JP5446745B2 - Sound signal processing method and sound signal processing apparatus - Google Patents

Description

  The present invention relates to noise suppression processing for sound signals, and more particularly to noise suppression processing for sound signals in accordance with the magnitude of noise.

  The noise suppression device discriminates a section including speech in the input sound signal and a section including only background noise (VAD determination), detects background noise power in the background noise section, and detects the detected noise in the sound section. The power corresponding to the background noise power can be suppressed.

  The microphone array uses an array of at least two microphones to process the sound signal received and converted, thereby limiting the sound receiving range to the sound source direction of the desired target sound or controlling the directivity, Noise suppression or target sound enhancement can be performed.

  In a certain known voice signal transmission device, the voice from the transmission position M and the noise from the noise source are input to each microphone through each path and further input to each delay unit. In this case, the delay amounts of the audio signals output from the delay devices are made equal, and the adder outputs the audio signals and noises of those delay amounts together. Since the noise has a small time correlation, the amplitude of the white noise is not tripled, and the amplitude of the voice signal is tripled, thereby improving the clarity of the voice.

  In one known conference hearing aid, the sound source direction is detected by the sound source direction detection means using the difference in arrival time of the sound waves input to the microphones at both ends of the microphone array, and the delay amount of each variable delay element is adjusted. The directivity is improved by making the outputs of the microphones in phase, and the gain of the amplification means is automatically adjusted by the AGC means each time the input sound source changes. As a result, background noise is suppressed, and only the target voice can be heard at an appropriate volume without distortion.

  A known sound source separation system generates a target sound dominant signal by performing linear combination processing for target sound enhancement using two microphones arranged side by side in the target sound arrival direction and these received signals. A target sound dominant signal generating means, a target sound inferior signal generating means for generating a target sound inferior signal by performing a linear combination process for suppressing the target sound using the received sound signal of the microphone, and a target sound dominant signal Separation means for separating the target sound and the interference sound using the spectrum and the spectrum of the target sound inferior signal. Thereby, the target sound and the interference sound coming from an arbitrary direction can be separated with high accuracy, and the apparatus can be miniaturized.

JP 07-38884 A Japanese Patent Laid-Open No. 09-140000 JP 2006-197552 A

  In the above-described noise suppression device, when the noise power is relatively large such that the speech component is buried in the noise component, the accuracy of the VAD determination tends to be low. On the other hand, in the above-described microphone array, there may be a difference in characteristics between the individual microphones at the time of shipment and due to aging. For example, since the microphone sensitivity varies, the phase difference between the microphones can be obtained theoretically. There may not be.

  The inventor has recognized that, when different suppression methods are applied according to the noise level of the input sound signal, higher voice quality with suppressed noise can be realized.

  An object of embodiments of the present invention is to achieve higher voice quality with noise suppressed.

According to one aspect of the embodiment of the present invention, a sound signal processing method in an information processing apparatus includes: a sound section of a first input sound signal in first and second input sound signals in a certain time period; Determining a noise interval; determining a power level of the first input sound signal in the noise interval; and determining a power level of the first input sound signal in the noise interval A step of determining whether or not the threshold value is greater than the threshold value; and if it is determined that the power of the first input sound signal in the noise interval is not greater than the first threshold value, the determination in the noise interval A step of suppressing noise in the speech section and the noise section of the first input sound signal by the first suppressor based on the magnitude of the power, and the first input sound in the noise section. Of signal power When it is determined that the pitch is not greater than the first threshold value, a phase difference between the first input sound signal and the second input sound signal is obtained in the sound section, and the first Obtaining an error between the theoretical phase difference between the input sound signal and the second input sound signal and the obtained phase difference, and the magnitude of the power of the first input sound signal in the noise interval Is determined to be larger than the first threshold, the phase difference between the first input sound signal and the second input sound signal is corrected according to the obtained error, and the second And a step of suppressing noise of the first input sound signal in accordance with the corrected phase difference.

  According to the embodiment of the present invention, it is possible to improve the voice quality after noise suppression of the sound signal.

FIG. 1 shows an example of a schematic configuration of a voice information device according to an embodiment of the present invention. FIG. 2 shows an example of a schematic configuration of a digital signal processor. FIG. 3 shows an example of the power spectrum of the voice section and the power spectrum of the noise section. FIG. 4 shows an example of a spatial arrangement of two microphones with respect to the target sound source. FIG. 5 shows an example of the suppression angle range of the microphone in FIG. 6 and 7 show examples of a phase difference and an error phase difference of an input signal regarding a certain frequency with respect to an angular direction of sound reception from a sound source. (Explained in Fig. 6) 8A to 8C show examples of flowcharts for noise suppression executed by two noise suppressors and a control unit. (Explained in Figure 8A) (Explained in Figure 8A)

  The objects and advantages of the invention will be realized and attained by means of the elements and combinations particularly pointed out in the appended claims.

