a sound source existing at 360 °, there is no difference in the amplitudes of the sound wave signals InM1 (k) and InM2 (k).

  FIG. 6 is a diagram showing the amplitude ratio of the sound wave signals InM1 (k) and InM2 (k). Since there is almost no difference between the amplitudes of the sound wave signals InM1 (k) and InM2 (k), the amplitude ratio of the sound wave signals InM1 (k) and InM2 (k) is in the range of 270 deg to 90 deg with the sound wave signal InM1 (k). The difference in amplitude of the sound wave signal InM2 (k) is considerably small. That is, the amplitude ratio between the sound wave signal InM1 (k) and the sound wave signal InM2 (k) approaches 1: 1.

  The sound wave signals InM1 (k) and InM2 (k) input to the amplitude ratio setting unit 1 are filtered by the filters H1 to H4 and the adder. FIG. 7 is a diagram showing the combined signal InA (k) output from the adder 1a after being filtered and the combined signal InB (k) output from the adder 1b. As shown in FIG. 7, the filtered combined signal InA (k) and combined signal InB (k) have different polar patterns. For this reason, the amplitudes of the combined signal InA (k) and the combined signal InB (k) differ between 270 deg and 90 deg.

FIG. 8 is a diagram illustrating the amplitude ratio between the combined signal InA (k) and the combined signal InB (k). As shown in FIG. 8, the amplitude ratio between the combined signal InA (k) and the combined signal InB (k) is 1.1 to 8.1 in the range of 270 deg to 15 deg. In this range, the amplitude ratio does not overlap, and the amplitude ratio is a unique value for each angle.
Similarly, the amplitude ratio between the combined signal InA (k) and the combined signal InB (k) is 0.1 to 8.1 in the range of 15 deg to 90 deg. In this range, the amplitude ratio does not overlap, and the amplitude ratio is a unique value for each angle.

(Operation of amplitude changer)
The amplitude changing unit 2 filters the combined signal InA (k) and the combined signal InB with the filters T1 and D1, thereby changing the amplitude ratio at a constant rate and making the signals of the amplitude ratio in a specific direction the same. For example, the value of the filter T1 for processing the combined signal InB (k) is set so that the amplitude ratio in the 50 deg direction, which is a specific direction, becomes 1.
In this case, the amplitude ratio in the 50 deg direction becomes 1 by the filter T1. On the other hand, the amplitude ratio in the direction other than the 50 deg direction is different from the amplitude ratio in the 50 deg direction and thus does not become 1, and becomes a value away from 1 depending on the difference from the amplitude ratio in the 50 deg direction. FIG. 9 is a graph in which the amplitude ratio of the synthesized signal InA (k) and the synthesized signal InB (k) in FIG. As shown in FIG. 9, in the range of 15 deg to 90 deg, by setting the amplitude ratio in the specific direction 30 deg to 1, the amplitude ratio in the other direction deviates from 1 every time it departs from 30 deg.

  FIG. 10 is a diagram illustrating polar patterns of the sound wave signals InM1 (k) and InM2 (k) when the sound wave signals InM1 (k) and InM2 (k) are adjusted so that the amplitude difference in the 50 deg direction is eliminated. The polar patterns of the sound wave signals InM1 (k) and InM2 (k) overlap in the direction of 50 deg. In this case, as shown in FIG. 12, the coefficient m (k) becomes a coefficient m (k) closer to 1 as it is closer to 50 deg which is the specific direction.

  On the other hand, FIG. 11 is a diagram showing polar patterns of the combined signals InA (k) and InB (k) when adjustment is made so that the amplitude difference in the 50 deg direction of the combined signals InA (k) and InB (k) is eliminated. is there. The polar patterns of the combined signals InA (k) and InB (k) overlap in the direction of 50 deg. In this case, as shown in FIG. 13, the coefficient m (k) becomes a coefficient m (k) that is closer to 1 as it is closer to 50 deg that is the specific direction. In particular, by enhancing the amplitude difference in each direction in the amplitude ratio setting unit 1, a large difference can be made in the value of the coefficient m (k) in the vicinity of 50 deg. That is, sharp directivity is obtained.

  As a result, the output signal OutL (k) and the output signal OutR (k) have a gain that is relatively emphasized by a coefficient m (k) closer to 1 as the sound source location is closer to the direction of 50 deg. Given. On the other hand, the further away the sound source location is from the direction of 50 deg, the more gain that is relatively suppressed by the coefficient m (k) less than 1.

