JP6528438B2 - Sound signal processing apparatus and sound signal processing program - Google Patents

Description

  The present invention relates to a sound signal processing apparatus and a sound signal processing program, and can be applied to, for example, suppressing noise components included in a sound signal such as a sound signal or an acoustic signal.

  In a voice communication system represented by a cellular phone system, there are many cases where ambient noise components are superimposed on an input voice signal, which has caused deterioration of the speech quality. In order to solve such problems, various methods for suppressing noise components contained in an input speech signal have been conventionally proposed. For example, Non-Patent Document 1 describes a method of suppressing noise components by the Wiener filter method. The Wiener filter method is a method of suppressing noise components by estimating noise components in the input speech signal and subtracting the noise components from the input speech signal in the frequency domain.

Institute of Electronics, Information and Communication Engineers Knowledge Base, 1 group 5 volume 5 chapter 5-1

  However, when the signal-to-noise ratio (S / N ratio) of the input speech signal is poor, when the Wiener filter method is applied, the noise component is suppressed, and as a result, the component of the target voice (hereinafter referred to as target sound) Also had the problem of deterioration. There is a problem that noise components can not be sufficiently suppressed if deterioration of the target sound component is suppressed in order to avoid this problem. That is, when the S / N ratio of the input voice signal is poor, there is a problem that the degradation of the speech quality can not be improved even if the noise component suppression processing is performed.

  There are various noise reduction methods other than the Wiener filter method, such as a spectral subtraction method and a voice switch, but the same problem occurs when the S / N ratio of the input speech signal is poor.

  By the way, when applying a sound producing device divided into a speaker unit for the right ear such as a headset and a speaker unit for the left ear as an audio output device, an audio signal for the R channel and an audio signal for the L channel In the prior art, it is general to convert a monaural sound signal subjected to noise suppression processing into a stereo sound signal, and the above-mentioned problems have occurred as well.

  The present invention has been made in view of the above points, and an object of the present invention is to provide a sound signal processing device and a sound signal processing program capable of making it easy to listen to a target sound in an input sound signal.

A sound signal processing apparatus according to a first aspect of the present invention includes: (1) a branch processing unit that bifurcates an input sound signal in which a target sound component and a noise component are mixed to generate first and second branched sound signals; (2) An S / N ratio change processing unit that generates first and second S / N ratio change sound signals by changing the S / N ratio for the first and second branch signals , (3) the first and synchronizes the second S / N ratio changing sound signal, one of them have a synchronized unit for outputting L-channel signal, the other as R-channel signal, the S / N ratio changing The processing unit suppresses the noise component included in the first branched sound signal by the suppression amount defined by the first suppression amount parameter to obtain the S / N ratio in the first branched sound signal. The noise component included in the second branched sound signal is changed to a second one different from the first suppression amount parameter. By suppressing only the suppression amount suppression amount parameter defines, and changes the S / N ratio in the second branch sound signal.

A sound signal processing apparatus according to a second aspect of the present invention comprises: (1) a branch processing unit that branches an input sound signal in which a target sound component and a noise component are mixed into two to generate first and second branched sound signals; (2) an S / N ratio change processing unit that generates a first S / N ratio change sound signal by changing the S / N ratio with respect to the first branch signal ; (3) the first the S / N ratio changing sound signal and to synchronize the said second branch sound signal, have a synchronized unit which outputs one L-channel signal, the other as R-channel signal, (4) the S / The N ratio change processing unit suppresses S / N in the first branched sound signal by suppressing the noise component included in the first branched sound signal by the suppression amount defined by the predetermined suppression amount parameter. It is characterized by changing the ratio .