  The foregoing general description and the following detailed description are exemplary and explanatory only and are not intended to limit the invention.

  Non-limiting embodiments of the present invention will be described with reference to the drawings. In the drawings, similar components are given the same reference numerals.

  FIG. 1 shows an example of a schematic configuration of a voice information device 10 according to an embodiment of the present invention.

  The audio information device 10 includes a microphone array device 100 including at least two microphones MIC1 and MIC2, a digital signal processor (DSP) 200, and a utilization application unit 400. The voice information device 10 may be an information device such as an in-vehicle device or a car navigation device having a voice recognition function, a hands-free phone, or a mobile phone.

  The microphone array apparatus 100 includes microphones MIC1, MIC2, amplifiers (AMP) 122, 124, low-pass filters (LPF) 142, 144, and analog-digital converters (A / A) as sound receiving units or sound signal input units. D) 162 and 164 are provided. A digital signal processor (DSP) 200 is coupled to a memory 202 including, for example, a RAM or the like.

  Analog input signals ina1 and ina2 converted from sound waves by the microphones MIC1 and MIC2 are supplied to amplifiers (AMP) 122 and 124, respectively, and are amplified by the amplifiers 122 and 124. The amplified analog sound signals INa1 and INa2 output from the amplifiers 122 and 124 are respectively coupled to inputs of low-pass filters 142 and 144 having a cutoff frequency fc (for example, 3.9 kHz), respectively. Low pass filtered. Instead of the low-pass filter, for example, a band-pass filter having a pass frequency band of 0.4 to 3.9 kHz may be used.

  The filtered analog signals INp1, INp2 at the outputs of the low-pass filters 142, 144 are respectively coupled to the inputs of analog-to-digital converters 162, 164 with a sampling frequency fs (eg, 8 kHz) (fs> 2fc), Converted to a digital input signal. The time domain digital input signals IN1 (t) and IN2 (t) from the analog-to-digital converters 162 and 164 are respectively coupled to input terminals it1 and it2 as sound signal input units of the digital signal processor (DSP) 200. Is done.

  FIG. 2 shows an example of a schematic configuration of the digital signal processor 200.

  Digital signal processor 200 includes input buffer memories 212 and 214 coupled to input terminals it1 and it2, respectively, a noise suppressor (NS) 220 for microphone MIC1, a control unit or mode switching unit 230, and A noise suppressor (NS) 260 for the two microphones MIC1 and MIC2 is included. Input buffer memories 212 and 214 are coupled to microphone array apparatus 100 (microphones MIC1, MIC2), receive their digital input signals IN1 (t), IN2 (t), and buffer them for noise suppression processing. Ring. Input buffer memories 212 and 214 may be memory areas of memory 202.

  The noise suppressor 220 is coupled to the input buffer memory 212 that buffers the digital input signal IN1 (t), and is coupled to the control unit 230, the noise suppressor 260, and the output side switch SW_O. The noise suppressor 220 includes a voice interval detection unit (VAD) 222, a power determination unit or power detection and estimation unit 224, an input side switch SW_I1, and a suppression unit 226 that suppresses noise power. The voice section detection unit (VAD) 222 identifies, for example, a voice section in the input sound signal or a non-speech section (noise section) for a time period of 20 ms by a known VAD (Voice Activity Detection) method. .

  The noise suppressor 260 is coupled to the input buffer memories 212 and 214 for buffering the digital input signals IN1 (t) and IN2 (t), respectively, and is coupled to the control signal CTL of the control unit 230 and the output side switch SW_O. ing. The noise suppressor 260 includes a phase difference determination unit 262, an error phase difference determination unit 264, an input side switch SW_I2, a phase difference correction unit 266, a direction determination unit 268, and a suppression unit 270 that suppresses noise power in the noise angle direction. It is out.

  The power determination unit 224 of the noise suppressor 220 detects the average power level of the noise of the digital input signal IN1 (t) in the non-voice section from the voice section detection unit 222, and the detected magnitude of the noise average power. Is supplied to the control unit 230. As an indication of the magnitude of power, the average amplitude may be used without being limited to power.