(Operation of exchange circuit)
Next, the significance of the exchange circuit will be described. By passing through the exchange circuit, the coefficient update circuit alternately calculates the following formula (14).

In Equation (14), the square term of the signal acts to reduce the uncorrelated component such as white noise as time passes. On the other hand, the adjacent term is equivalent to the numerator part of the following formula (15) for sequentially calculating the correlation coefficient, and the influence of the correlation component is reflected on the coefficient m.

  That is, when the coefficient update circuit 3c attempts to approximate the composite signal InB (k) to the composite signal InA (k), the uncorrelated component of the composite signal InA (k) becomes the amplification direction, and the composite signal InB (k ) Of the uncorrelated component is in the suppression direction. Further, when the synthesized signal InA (k) is approximated to the synthesized signal InB (k), the uncorrelated component of the synthesized signal InB (k) becomes the amplification direction, and the uncorrelated component of the synthesized signal InA (k). Is the direction of suppression.

  Therefore, when the switching circuit 2 is installed in front of the coefficient update circuit 3c, the synthesized signal InA (k) is approximated to the synthesized signal InB (k) to perform synchronous addition, and the synthesized signal InB (k) is added. On the other hand, the function of approximating the synthesized signal InA (k) and attempting to add synchronously is repeated alternately. Therefore, the function of amplifying and suppressing the uncorrelated component cancels out alternately, and the coefficient m (k) reflects the influence of the correlated component deeply.

  FIG. 14 shows the convergence state of the coefficient m (k) with and without the exchange circuit 3b. In both the convergence states, a sound source is placed at the center position, and sounds are collected by the microphones M1 and M2. As shown by the curve F in FIG. 14, the coefficient m (k) converges to 1 about 1000 times when there is the exchange circuit 3b, but when there is no exchange circuit 2 as shown by the curve G, Even if the coefficient m (k) was updated 10,000 times, it still did not converge to 1, and the opening was 10 times. That is, when the exchange circuit 3b exists, it indicates that the sound source separation is completed quickly.

(effect)
As described above, in the sound source separation device according to the present embodiment, the combined signal InA having a unique amplitude ratio for each direction from the pair of sound wave signals InM1 (k) and InM2 (k) input to the microphones M1 and M2. (K) and the synthesized signal InB (k) are synthesized. And
A signal having an amplitude ratio in a specific direction in the combined signal InA (k) and the combined signal InB (k) is changed to the same amplitude. By emphasizing this same amplitude ratio, a filtering process is performed to emphasize the sound waves in a specific direction relative to the sound waves coming from other directions. Then, after the filtering process, the exchange signal 2 is alternately exchanged with the synthesized signal InA (k) and the synthesized signal InB (k) for each sample, whereby a pair of exchange signals InC (k) and InD (k) are obtained. Then, after multiplying one of the exchange signals InC (k) and InD (k) by a coefficient m, an error signal of the exchange signals InC (k) and InD (k) is produced. Further, the recurrence formula of the coefficient m including the error signal is calculated to update the coefficient m for each sample. Finally, the sequentially updated coefficient m is multiplied by a pair of sound wave signals and output.

  For example, an adder 6 that adds one pair of exchange signals after passing one of the exchange signals through the integrator 5 in which −1 times the past coefficient m calculated one sample before is set. After passing through the adder 6, it passes through the integrator 7 in which the constant μ is set. After passing through the integrator 7, the integrator 8 in which one exchange signal before being multiplied by the past coefficient m is set. Then, after passing through the integrator 8, the coefficient m is updated for each sample by passing through the adder 9 in which the past coefficient m calculated one sample before is passed.

  Thereby, the sound source separation device of the present embodiment focuses on the amplitude ratio of the sound wave signal from the sound source S1, can avoid complicated calculation, and the amplitude difference between the sound wave signal InM1 (k) and the sound wave signal InM2 (k). Even if is very small, directivity can be formed in a specific direction.

  Furthermore, it does not depend on a filter having a large number of taps in order to form the directivity, and can be realized by an exchange circuit and a single coefficient update circuit that calculates a recurrence formula. Therefore, the number of operations can be greatly reduced, and the final delay can be suppressed within several tens of microseconds to several milliseconds.