A sound signal processing program according to a third aspect of the present invention comprises: (1) branching processing for dividing an input sound signal in which a target sound component and a noise component are mixed into two to generate the first and second divided sound signals And (2) S / N ratio change processing for generating the first and second S / N ratio change sound signals by changing the S / N ratio with respect to the first and second branched signals. means, (3) to synchronize the first and second S / N ratio changing sound signal, one of the L-channel signal and the other to function as a synchronization means to output as R-channel signal, ( 4) The S / N ratio changing processing means suppresses the noise component contained in the first branched sound signal by the suppression amount defined by the first suppression amount parameter, thereby the first branch. The S / N ratio in the sound signal is changed, and the noise component contained in the second branched sound signal is By suppressing only suppression amount prescribed different second suppression amount parameter from the first suppression amount parameter, and changing the S / N ratio in the second branch sound signal.

A sound signal processing program according to a fourth aspect of the present invention comprises: (1) branching processing for dividing an input sound signal in which a target sound component and a noise component are mixed into two to generate first and second divided sound signals Means, (2) S / N ratio change processing means for generating a first S / N ratio change sound signal by changing the S / N ratio with respect to the first branched signal , (3) by synchronizing the said first S / N ratio changing sound signal and the second branch sound signal, one of the L-channel signal, the other one functioning as the synchronization means to output a R-channel signal, (4 The S / N ratio changing processing means suppresses the noise component contained in the first branched sound signal by the suppression amount defined by a predetermined suppression amount parameter, thereby the first branched sound signal It characterized that you change the S / N ratio in.

  According to the present invention, it is possible to realize a sound signal processing device, a sound signal processing program, and a sound signal processing method that can easily make the user listen to a target sound in an input sound signal.

It is a block diagram showing composition of a noise suppression device concerning a 1st embodiment. It is explanatory drawing which shows the signal waveform of the input-output signal of the noise suppression apparatus which concerns on 1st Embodiment. It is a block diagram which shows the structure of the noise suppression apparatus which concerns on 2nd Embodiment. It is a block diagram which shows the structure of the noise suppression apparatus which concerns on 3rd Embodiment.

(A) First Embodiment Hereinafter, a first embodiment of a sound signal processing device and a sound signal processing program according to the present invention will be described with reference to the drawings. The sound signal processing device and the sound signal processing program according to the first embodiment are a noise suppression device and a noise suppression program for suppressing a noise component contained in an input sound signal.

(A-1) Configuration of First Embodiment FIG. 1 is a block diagram showing a configuration of a noise suppression apparatus according to the first embodiment.

  The noise suppression device according to the first embodiment can also be configured by hardware as the component shown in FIG. 1, and can be implemented by software (sound processing program) executed by the CPU and a CPU (central processing unit). Although it is possible to realize it, even in the case of adopting any of the realization methods, functionally it can be represented by FIG.

  In FIG. 1, the noise suppression apparatus 100 according to the first embodiment includes an L channel noise suppression circuit 101, an R channel noise suppression circuit 102, a time difference detection circuit 103, an L channel time correction circuit 104, and an R channel time. A correction circuit 105 is provided.

  Although not shown in FIG. 1, from a different point of view, it can be said that a stereo audio signal output device such as a headset is also a component of the noise suppression device 100 of the first embodiment (this is not the case) Will be clarified in the operation explanation section). However, the noise suppression apparatus 100 according to the first embodiment may be provided in an apparatus on the transmitting side of an audio signal, or may be provided in an apparatus on the receiving side of an audio signal.

  The audio signal (input audio signal) S input to the noise suppression apparatus 100 according to the first embodiment is branched into two, and the branched input audio signals SL and SR respectively correspond to one of the stereo audio signals for the L channel. It is supplied to the noise suppression circuit 101 or the R channel noise suppression circuit 102. Although FIG. 1 describes that the input audio signal S is branched into two at a branch of the signal line (a kind of signal branch circuit), it is realized by various existing techniques for dividing the input audio signal S into two. The signal branching circuit described above may be applied.

  The L channel noise suppression circuit 101 and the R channel noise suppression circuit 102 are noise suppression circuits provided as S / N ratio change circuits.