  The control unit 230 performs switching between the noise power suppression mode and the noise direction suppression mode for each time interval of the speech segment and the non-speech segment and according to the magnitude of the noise (N) power in the non-speech segment. A control signal CTL is generated. The control unit 230 converts the control signal CTL into a power determination unit (or power detection and estimation unit) 224, a phase difference determination unit 262, an error phase difference determination unit 264, input side switches SW_I1 and SW_I2, and output side switch SW_O. To supply. The control unit 230 may further supply the control signal CTL to the suppression units 226 and 270, the phase difference correction unit 266, and the direction determination unit 268.

  In the noise suppressor 220, the voice interval detector (VAD) 222 converts the time domain digital input signal IN1 (t) into a frequency domain digital input signal or a complex spectrum IN1 (f) by, for example, Fourier transform. The voice segment detection unit 222 identifies a voice segment and a non-speech segment (noise segment, silence segment) based on the characteristics of the power distribution, and uses the power determination unit 224 to identify identification information of the voice segment and the non-speech segment. This is supplied to the control unit 230, the phase difference determination unit 262, and the error phase difference determination unit 264.

  FIG. 3 shows an example of the power spectrum of the voice section and the power spectrum of the noise section. The power spectrum of the speech section has a non-uniform distribution and a relatively high regularity (small entropy). The power spectrum of the noise section has a substantially uniform distribution over the entire frequency, and is relatively low in regularity (high entropy). A voice interval and a non-voice interval are identified using such a distribution difference. Further, for example, a pitch (harmonic) characteristic peculiar to voice or a formant distribution characteristic may be obtained and identified.

  The power determination unit 224 estimates the noise power distribution based on the detected noise power distribution with respect to the pass frequency band in the noise power suppression mode, and supplies the estimated noise power distribution to the suppression unit 226. The power spectrum in the noise section in FIG. 3 may be estimated as the distribution of the noise power component with respect to the frequency. In the noise power suppression mode, the suppression unit 226 subtracts the estimated noise power component from the input signal IN1 (t), thereby suppressing the noise component and switching the audio signal INns (t) from which the noise has been removed to the switch SW_O. To supply.

FIG. 4 shows an example of a spatial arrangement of two microphones MIC1 and MIC2 with respect to the target sound source SS.
FIG. 5 shows an example of suppression angle ranges + α to π to (2π−α) of the microphones MIC1 and MIC2 in FIG.

  In FIG. 4, generally, a plurality of microphones MIC1, MIC2,. . . Are arranged at a known distance d from each other on a straight line. Here, as a typical example, it is assumed that two adjacent microphones MIC1 and MIC2 are arranged on a straight line at a distance d from each other. The distance d between adjacent microphones only needs to satisfy the sampling theorem.

  In FIG. 4, the target sound source SS is on a straight line connecting the microphones MIC1 and MIC2, the target sound source is on the left front surface of the microphone MIC1, and the direction of the target sound source SS is the sound receiving direction or the target direction of the microphone arrays MIC1 and MIC2. And Typically, the sound source SS for receiving sound is the speaker's mouth, and the sound receiving direction is the direction of the speaker's mouth. A sound reception angle range (−α to 0 to + α) is provided in the vicinity of the sound reception angle direction (0). A range (+ α to π to (2π−α)) other than the sound receiving angle range (−α to 0 to + α) may be set as the noise suppression angle range.

  The distance d between the microphones MIC1 and MIC2 is preferably set so as to satisfy the condition of distance d <sound speed c / sampling frequency fs so as to satisfy the sampling theorem. The input sound signal received and processed by the microphone arrays MIC1 and MIC2 depends on the incident angle θ (= −α to 0 + α) of the sound wave with respect to the straight line passing through the microphone arrays MIC1 and MIC2, and is perpendicular to the straight line. It does not depend on the incident direction (0 to 2π) in the radial direction on a flat surface.

  The sound of the target sound source SS is detected by the right microphone MIC2 with a delay time τ = d / c from the left microphone MIC1. The sound from the sound source in the direction of the angle θ on the left front is detected by the right microphone MIC2 with a delay time τ = d · sin (π / 2−θ) / c from the left microphone MIC1 ( −π / 2 ≦ θ ≦ + π / 2). On the other hand, the noise N in the suppression angle range is detected by the left microphone MIC1 with a delay time τ = d / c from the right microphone MIC2. Noise from the sound source in the angle θ direction on the right rear surface is detected by the left microphone MIC1 with a delay time τ = d · sin (θ−π / 2) / c from the right microphone MIC2. + Π / 2 ≦ θ ≦ + π).

