JP2008072600A - Acoustic signal processing apparatus, acoustic signal processing program, and acoustic signal processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To correct, in a limited manner, a component of a specific acoustic signal (typically, a signal component of vocal audio) approximately uniformly contained in signals of channels without disturbing stereophonic pitch or continuity of stereophonic signals. <P>SOLUTION: Spectrum data of input acoustic signals of two channels are generated (S3) and if data of a plurality of frequency bins belonging to a set frequency band corresponding to the specific acoustic signal (vocal audio signal) in the spectral data satisfy predetermined approximation conditions mutually between the two channels, degeneration correction is performed on power of the data of the frequency bins (S5). The corrected acoustic signal of a time domain based on the corrected spectral data is composed with a difference signal for another channel of the two channels, thereby generating output acoustic signals of two channels constituting a stereophonic signal (S8). <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an acoustic signal processing apparatus that corrects a component of a specific acoustic signal that is equally contained in both signals from two channel (L and R channel) input acoustic signals that constitute a stereo acoustic signal, and the processing to the processor. The present invention relates to an acoustic signal processing program to be executed and an acoustic signal processing method for executing the processing.

In general, an acoustic signal that is a source of acoustic data recorded on a recording medium such as a music CD is a sound from a plurality of sound sources such as a main vocal sound, a sound of a musical instrument, and a background vocal sound through two microphones. It is the collected stereo sound signal. Normally, when recording such a stereo sound signal, the main vocal is localized at the approximate center of the two microphones. Thereby, the signal component of the main vocal sound is included in each of the two-channel input sound signals constituting the stereo sound signal substantially equally (the signal level and phase are substantially the same), and the other sound signal components are Only included in one of the two channels or mainly in one.
An acoustic signal processing device provided in an acoustic device such as a music player has a signal component (specific identification) of a main vocal sound that is almost equally contained in both signals from each of the two-channel input acoustic signals constituting the stereo acoustic signal. Some of them have a function (hereinafter referred to as a vocal cancellation function) for performing a process of reducing (removing) an example of the component of the acoustic signal. This acoustic signal processing device equipped with the vocal cancellation function automatically converts, for example, an acoustic signal for karaoke from an acoustic signal including a signal component of the main vocal voice without newly recording a special acoustic signal for karaoke. Can be generated.

On the other hand, some conventional acoustic signal processing apparatuses realize a vocal cancellation function by using a difference signal between two channel signals (time domain signals) constituting a stereo acoustic signal as an output signal.
In Patent Document 1, a stereo sound signal is divided into a plurality of frequency bands for each channel (generates spectrum data) by FFT processing, and the similarity of signals between channels is divided for each frequency band (frequency bin). Depending on the degree, the sound source signal localized near the center of the two channels (that is, the power of frequency bins similar between the channels) is suppressed (reduced) or emphasized (enhanced), and the signal after the suppression processing is a signal in the time domain. A stereo sound signal processing device is shown that returns to the output signal as an output signal.
JP 2002-78100 A

However, if the difference signal between the two channel signals in the stereo sound signal is an output signal, the output signal is substantially a monaural sound signal even though the original sound signal is a stereo sound signal. There was a problem that the stereo sound was lost in sound. Further, in this acoustic signal processing, there is a problem that even a component of an acoustic signal having a frequency band different from that of a specific voice (main vocal voice) desired to be corrected (reduced or enhanced) is corrected.
On the other hand, as shown in Patent Document 1, when the power of frequency bins (divided frequency bands) similar between two channels in a stereo acoustic signal is simply corrected, a sequence of audio signals returned to the time domain There is a problem that the discontinuous output sound is heard as sound including noise. Further, the acoustic signal processing disclosed in Patent Document 1 is also corrected (reduced, etc.) because there is no particular limitation on the frequency band for performing the similarity calculation of signals between channels and the suppression (reduction) processing based on the calculation result. ), A component of an acoustic signal having a frequency band different from that of the specific voice (main vocal voice) desired is corrected.
In addition, when an acoustic signal processing function such as a vocal cancel function is incorporated in an electronic device having a relatively low arithmetic processing capability of a processor such as a portable recording device (so-called IC recorder) or a portable music player, the acoustic signal It is desirable that the processing load of processing is as small as possible. On the other hand, in the technique disclosed in Patent Document 1, for a very large number of frequency bins (all frequency bands), the calculation of the similarity and the calculation of the attenuation coefficient based on the signal component size ratio and the phase difference are performed. Further, there is a problem that it is necessary to perform an attenuation calculation and the calculation load of the acoustic signal processing is high.
Accordingly, the present invention has been made in view of the above circumstances, and the object of the present invention is to specify that the signals of each channel are included almost evenly without impairing the stereo sense and continuity of the stereo sound signal. An acoustic signal processing apparatus, a program thereof, and an acoustic signal processing method capable of executing, with a low computational load, a process of limitedly correcting a component of an acoustic signal (a signal component of vocal speech is a typical example) It is in.