  The L channel noise suppression circuit 101 suppresses the noise component contained in the L channel input speech signal SL to itself and outputs the noise-suppressed L channel speech signal SSL. As the L channel noise suppression circuit 101, a known noise suppression circuit to which an existing noise suppression method such as a Wiener filter method, a spectral subtraction method, or a voice switch is applied can be applied. However, the L channel noise suppression circuit 101 is required to suppress the noise component in accordance with an externally applied suppression amount parameter (in other words, an S / N ratio change parameter) PL. For example, in the Wiener filter method or spectral subtraction method, the value of the noise component is obtained in the frequency domain, and the amount of suppression specified by the parameter is obtained for each noise component from the input speech signal in the frequency domain. This is a method of suppressing the noise component by subtraction, and the amount of suppression of the noise component can be changed according to the suppression amount parameter PL.

  The R channel noise suppression circuit 102 suppresses the noise component contained in the R channel input speech signal SR to itself, and outputs the L channel speech signal SSR after the suppression. The R channel noise suppression circuit 102 also needs to suppress the noise component in accordance with the externally applied suppression amount parameter PR.

  The noise suppression method and internal configuration applied to the R channel noise suppression circuit 102 may be the same as or different from the noise suppression method and internal configuration applied to the L channel noise suppression circuit 101. It is good. For example, both the L channel noise suppression circuit 101 and the R channel noise suppression circuit 102 may be in accordance with the Wiener filter method, and the L channel noise suppression circuit 101 may be in accordance with the Wiener filter method. The noise suppression circuit 102 may follow the spectral subtraction method.

  In the case of the first embodiment, the suppression amount parameter PL for the L channel noise suppression circuit 101 and the suppression amount parameter PR for the R channel noise suppression circuit 102 are different as the S / N ratio change parameter. . By making the suppression amount parameters PL and PR different from each other, the target sound component in the L channel audio signal SSL after the noise suppression from the L channel noise suppression circuit 101 and the target sound component from the R channel noise suppression circuit 102 are obtained. For the R channel, the noise component level in the L channel audio signal SSL after noise suppression from the L channel noise suppression circuit 101 while the target sound component in the R channel audio signal SSR after noise suppression is at the same level The level of the noise component in the R channel audio signal SSR after noise suppression from the noise suppression circuit 102 is made different. For example, when the suppression amount parameter PL is selected to reduce the suppression amount more than the suppression amount parameter PR, the level of the noise component in the L channel audio signal SSL after noise suppression is the R after noise suppression. It becomes larger than the level of the noise component in the channel audio signal SSR.

  The L channel noise suppression circuit 101 and the R channel noise suppression circuit 102 may output time domain signals as their output signals SSL and SSR, respectively, and output frequency domain signals. It may be something.

  The time difference detection circuit 103, the L channel time correction circuit 104, and the R channel time correction circuit 105 are provided in consideration of processing delays in the L channel noise suppression circuit 101 and the R channel noise suppression circuit 102. It is

  If the processing delays in L channel noise suppression circuit 101 and R channel noise suppression circuit 102 are the same, in other words, L channel speech signal SSL after noise suppression and R channel speech signal SSR after noise suppression are synchronized. In this case, the time difference detection circuit 103, the L channel time correction circuit 104, and the R channel time correction circuit 105 can be omitted.

  The time difference detection circuit 103 calculates the cross-correlation between the noise-suppressed L channel audio signal SSL and the noise-suppressed R channel audio signal SSR, for example, to calculate the noise-suppressed L channel audio signal SSL and the noise The time difference D with the later R channel audio signal SSR is calculated. For example, when the signals SSL and SSR are signals in the frequency domain, a frame which is a sample of a predetermined period (for example, 20 ms) to be subjected to FFT (Fast Fourier Transform) may be used as a unit of time, and the signals SSL and SSR In the case of a signal in the time domain, a sampling cycle or a frame may be used as a unit of time.