  In the noise suppressor 260, the phase difference determination unit 262 obtains the phase difference PD between the time domain digital input signals IN1 (t) and IN2 (t). The phase difference PD may be obtained based on the time difference between temporal power or amplitude changes of the two digital input signals IN1 (t) and IN2 (t) for each frequency f. As an alternative, the phase difference PD may be obtained based on the phases of the two digital input signals IN1 (t) and IN2 (t) in the frequency domain for each frequency f.

  In order to obtain the phase difference PD, the time domain digital input signals IN1 (t) and IN2 (t) are converted into frequency domain digital input signals or complex spectra IN1 (f) and IN2 (f) by, for example, Fourier transform. May be.

In order to process the input signal in the frequency domain, the time domain digital input signals IN1 (t), IN2 (t) in the input buffer memories 212 and 214 are fed to the fast Fourier transform (FFT) of the digital signal processor 200. Supplied respectively. In a known form, the fast Fourier transformer multiplies each signal section of the digital input signals IN1 (t) and IN2 (t) by an overlap window function and Fourier-transforms or orthogonally transforms the product to obtain a frequency domain. The complex spectra IN1 (f) and IN2 (f) are generated. Where IN1 (f) = A ₁ e ^{j (2πft + φ1 (f))} , IN2 (f) = A ₂ e ^{j (2πft + φ2 (f))} , f is frequency, A ₁ and A ₂ are amplitude, j is unit The imaginary numbers, φ1 (f) and φ2 (f) are delay phases that are a function of frequency f. As the overlap window function, for example, a Hamming window function, a Hanning window function, a Blackman window function, a 3 sigma gauss window function, or a triangular window function can be used.

The phase difference determination unit 262 includes a phase difference PD (f) of a phase spectrum component indicating a sound source direction for each frequency f (0 <f <fs / 2) between two adjacent microphones MIC1 and MIC2 separated by a distance d. ) (Radian, rad) is obtained by the following equation.
PD (f) = tan ⁻¹ (IN1 (f) / IN2 (f))
= Tan ⁻¹ (J {IN1 (f) / IN2 (f)} / R {IN1 (f) / IN2 (f)})
Here, a sound source having a specific frequency f or a sound source is approximated as having only one sound source. J {x} represents the imaginary component of the complex number x, and R {x} represents the real component of the complex number x.

This phase difference PD (f) is expressed as follows by the delay phases (φ1 (f), φ2 (f)) of the digital input signals IN1 (t) and IN2 (t).
PD (f) = tan ⁻¹ (J {(A ₁ e ^{j (2πft + φ1 (f))} / A ₂ e ^{j (2πft + φ2 (f))} } / R {(A ₁ e ^{j (2πft + φ1 (f))} / A ₂ e ^{j (2πft + φ2 (f))} })
= Tan ⁻¹ (J {(A ₁ / A ₂ ) e ^{j (φ1 (f) −φ2 (f))} } / R {(A ₁ / A ₂ ) e ^{j (φ1 (f) −φ2 (f) )} })
= Tan ⁻¹ (J {e ^{j (φ1 (f) −φ2 (f))} } / R {e ^{j (φ1 (f) −φ2 (f))} })
= Tan ⁻¹ (sin (φ1 (f) −φ2 (f)) / cos (φ1 (f) −φ2 (f)))
= Tan ⁻¹ (tan (φ1 (f) −φ2 (f))
= Φ1 (f) −φ2 (f)

  6 and 7 show examples of the phase difference PD and the error phase difference ΔPD of the input signal related to a certain frequency f with respect to the angular direction θ of the sound received from the sound source. 6 and 7, the theoretical phase difference is exemplified by a solid line, and the actually measured phase difference is exemplified by a broken line.

  6 and 7, in the angular directions θ = 0 to + π / 2 and + 3π / 2 to 2π of the sound source, the phase difference PD has a positive value, that is, indicates a leading phase. In the angle direction θ = + π / 2 to + 3π / 2 of the sound source, the phase difference PD has a negative value, that is, indicates a delayed phase. In the angular directions θ = + π / 2 and + 3π / 2 of the sound source, the phase difference PD has 0, that is, shows the same phase.

  In FIG. 6, the phase difference PD between the actually measured input signals IN1 (t) and IN2 (t) is a desirable threshold value of the phase difference corresponding to the angle boundary α of the suppression angle range due to variations in sensitivity of the microphones MIC1 and MIC2. It may have a phase difference lower than Dth. In this case, the input signal IN1 (t) having the phase difference PD lower than the threshold value Dth is not desirable and is all suppressed.