  The combined signal InA (k) and the combined signal InB (k) only need to have a correlation such that the amplitude ratio for each direction becomes an individual value, and each combined signal has an arbitrary polar pattern. doing. As an example of the polar pattern, a cardioid type, a bi-directional type, and an acute-directional type polar pattern can be cited. Furthermore, the polar pattern of the combined signal InA (k) and the combined signal InB (k) may be line symmetric. The polar patterns of the synthesized signal InA (k) and the synthesized signal InB (k) may have different shapes.

  In the present embodiment, as shown in FIG. 8, the amplitudes of the combined signal InA (k) and the combined signal InB (k) are set so that the amplitude ratio of 15 deg is maximized. Can be in any direction. For example, when it is desired to increase the directivity in a specific direction, the amplitude ratio in that direction can be set to increase. As a result, the amplitude ratio is particularly increased in the specific direction, so that a relatively emphasized gain is given when the gain is determined later. Furthermore, the amplitude ratio may be set not only to increase the amplitude ratio but also to emphasize the amplitude ratio at a specific angle. In other words, it is easy to distinguish the amplitude ratio in a specific direction from the amplitude ratio in other directions. For example, when the amplitude ratio for each angle is graphed, the graph has the amplitude ratio in a specific direction as a peak, and the peak is steep. Thus, the greater the difference in amplitude ratio between the specific direction and the other direction, the stronger the directivity can be.

  On the other hand, when it is desired to attenuate a directivity that is negative with respect to a specific direction, that is, a sound in a specific direction, the amplitude ratio in the direction to be attenuated is set to be smaller than other amplitude ratios. As a result, the difference between the amplitude ratio in the direction to be attenuated and the amplitude ratio in the other direction becomes large, so that a gain g for emphasizing the sound in the other direction is given in a later gain determination circuit. A relatively suppressed gain is provided for the direction.

  In this embodiment, in the range of 270 deg to 90 deg, the specific amplitude ratio is set for each direction in the range of 270 deg to 15 deg and 15 deg to 90 deg. However, it can be set as a specific amplitude ratio for each direction in the range of 270 deg to 90 deg. Further, it may be set as a specific amplitude ratio for each direction in the range of 0 deg to 360 deg.

  In the present embodiment, a circuit in which the characteristic correction circuit 3a, the exchange circuit 3b, and the coefficient update circuit 3c are connected in series is used as the coefficient determination circuit 3, but the combined signal InA (k) and the combined signal InB (k As long as it is a circuit that determines the gain according to the difference between the waveforms of (), it can be realized by other circuits and methods without being limited to the above embodiment.

In the present embodiment, the combining circuit 4 multiplies the combined signal InA (k) and the combined signal InB (k) by a coefficient m (k) at an arbitrary ratio, and adds the arbitrary ratio to determine the gain g. To do. Examples of the gain g include the following (1) to (5).
(1) The coefficient m (k) as it is.
(2) A value obtained by multiplying a coefficient m (k) by a specified number, such as g = m (k) ^ 2 and g = m (k) ^ 3.
(3) A value obtained by adding a specified number to the coefficient m (k) such as g = m (k) + a and g = m (k) −a.
(4) A value obtained by multiplying the coefficient m (k) by a certain number b, such as g = m (k) * b.
(5) A value obtained by combining (2) to (4) as in g = ((m (k) + a) * b) ^ 2.

Further, the output signal OutL (k) and the output signal OutR (k) can be combined as in the following equations (16) and (17).
Output signal outL = (InM1 (k) + outL (k)) * g
Output signal outR = (InM2 (k) + outR (k)) * g (16)

Output signal outL = outR = (outL + outR) * g + InM1 (k) + InM2 (k) (17)

(Other embodiments)
As mentioned above, although embodiment of this invention was described, this embodiment is shown as an example and is not intending limiting the range of invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

  For example, in the embodiment, the sound source separation device has been described on the assumption that the sound source separation device is mounted on a device having a recording function such as an IC recorder, a PC, or a portable terminal. However, the sound source separation device can be mounted on any other audio device, Instead, the sound wave signals InM1 and InM2 can be received from the memory storing the sound wave data. For example, in addition to a sound wave signal input in real time from a microphone, a sound wave signal recorded in advance by a pair of microphones connected to a sound source separation device, a sound wave obtained in advance by a completely different pair of microphones This includes forming directivity in a specific direction with respect to a sound wave signal that is artificially generated by using a signal, a computer, or the like as a sound wave recorded by a pair of microphones.