  The time difference detection circuit 103 forms correction delay amounts DL and DR for eliminating the time difference D between the noise-suppressed L-channel audio signal SSL and the noise-suppressed R-channel audio signal SSR, and corresponds them to each other. The time correction circuits 104 and 105 are supplied. When the time of the L channel audio signal SSL is behind the time of the R channel audio signal SSR by the time difference D, the time difference detection circuit 103 sets the correction delay amount DL to be supplied to the L channel time correction circuit 104 to 0. It is assumed that the correction delay amount DR given to the R channel time correction circuit 105 is D. Conversely, when the time of the R channel audio signal SSR is behind the time of the L channel audio signal SSL by the time difference D, the time difference detection circuit 103 corrects the correction delay amount to be given to the R channel time correction circuit 105. Let DR be 0 and the correction delay amount DL to be supplied to the L channel time correction circuit 104 be D.

  Even when the processing delays in the L channel noise suppression circuit 101 and the R channel noise suppression circuit 102 are different, the time difference D is determined by the configurations of the L channel noise suppression circuit 101 and the R channel noise suppression circuit 102, etc. When it can be specified in advance, the time difference detection circuit 103 and one of the time correction circuits 104 or 105 can be omitted.

  The L-channel time correction circuit 104 delays the noise-suppressed L-channel audio signal SSL by the correction delay amount DL and sends it out as an L-channel output audio signal OUTL from the noise suppression device 100. Similarly, the R-channel time correction circuit 105 delays the noise-suppressed R-channel audio signal SSR by the correction delay amount DR and sends it out as an R-channel output audio signal OUTR from the noise suppression device 100.

  Although not shown in FIG. 1, when the output audio signals OUTL and OUTR are signals in the frequency domain, it is necessary to provide an IFFT (Inverse Fast Fourier Transform) circuit before reaching the output device of the stereo audio signal. .

(A-2) Operation of First Embodiment Next, the operation of the noise suppression device 100 according to the first embodiment having the above-described detailed configuration will be described.

  In the noise suppression apparatus 100 of the first embodiment, the input speech signal S is branched into two, and the L-channel input speech signal SL obtained by branching is given to the L channel noise suppression circuit 101 and is obtained by branching. An R-channel input speech signal SR is applied to an R-channel noise suppression circuit 101.

  The noise component contained in the L channel input speech signal SL is suppressed in the L channel noise suppression circuit 101 by the amount indicated by the suppression amount parameter PL, and the R channel noise suppression circuit 102 The noise component contained in the input speech signal SR is suppressed by the amount indicated by the suppression amount parameter PR.

  The time difference detection circuit 103 detects the time difference D between the noise suppressed L channel audio signal SSL and the noise suppressed R channel audio signal SSR, and the correction delay amounts DL and DR for eliminating the time difference D are formed. Then, they are supplied to the corresponding L channel time correction circuit 104 and R channel time correction circuit 105 respectively.

  The L channel audio signal SSL after noise suppression is delayed by the correction delay amount DL by the L channel time correction circuit 104, and is sent out as an L channel output audio signal OUTL from the noise suppression device 100, and similarly noise suppression is performed. The subsequent R channel audio signal SSR is delayed by the correction delay amount DR by the R channel time correction circuit 105, and is sent out as the R channel output audio signal OUTR from the noise suppression device 100.

  FIG. 2 is an explanatory view showing signal waveform examples of an input speech signal S to the noise suppression apparatus 100 and an output speech signal (L channel output speech signal and R channel output speech signal) OUTL and OUTR from the noise suppression apparatus 100. is there. In FIG. 2, "V" represents a section in which the target sound component and the noise component are superimposed, and "N" represents a section including only the noise component. FIG. 2 shows a signal waveform in the case where the suppression amount parameter PL is selected to make the suppression amount smaller than the suppression amount parameter PR.

  Since the suppression amount parameter PL is selected to reduce the suppression amount, the target sound component in the L channel output audio signal OUTL and the target sound component in the R channel output audio signal OUTR have similar levels. The level of the noise component in the L channel output audio signal OUTL is greater than the level of the noise component in the R channel output audio signal OUTR.