  In FIG. 7, the phase difference PD between the actually measured input signals IN1 (t) and IN2 (t) is the phase difference threshold Dth corresponding to the angle boundary α of the suppression angle range due to variations in sensitivity of the microphones MIC1 and MIC2. May have a higher phase difference. In this case, the suppression angle range becomes undesirably narrow, that is, the input signal IN1 (t) having a phase difference PD higher than the threshold value Dth has a narrow suppression angle range, and noise suppression becomes insufficient.

  On the other hand, in the speech section determined by the speech section detection unit 222, it is expected that the sound is received from the direction of the target sound source SS, and thus the phase difference PD is considered to have a theoretical maximum positive value. Therefore, the difference between the actually measured phase difference PD and the theoretical phase difference PD is considered to be the error phase difference ΔPD.

  The phase difference determining unit 262 is configured to output a phase difference PD for each frequency f between two adjacent input signals IN1 (t) and IN2 (t), or a phase difference PD of the maximum power frequency or an upper limit frequency (for example, 3.8 kHz). ) Is supplied to the error phase difference determination unit 264 and the phase difference correction unit 266.

  In the noise power suppression mode, the error phase difference determination unit 264 is the difference between the theoretical phase difference PD and the actually measured phase difference PD from the phase difference determination unit 262 in the speech segment determined by the speech segment detection unit 222. Is obtained as an error phase difference ΔPD. The error phase difference determination unit 264 holds the obtained error phase difference ΔPD in the noise power suppression mode and supplies it to the phase difference correction unit 266 in the subsequent noise direction suppression mode. The phase difference correction unit 266 corrects or compensates the phase difference PD from the phase difference determination unit 262 by the error phase difference ΔPD from the error phase difference determination unit 264 in the noise direction suppression mode, and corrects the corrected phase difference PDc. Is generated.

  The direction determination unit 268 determines the corrected phase difference PDc based on the threshold value Dth in the noise direction suppression mode, and has an input signal having a phase difference PDc equal to or less than the desired threshold value Dth corresponding to the angle boundary α of the suppression angle range. Are determined as noise intervals. In addition, the direction determination unit 268 determines that the time interval of the input signal having the phase difference PDc larger than the desirable threshold value Dth corresponding to the angle boundary α of the suppression angle range is the audio interval. The direction determination unit 268 supplies the speech section or the noise section identification information to the suppression unit 270 in the noise direction suppression mode. In the noise direction suppression mode, the suppression unit 270 suppresses the input signal IN1 (t) or IN1 (f) in the noise interval by attenuating. The suppression unit 270 passes the input signal IN1 (t) or IN1 (f) in the speech section in the noise direction suppression mode. In the noise direction suppression mode, the suppression unit 270 supplies the audio signal INns (t) from which the noise has been removed to the switch SW_O (T1).

  The noise suppressors 220 and 260 may perform the processing for noise suppression for each frequency in the frequency domain. In this case, the suppression units 226 and 270 of the noise suppressors 220 and 260 further convert the processed digital input signal INns (f) in the frequency domain into a digital audio signal INns (t) in the time domain by, for example, inverse Fourier transform. And the noise-suppressed digital sound signal INns (t) is supplied to the switch SW_O.

  The output of the digital sound signal INns (t) is used, for example, for voice recognition or a mobile phone call. The digital sound signal INns (t) is supplied to a subsequent application 400 where, for example, it is digital-analog converted by a digital-analog converter 404 and low-pass filtered by a low-pass filter 406 to generate an analog signal. Or stored in the memory 414 and used by the voice recognition unit 416 for voice recognition. The speech recognition unit 416 may be a processor implemented as hardware, or a processor that operates according to a program stored in a memory 414 including, for example, a ROM and a RAM implemented as software.

  The digital signal processor 200 may be a signal processing circuit implemented as hardware or a signal processing circuit that operates according to a program stored in a memory 202 including, for example, ROM and RAM, implemented as software. .

  8A to 8C show examples of flowcharts for noise suppression executed by the two noise suppressors 220 and 260 and the control unit 230. FIG.

  Referring to FIG. 8A, in step 802, the voice interval detector 222 receives from the input buffer memory 212 the digital input signal IN 1 (t) of the current time interval extracted from the microphone MIC 1.