As shown in FIG. 15, an amplitude ratio measuring unit 12 that measures the amplitude ratio of the combined signal can be provided at the subsequent stage of the amplitude ratio setting unit 1. The amplitude ratio measurement unit 12 stores in advance data of the ratio of the amplitude rate for each direction in the amplitude ratio setting unit 1 and measures the direction of the sound wave signal based on the amplitude ratio of a pair of newly input sound wave signals.
That is, the ratio data of the amplitude ratio for each direction in the amplitude ratio setting unit 1 when the sound wave signal InM1 (k) and the sound wave signal InM2 (k) are input in advance is recorded in the memory. Then, the amplitude ratio of the combined signal InA ′ (k) and the combined signal InB ′ (k) of the input sound wave signal, InM1 ′ (k) and the sound wave signal InM2 ′ (k) is measured. Then, the measured amplitude ratio is compared with the ratio of the amplitude change rate of the sound wave signal InM1 (k) and the sound wave signal InM2 (k) in the amplitude ratio setting unit 1 stored in the memory in the amplitude ratio measurement unit 12.
Since the ratio of the amplitude change rate of the sound wave signal InM1 (k) and the sound wave signal InM2 (k) is a unique value for each direction, the measured synthesized signal InA ′ (k) and synthesized signal InB ′ (k) , The direction of the sound source that generates the sound wave signal, InM1 ′ (k), and sound wave signal InM2 ′ (k) can also be known.

  Further, as shown in FIG. 16, the coefficient updating circuit multiplies one of the exchange signals by a coefficient m, generates an error signal of the exchange signal, and calculates a recurrence formula of the coefficient m including the error signal. If the coefficient m is updated for each sample, the present invention is not limited to the above embodiment and can be realized in other modes.

  The sound source separation device may be realized as software processing of a CPU or DSP, or may be configured by a dedicated digital circuit. When implemented as software processing, in a computer having a CPU, external memory, and RAM, a program describing the same processing contents as the amplitude ratio setting unit 1, amplitude changing unit 2, coefficient determination circuit 3, and synthesis circuit 4 is read into ROM Or stored in an external memory such as a hard disk or a flash memory, appropriately expanded in a RAM, and the CPU may perform calculations according to the program.

DESCRIPTION OF SYMBOLS 1 Amplitude ratio setting part 1a Adder 1b Adder 2 Amplitude change part 3 Coefficient determination circuit 3a Characteristic correction circuit 3b Exchange circuit 3c Coefficient update circuit 4 Synthesis circuit 5 Accumulator 6 Adder 7 Accumulator 8 Accumulator 9 Adder 10 Delay 11 Accumulator 12 Amplitude ratio measurement unit

Claims (19)