  Here, as an output device of a stereo audio signal including an L channel output audio signal OUTL and an R channel output audio signal OUTR (in the case of a signal in the frequency domain, conversion to a time domain signal is required) The headset listens to both ears (note that the output device is not limited to the headset). The target sound component in the pronunciation sound has almost the same sound pressure on the left and right, so it can be localized at the center of the head and heard. The noise components in the pronunciation speech have different amounts of suppression as described above, so the sound pressures on the left and right are different, and can be localized and heard at a position shifted from the center in the head to either the left or the right. In the case of the example of FIG. 2, since the sound pressure of the noise component to the left ear is larger than the sound pressure of the noise component to the right ear, the noise is localized at a position shifted from the center to the left in the head and heard.

  That is, the localization positions (sound source positions) of the target sound and the noise can be made different, and when perceiving in the brain, the target sound and the noise can be easily separated. Humans can distinguish target sounds from a plurality of sound sources consciously, and taking into consideration human auditory characteristics, noise suppression effects can be enhanced even if the S / N ratio of the input speech signal is poor. .

(A-3) Effects of the First Embodiment As described above, according to the noise suppression device of the first embodiment, a large noise suppression effect can be obtained even if the S / N ratio of the input speech signal is bad. it can. That is, according to the noise suppression device of the first embodiment, it is possible to easily listen to the target sound.

(B) Second Embodiment Next, a second embodiment of the sound signal processing device and the sound signal processing program according to the present invention will be described with reference to the drawings. The sound signal processing device and the sound signal processing program according to the second embodiment are also a noise suppressing device and a noise suppressing program for suppressing a noise component contained in an input sound signal.

  FIG. 3 is a block diagram showing the configuration of the noise suppression apparatus according to the second embodiment. The same reference numerals as in FIG. 1 according to the first embodiment denote the same or corresponding parts.

  As is clear from the comparison of FIG. 3 and FIG. 1, the noise suppression apparatus 200 of the second embodiment is different from the noise suppression apparatus 100 of the first embodiment in the noise suppression circuit 101 for L channel and the time correction for R channel. The circuit 105 is omitted.

  Omitting the L channel noise suppression circuit 101 instructs the L channel noise suppression circuit 101 according to the first embodiment that the suppression amount parameter PL to be given is not suppressed, and the branch L channel audio signal SL is transmitted as it is. This is equivalent to the L channel audio signal SSL after suppression.

  In the second embodiment, since the L channel noise suppression circuit 101 is omitted, a processing delay occurs only in the R channel audio signal SSR after suppression (L channel audio after suppressing the branched L channel audio signal SL. There is no processing delay in the L channel audio signal SSL after suppression when it is regarded as the signal SSL). Therefore, the time correction circuit (time correction circuit for L channel) only in the L channel processing system where no processing delay occurs 104 is provided.

  The other points in the noise suppression apparatus 200 of the second embodiment are similar to those of the first embodiment.

  In the case of the noise suppression device 200 of the second embodiment, the input speech signal S is applied as it is as the L channel output speech signal OUTL, but as described in the first embodiment with reference to FIG. The target sound component in the output audio signal OUTL and the target sound component in the R channel output audio signal OUTR can be made to the same level, and the noise component in the L channel output audio signal OUTL is the noise component in the R channel output audio signal OUTR Can be greater than As a result, when listening to these stereo audio signals, the localization positions (sound source positions) of the target sound and the noise can be made different, and the target sound can be easily separated by being conscious of human beings .

  Even with the noise suppression device of the second embodiment, a large noise suppression effect can be achieved even if the S / N ratio of the input speech signal is poor, and it becomes easy to listen to the target sound.

(C) Third Embodiment Next, a third embodiment of the sound signal processing device and program according to the present invention will be described with reference to the drawings. The sound signal processing device of the third embodiment is also a noise suppression device that suppresses noise components included in the input sound signal.

  FIG. 4 is a block diagram showing the configuration of the noise suppression device according to the third embodiment.