  In step 804, the voice interval detector 222 determines whether the digital input signal IN1 (t) in the time interval is a voice interval, a non-voice interval, or a noise interval by a VAD (Voice Activity Detection) method. The voice section detection unit 222 supplies the identification information of the voice section and / or the non-voice section (noise section) for the time section of the digital input signal IN1 (t) to the control unit 230 and the power determination unit 224.

  In step 806, the power determination unit 224 extracts the digital input signal IN1 (t) from the input buffer memory 212, and the noise power (or amplitude) of the digital input signal IN1 (t) in the determined noise section. Detect the size. The power determination unit 224 supplies the detected magnitude of the noise power to the control unit 230.

  In step 808, the control unit 230 determines whether or not the magnitude of the noise power is greater than a threshold Pth for noise suppression. As the magnitude of the noise power, for example, an average value of power or amplitude may be used. The threshold value Pth may be set to a value that is sufficiently smaller than the desired amount of voice power or higher than the expected general background noise. If it is determined that the magnitude of the noise power is greater than the threshold value Pth, the procedure proceeds to Step 832 in FIG. 8C. When it is determined that the magnitude of the noise power is equal to or less than the threshold value Pth, the procedure proceeds to Step 812 in FIG. 8B.

When the magnitude of the noise power of the input signal IN1 (t) in the noise section is not larger than the threshold value Pth, the input signal in the noise section is suppressed and the noise power component is subtracted from the input signal power including the voice power component. Thus, the voice quality of the input signal can be improved. When the magnitude of the noise power of the input signal IN1 (t) in the noise section is larger than the threshold value Pth or not sufficiently smaller than the voice component, it is difficult to distinguish voice and noise by the magnitude of power. In this case, if the noise power component is subtracted from the input signal power including the audio power component, the audio power component is too small, and the audio quality of the input signal may be degraded. Therefore, noise may be suppressed based on the phase difference between the two input signals or the sound source direction.

  Referring to FIG. 8B, in step 812, the control unit 230 sets the noise suppressor 220 to the noise power suppression mode in at least the current noise period or a time period including the current noise period and the subsequent voice period. Therefore, the control unit 230 sets the switch SW_I1 of the noise suppressor 220 to the on state (T1) and sets the switch SW_I2 of the noise suppressor 260 to the off state (T1).

  In step 814, the power determination unit 224 estimates noise power based on the detected noise power in the determined noise section, and supplies the estimated noise power to the suppression unit 226. The estimated noise power may be, for example, an average value of the detected noise power.

  In step 816, the suppression unit 226 suppresses the noise power (component) in the noise interval by attenuating the power of the digital input signal IN1 (t) to zero (0). As an alternative, the power of the input signal IN1 (t) may be attenuated at a certain rate, for example 1/10. As a result, it is possible to reduce the risk of erasing the voice when the voice is buried in the noise power. Further, the suppression unit 226 suppresses the noise component in the speech section by subtracting the estimated noise power (component) for each frequency from the digital input signal IN1 (t).

  In step 818, the phase difference determination unit 262 receives the identification information of the voice segment (V) and the non-speech segment (N) from the voice segment detection unit 222, and determines whether the voice segment (V) is detected. If it is determined that a voice segment has been detected, the procedure proceeds to step 820. If it is determined that no voice segment has been detected, the procedure proceeds to step 824.

  In step 820, the phase difference determination unit 262 takes out the digital input signals IN1 (t) and IN2 (t) in the voice section from the input buffer memories 212 and 214, and the voice digital input signal IN1 (t) in the voice section. And the phase difference between IN2 (t). In the voice section, it is considered that the sound from the target sound source SS in the direction θ = 0 is picked up by the microphones MIC1 and MIC2.

  In step 822, the error phase difference determination unit 264 determines the determined phase difference PD between the audio digital input signals IN1 (t) and IN2 (t) and the corresponding theoretical phase difference for the audio from the target sound source direction. An error (difference) between them is obtained by comparing with PD. For example, the error phase difference ΔPD may be obtained by subtracting the theoretical phase difference PD from the determined phase difference PD. The error phase difference determination unit 264 supplies the determined error to the phase difference correction unit 266 as an error phase difference ΔPD. This error phase difference ΔPD can be used later in step 836.

  In step 824, the control unit 230 connects the output side switch SW_O to the output of the suppression unit 226 of the noise suppressor 220, and the noise-suppressed output audio signal INns (t) from the suppression unit 226 (in step 816). Is output.