A sound source separation method for forming directivity in a specific direction with respect to a pair of sound wave signals,
And optional filter, Ru used and any number of adders provided in a subsequent stage,
An amplitude ratio setting step for changing the amplitude ratio of the pair of sound wave signals and setting the specific amplitude ratio for each direction;
An amplitude changing step of changing an amplitude ratio in the specific direction of the pair of sound wave signals so that the amplitude in the specific direction is the same ;
Including
A gain determining step for determining a gain according to a difference between waveforms of the pair of sound wave signals that have undergone the amplitude changing step;
An output step of multiplying the sound wave signal by the gain and outputting;
Including ,
A sound source separation method characterized by the above. In the amplitude ratio setting step,
2. The sound source separation method according to claim 1, wherein the amplitude of the pair of sound wave signals is changed at a specific rate of change depending on a direction, and the amplitude derived from the same direction is changed at a different rate of change. The polar pattern indicating the ratio of the change rate is
The sound source separation method according to claim 2, wherein the sound source separation method is different for each pair of sound wave signals. The rate of change of the amplitude is in a specific direction for emphasis or attenuation.
4. The sound source separation method according to claim 3, wherein the rate of change is larger than the amplitude rate in other directions. The gain determining step includes:
An exchange step for generating a pair of exchange signals by alternately exchanging the pair of sound wave signals for each sample by an exchange circuit for the sound wave signals that have undergone the amplitude changing step;
A step of generating an error signal of the exchange signal after multiplying one of the exchange signals by a coefficient m;
An update step of calculating a recurrence formula of the coefficient m including the error signal and updating the coefficient m every sample;
An output step of multiplying and outputting the sequentially updated coefficient m to the pair of sound wave signals;
Providing
The sound source separation method according to claim 4.   The sound source separation method according to claim 1, wherein the pair of sound wave signals is a pair of sound wave signals input from a pair of microphones.   6. The pair of sound wave signals according to any one of claims 1 to 5, wherein a system represented by a transmission coefficient when a system from a sound source to a microphone is simulated is convoluted. Sound source separation method. A sound source separation device that forms directivity in a specific direction with respect to a pair of sound wave signals,
An arbitrary number of filters and an arbitrary number of adders provided in the subsequent stage ,
An amplitude ratio setting unit that changes the amplitude ratio of the pair of sound wave signals and sets the specific amplitude ratio for each direction ;
An amplitude changing unit that changes an amplitude ratio of the pair of sound wave signals in the specific direction so that the amplitude in the specific direction is the same ;
With
A gain determining unit that determines a gain according to a difference between waveforms of the pair of sound wave signals that have passed through the amplitude changing unit ;
An output unit that multiplies the sound wave signal by the gain and outputs it,
Providing
A sound source separation device characterized by the above.   The amplitude ratio setting unit changes the amplitude of the pair of sound wave signals at a specific rate of change depending on the direction, and changes the amplitude derived from the same direction at a different rate of change. The sound source separation device according to 1. The polar pattern indicating the ratio of the change rate is
The sound source separation device according to claim 9, wherein the sound source separation device is different for each pair of sound wave signals. The rate of change of the amplitude is in a specific direction for emphasis or attenuation.
The sound source separation device according to claim 10, wherein the rate of change is larger than the amplitude rate in the other direction. The gain determining unit
An exchange unit that generates a pair of exchange signals by alternately exchanging the pair of sound wave signals for each sample by an exchange circuit with the sound wave signal that has passed through the amplitude changing unit;
A generator for generating an error signal of the exchange signal after multiplying one of the exchange signals by a coefficient m;
An updating unit that calculates a recurrence formula of the coefficient m including the error signal and updates the coefficient m for each sample;
An output unit for multiplying and outputting the pair of sound wave signals by the sequentially updated coefficient m;
Providing
The sound source separation device according to claim 11.   The sound source separation device according to claim 8, wherein the pair of sound wave signals is a pair of sound wave signals input from a pair of microphones.   13. The pair of sound wave signals is a system in which a system represented by a transfer coefficient when a system from a sound source to a microphone is simulated is convoluted. Sound source separation device. A sound source separation program that forms directivity in a specific direction with respect to a pair of sound wave signals,
Computer
Amplitude ratio setting means for changing the amplitude ratio of each sound wave signal in the pair of sound wave signals and generating a pair of sound wave signals having a specific amplitude ratio for each direction ;
Amplitude changing means for changing an amplitude ratio in the specific direction of the pair of sound wave signals so that the amplitude in the specific direction is the same ;
Gain determining means for determining a gain according to a difference between waveforms of the pair of sound wave signals that have passed through the amplitude changing means ;
Output means for multiplying the sound wave signal by the gain and outputting;
A sound source separation program characterized by being made to function. The amplitude ratio setting means changes the amplitude of the pair of sound wave signals at a specific rate of change depending on the direction, and changes the amplitude derived from the same direction at a different rate of change. The sound source separation program described in 1. The polar pattern indicating the ratio of the change rate is
The sound source separation program according to claim 16, wherein the sound source separation program is different for each pair of sound wave signals. The rate of change of the amplitude is in a specific direction for emphasis or attenuation.
The sound source separation program according to claim 17, wherein the rate of change is larger than the amplitude rate in the other direction. The sound wave signal that has passed through the amplitude changing means is generated by alternately exchanging the pair of sound wave signals for each sample by an exchange circuit, and generating a pair of exchange signals,
After multiplying one of the exchange signals by a coefficient m, an error signal of the exchange signal is generated,
Calculating a recurrence formula of the coefficient m including the error signal to update the coefficient m every sample;
Multiplying the pair of sound wave signals by the sequentially updated coefficient m, and outputting them.
The sound source separation program according to claim 18, which realizes a function.