  In FIG. 4, the noise suppression apparatus 300 according to the third embodiment includes a noise suppression apparatus main body 301 and a sound source separation circuit 302.

  The noise suppressor main body 301 corresponds to the noise suppressor 100 or 200 according to the first embodiment or the second embodiment (which may be a modified embodiment thereof). In the case of the third embodiment, the L channel output audio signal OUTL and the R channel output audio signal OUTR from the noise suppressor main body 301 are supplied to the sound source separation circuit 302.

  The sound source separation circuit 302 performs sound source separation processing on the L channel output sound signal OUTL and the R channel output sound signal OUTR as sound signals captured by two microphones, and generates a sound source of the target sound component and a sound source of the noise component. An output audio signal OUT which includes only the target sound component separated and separated is obtained and output.

  The sound source separation circuit 302 can apply an existing sound source separation circuit that performs sound source separation from a plurality of input audio signals, and therefore the description of the detailed configuration is omitted. As the sound source separation circuit 302, for example, a sound source separation circuit described in JP-A-2014-215461 or JP-A-2014-157261 may be applied.

  Even with the noise suppression device of the third embodiment, a large noise suppression effect can be achieved even if the S / N ratio of the input speech signal is bad. Moreover, according to the noise suppression device of the third embodiment, the output sound signal of the monaural signal in which the noise component is suppressed can be obtained, and the target sound can be easily listened to.

(D) Other Embodiments In the description of each of the above embodiments, various modified embodiments are referred to, but further, modified embodiments as exemplified below can be mentioned.

  In the above embodiments, the suppression amount parameters PL and PR have fixed values. However, at least one of the suppression amount parameters PL and PR may be dynamically changed. For example, when there are many noise components in the input audio signal S, the difference between the suppression amounts indicated by the suppression amount parameters PL and PR is increased, and the deviation between the localization position of the target sound component and the localization position of the noise components is large. It may be possible for a human to easily recognize the target sound component. For example, the power of the noise component belongs by separating the target sound section and the noise section by a known method, calculating the power of the noise section by a known method, and comparing the power of the noise section with the boundary value of the range. The range may be detected, and the suppression amount parameters PL and PR associated with the detected range may be read out from the storage table to dynamically change the suppression amount parameters PL and PR.

  In each of the above embodiments, the noise suppression circuit (101, 102) for suppressing the noise component is applied as the S / N ratio change circuit. However, as the S / N ratio change circuit, the voice component (target sound component A voice emphasis circuit may be applied to emphasize the In this case, although the target sound component in the sound generation voice is localized at a position other than the center in the head, it is different from the localization position of the noise component, so that the same effect as each embodiment described above can be obtained.

  Also, noise suppression processing is performed on one of the L channel input audio signal SL and the R channel input audio signal SR, and audio enhancement processing is performed on the other of the L channel input audio signal SL and the R channel input audio signal SR. You may

  Furthermore, although the S / N ratio changing circuit described above suppresses noise components or enhances the target sound component to increase the S / N ratio, the S / N ratio is changed. It is also possible to apply an S / N ratio changing circuit that reduces. As the S / N ratio changing circuit in this case, there can be mentioned a circuit for suppressing an audio component and a circuit for emphasizing a noise component. Even when an S / N ratio change circuit that reduces the S / N ratio is applied, the localization positions of the target sound component and the noise component in the sound production voice are different by making the change degree different for the L channel and the R channel. It can be a thing, and the target sound can be separated by human being conscious.

  In each of the above embodiments, the case where the target sound is voice is shown, but the present invention can be applied to the case where the target sound is a sound (for example, a sound emitted by a machine).

  100, 200, 300 ... noise suppression device, 101 ... noise suppression circuit for L channel, 102 ... noise suppression circuit for R channel, 103 ... time difference detection circuit, 104 ... time correction circuit for L channel, 105 ... time for R channel Correction circuit, 301: Noise suppression device main body, 302: Source separation circuit.