  In step 852, the voice interval detector 222 determines whether or not the input buffer memory 212 has a digital input signal IN 1 (t) for the time interval to be processed at present. If it is determined that there is a digital input signal IN1 (t) for such a time interval, the procedure returns to step 814. If it is determined that there is no digital input signal for such a time interval, the procedure exits the routine of FIGS. The flowcharts of FIGS. 8A-8C are executed again for the digital input signal in the new time interval.

  Steps 814-824 and 852 of FIG. 8B may be repeated until the determination of step 808 of FIG. 8A is then made and thereafter step 832 of FIG. 8C is executed.

  Thus, when the magnitude of the noise power of the digital input signal in the noise interval is not larger than the threshold value Pth, the noise is suppressed by the noise suppressor 220 based on the noise power.

  Referring to FIG. 8C, in step 832, the control unit 230 sets the noise suppressor 260 to the noise direction suppression mode in at least the current noise period or a time period including the current noise period and the subsequent voice period. Therefore, the control unit 230 sets the switch SW_I2 of the noise suppressor 260 to the on state (T2) and sets the switch SW_I1 of the noise suppressor 220 to the off state (T2).

  In step 834, the phase difference determination unit 262 extracts the digital input signals IN1 (t) and IN2 (t) from the input buffer memories 212 and 214, and the audio digital input signal IN1 (in the noise interval or the audio interval time interval). The phase difference PD between t) and IN2 (t) is obtained.

  In step 836, the phase error correction unit 264 corrects or compensates the phase difference PD obtained by the phase difference determination unit 262 using the error phase difference ΔPD received from the error phase difference determination unit 264 in step 822. The phase difference PDc thus generated is generated (FIGS. 6 and 7). Therefore, the error phase difference ΔPD or a certain ratio α (for example, α = 1 or 0.8) of the error phase difference ΔPD may be subtracted from the obtained phase difference PD (PD−ΔPD × α). As a result, errors that can be corrected for phase difference errors can be reduced.

  In step 838, the direction determination unit 268 determines whether the direction of the sound source in the time interval is spatially within the sound reception angle range (−α to 0 + α) based on the corrected phase difference PDc, or It is determined whether or not the corrected phase difference PDc exceeds the threshold value Dth. In this case, each time interval may be one speech interval or noise interval, or may be a shorter time interval. When the phase difference PDc exceeds the threshold value Dth, the direction determination unit 268 determines that the direction of the sound source is spatially within the sound reception angle range (−α to 0 ++ α). When the phase difference PDc does not exceed the threshold value Dth, the direction determination unit 268 determines that the direction of the sound source is spatially within the suppression angle range (+ α to π to (2π−α)).

  In Steps 840 to 844, the suppression unit 270 determines whether the direction of the sound source is a sound reception angle range (−α to 0 to + α) or a phase difference within a sound reception angle range (−α to 0 to + α). Let IN1 (t) pass. Further, the suppression unit 270 has a direction of the sound source in the suppression angle range (+ α to π to (2π−α)) or the phase difference PDc in the suppression angle range (+ α to π to (2π−α)). When there is a phase difference, the power of the digital input signal IN1 (t) is attenuated to zero (0). In this way, the power of the digital input signal IN1 (t) in the time interval in which the direction of the sound source is in the suppression angle range (+ α to π to (2π−α)) is suppressed.

  In step 840, the suppression unit 270 determines whether the determined sound source direction is within the sound reception angle range (−α to 0 to + α), or the phase difference PDc is within the suppression angle range (+ α to π to (2π−α). ) Is determined. If it is determined that it is within the sound receiving angle range (−α to 0 to + α), the procedure proceeds to step 842. If it is determined that it is not within the sound reception angle range (−α to 0 to + α), the procedure proceeds to step 844.

  In step 842, the suppression unit 270 passes the digital input signal IN1 (t). Thereafter, the procedure proceeds to Step 850.

  In step 844, the suppression unit 270 attenuates and suppresses the digital input signal IN1 (t) to zero (0). As an alternative, the power of the input signal IN1 (t) may be attenuated at a certain rate, for example 1/10. As a result, it is possible to reduce the risk of erasing the voice when the voice is buried in the noise power. Thereafter, the procedure proceeds to Step 850.

  In step 850, the control unit 230 connects the output side switch SW_O to the suppression unit 270 of the noise suppressor 260 and outputs the noise-suppressed output audio signal INns (t) from the suppression unit 270. Thereafter, the procedure proceeds to Step 842.

  Step 852 is similar to that of FIG. 8B. If it is determined that there is a digital input signal IN1 (t) for the current time interval to be processed, the procedure returns to step 834 in FIG. 8A. The flowcharts of FIGS. 8A-8C are executed again for the digital input signal in the new time interval.

  Steps 834-850 and 852 of FIG. 8C may be repeated until the determination of step 808 of FIG. 8A is then made and thereafter step 812 of FIG. 8B is executed.

  Thus, when the magnitude of the noise power of the digital input signal in the noise interval is larger than the threshold value Pth, the noise is suppressed by the noise suppressor 260 based on the phase difference between the two digital input signals or the sound source direction. .

  In this way, the noise-suppressed output audio signal INns (t) from the suppression unit 226 or the suppression unit 270 is output and further supplied to the usage application unit 400.

  All examples and conditional expressions given here are intended to help the reader understand the inventions and concepts that have contributed to the promotion of technology, such examples and Interpretation is not limited to conditions. Also, the organization of such examples in the specification is not related to showing the superiority or inferiority of the present invention. While embodiments of the present invention have been described in detail, various changes, substitutions and variations can be made thereto without departing from the spirit and scope of the present invention.

200 Digital Signal Processor 220 Noise Suppressor for One Microphone 222 Voice Section Detection Unit (VAD)
224 Power determination unit 226 Suppression unit 230 Control unit 260 Noise suppressor for 2 microphones 262 Phase difference determination unit 264 Error phase difference determination unit 266 Phase difference correction unit 268 Direction determination unit 270 Suppression unit SW_I1, SW_I2 Input side switch SW_O Output Side switch

Claims (4)

A sound signal processing method in an information processing apparatus,
Determining a speech section and a noise section of the first input sound signal in the first and second input sound signals in a certain time period;
Determining the magnitude of power of the first input sound signal in the noise interval;
Determining whether the magnitude of the power of the first input sound signal in the noise interval is greater than a first threshold;
When it is determined that the power level of the first input sound signal in the noise section is not larger than the first threshold, based on the determined power level in the noise section, A step of suppressing noise in the voice section and the noise section of the first input sound signal by one suppression unit;
When it is determined that the power level of the first input sound signal in the noise section is not larger than the first threshold value, the first input sound signal and the second input in the voice section Obtaining a phase difference between the sound signals and obtaining an error between the theoretical phase difference between the first input sound signal and the second input sound signal and the obtained phase difference;
When it is determined that the power level of the first input sound signal in the noise section is larger than the first threshold value, the first input sound signal and the second input sound signal are between the first input sound signal and the second input sound signal. Correcting a phase difference according to the obtained error, and suppressing noise of the first input sound signal according to the corrected phase difference by a second suppression unit;
A sound signal processing method including: The step of suppressing noise of the first input sound signal in accordance with the corrected phase difference,
Determining whether the phase difference between the first input sound signal and the second input sound signal is greater than a second threshold;
If the phase difference between the is determined to be greater than the second threshold value, and supplies the first input sound signal to the output unit, when the phase difference is not greater than said second threshold value The sound signal processing method according to claim 1, wherein the second suppression unit includes suppressing noise of the first input sound signal. First and second sound signal input units for receiving first and second input sound signals, respectively;
A section determination unit for determining a voice section and a noise section of the first input sound signal received from the first sound signal input unit in a certain time period received;
A power determination unit that determines the magnitude of the power of the first input sound signal in the noise section;
A control unit for determining whether the magnitude of the power of the first input sound signal in the noise section is greater than a first threshold;
When it is determined that the power level of the first input sound signal in the noise section is not greater than the first threshold, based on the determined power level in the noise section, A first suppression unit that suppresses noise in the speech section and the noise section of the first input sound signal;
When it is determined that the power level of the first input sound signal in the noise section is not larger than the first threshold value, the first input sound signal and the second input in the voice section An error phase difference determination unit that obtains a phase difference between sound signals and obtains an error between the theoretical phase difference between the first input sound signal and the second input sound signal and the obtained phase difference. When,
When it is determined that the power level of the first input sound signal in the noise section is larger than the first threshold value, the first input sound signal and the second input sound signal are between the first input sound signal and the second input sound signal. A second suppression unit that corrects a phase difference according to the obtained error, and suppresses noise of the first input sound signal according to the corrected phase difference ;
A sound signal processing apparatus. The second suppressor is
Determining whether the phase difference between the first input sound signal and the second input sound signal is greater than a second threshold;
If the phase difference between the is determined to be greater than the second threshold value, and supplies the first input sound signal to the output unit, when the phase difference is not greater than said second threshold value The second suppressor suppresses the noise of the first input sound signal.
The sound signal processing apparatus according to claim 3 .

Images