Claims (8)

A branch processing unit configured to bifurcate an input sound signal in which a target sound component and a noise component are mixed to generate first and second branched sound signals;
An S / N ratio change processing unit that generates a first and second S / N ratio change sound signal by changing the S / N ratio with respect to the first and second branch signals ;
To synchronize the first and second S / N ratio changing sound signal, have a synchronized unit which outputs one L-channel signal, the other as R-channel signal,
The above S / N ratio change processing unit
The S / N ratio in the first branched sound signal is changed by suppressing the noise component contained in the first branched sound signal by the suppression amount defined by the first suppression amount parameter,
The second branched sound signal is suppressed by suppressing the noise component included in the second branched sound signal by a suppression amount defined by a second suppression amount parameter different from the first suppression amount parameter. A sound signal processing apparatus characterized by changing the S / N ratio in . The synchronization unit
A time difference detection unit that detects a time difference between the first S / N ratio change sound signal and the second S / N ratio change sound signal;
A delay unit for delaying at least one of the first S / N ratio change sound signal and the second S / N ratio change sound signal based on the detected time difference. the sound signal processing apparatus according to 1. A sound source separation unit that performs sound source separation on the first and second S / N ratio change sound signals generated by the S / N ratio change processing unit to obtain and output a sound signal having a target sound component the sound signal processing apparatus according to claim 1 or 2, characterized in that it further comprises a. A branch processing unit configured to bifurcate an input sound signal in which a target sound component and a noise component are mixed to generate first and second branched sound signals;
An S / N ratio change processing unit that generates a first S / N ratio change sound signal by changing the S / N ratio with respect to the first branched signal ;
A synchronization unit that synchronizes the first S / N ratio change sound signal with the second branch sound signal, and outputs one as an L channel signal and the other as an R channel signal ;
The above S / N ratio change processing unit
A noise component included in said first branch tone signal, by a predetermined amount of suppression parameter is suppressed by suppressing the amount that defines, that to change the S / N ratio in the first branch sound signal Sound signal processing device characterized by the above. The synchronization unit
A time difference detection unit that detects a time difference between the first S / N ratio change signal and the second branch sound signal;
The sound signal processing apparatus according to claim 4, further comprising: a delay unit that delays the second branched sound signal based on the detected time difference. Sound source separation is performed on the first S / N ratio change signal generated by the S / N ratio change unit and the second branch signal to obtain and output a sound signal having a target sound component The sound signal processing apparatus according to claim 4 , further comprising a separation unit. Computer,
Branch processing means for bifurcating an input sound signal in which a target sound component and a noise component are mixed to generate first and second branched sound signals;
S / N ratio change processing means for generating first and second S / N ratio change sound signals by changing the S / N ratio with respect to the first and second branch signals ;
The first and synchronizes the second S / N ratio changing sound signal, one of the L-channel signal and the other to function as a synchronization means to output as R-channel signal,
The above S / N ratio change processing means
The S / N ratio in the first branched sound signal is changed by suppressing the noise component contained in the first branched sound signal by the suppression amount defined by the first suppression amount parameter,
The second branched sound signal is suppressed by suppressing the noise component included in the second branched sound signal by a suppression amount defined by a second suppression amount parameter different from the first suppression amount parameter. The sound signal processing program characterized by changing the S / N ratio in . Computer,
Branch processing means for bifurcating an input sound signal in which a target sound component and a noise component are mixed to generate first and second branched sound signals;
S / N ratio change processing means for generating a first S / N ratio change sound signal by changing the S / N ratio with respect to the first branched signal ;
The first by synchronizing the S / N ratio changing sound signal and the second branch sound signal, one of the L-channel signal and the other to function as a synchronization means to output as R-channel signal,
The above S / N ratio change processing means
The S / N ratio in the first branched sound signal is changed by suppressing the noise component contained in the first branched sound signal by the suppression amount defined by a predetermined suppression amount parameter. Sound signal processing program to be.

Images