In order to achieve the above object, the present invention provides a specific sound that is substantially equally contained in both signals from each of two channel (L channel and R channel) input sound signals (time domain sound signals) constituting a stereo sound signal. An acoustic signal processing device for correcting a signal component, an acoustic signal processing program for causing a predetermined processor to execute the processing, or an acoustic signal processing method for executing the processing, and the following (1) to (6) An apparatus including each means shown, a program that causes a predetermined processor (may be called a computer) to function as each means, or a method that executes a procedure corresponding to the processing of each means.
(1) Spectrum generating means for generating spectrum data of each of the two-channel input acoustic signals.
(2) Approximation for determining whether data of each of a plurality of frequency bins belonging to a predetermined frequency band corresponding to the specific acoustic signal in the spectrum data satisfies a predetermined approximate condition between the two channels. Discriminating means.
(3) Spectral correction for correcting the power of frequency bin data determined by the approximation determining means to satisfy the approximation condition in the spectrum data of each of the two channels (typically, the power is corrected to zero). means.
(4) A corrected acoustic signal generating unit that generates a corrected acoustic signal in the time domain for each of the two channels based on the spectrum data corrected by the spectrum correcting unit.
(5) Difference signal generation means for generating a difference signal of the input acoustic signal with respect to the other channel for each of the two channels.
(6) Output acoustic signal synthesizing means for generating a two-channel output acoustic signal constituting a stereo acoustic signal by synthesizing the corrected acoustic signal and the difference signal for each of the two channels.

Here, it is conceivable that the approximate condition is a condition in which the power ratio or difference of frequency bin data in the spectrum data of each of the two channels is within a predetermined range.
Further, it is conceivable that the output acoustic signal synthesis means synthesizes the corrected acoustic signal and the difference signal by their weighted addition.
As described above, in the present invention, after the correction obtained by correcting the power in a frequency bin (a predetermined unit frequency band divided by a predetermined frequency width) approximated between two channels in a stereo acoustic signal. The output acoustic signal is generated by synthesizing the acoustic signal and the difference signal between the signals of the two channels (the signal obtained by taking the difference between the two signals).

According to the present invention, for each of the two channels in the stereo sound signal, a signal obtained by synthesizing the corrected sound signal in which the stereo tone is maintained and the difference signal in which the continuity of sound is maintained is output sound. Signal.
Further, since the frequency band for performing the correction processing based on the approximate discrimination of the signal between the channels and the discrimination result is limited to the frequency band corresponding to the specific acoustic signal to be corrected, the specific acoustic desired to be corrected is The components of acoustic signals in different frequency bands are not corrected, and are limited to a range where the frequency bandwidth of the signal to be subjected to the approximation determination or the correction process is narrow.
As a result of the above, according to the present invention, the component of the specific acoustic signal (the signal component of the vocal voice is included in the signal of each channel substantially without impairing the stereo sense and continuity of the stereo acoustic signal. (Typical example) can be corrected limitedly, and the processing can be executed with a low calculation load.

Embodiments of the present invention will be described below with reference to the accompanying drawings for understanding of the present invention. In addition, the following embodiment is an example which actualized this invention, Comprising: It is not the thing of the character which limits the technical scope of this invention.
1 is a block diagram showing a schematic configuration of the acoustic signal processing device X according to the embodiment of the present invention, FIG. 2 is a flowchart showing a procedure of voice cancellation processing by the acoustic signal processing device X, and FIG. 3 is an acoustic signal processing. FIG. 4 is a diagram illustrating an example of data obtained in the process of voice cancellation processing by the device X. FIG. 4 is an example of a spectrum of an output acoustic signal obtained by the acoustic signal processing device X and a spectrum of an output acoustic signal obtained by the conventional voice cancellation processing. It is a figure showing an example.

First, the configuration of an acoustic signal processing device X according to an embodiment of the present invention will be described with reference to FIG.
The acoustic signal processing apparatus X has two channels (L channel and R) that constitute a stereo acoustic signal obtained by collecting sounds from a plurality of sound sources such as main vocal sound, instrument sound, and background vocal sound through two microphones. Processing for reducing the main vocal audio signal component (an example of a specific acoustic signal component) contained almost equally in both signals from each of the input audio signals XL (t) and XR (t) of the channel) (vocal cancellation) Processing (vocal canceller).
As shown in FIG. 1, the acoustic signal processing apparatus X includes an arithmetic unit 10, two A / D converters 11 and 12 provided for each channel, and two buffers 13 and 14 for input signals provided for each channel. (Hereinafter referred to as an input buffer), two buffers 15 and 16 for output signals provided for each channel (hereinafter referred to as output buffers), D / A converters 17 and 18 provided for each channel, a memory 19 and the like. It is prepared for.
Further, the acoustic signal processing apparatus X receives the first and second input terminals In1 and In2 for inputting the L channel and R channel analog input signals constituting the stereo acoustic signal, and the stereo acoustic signal after the vocal cancellation processing. A first output terminal Ot1 and a second output terminal Ot2 that output analog signals of the L channel and the R channel, and a third input terminal that inputs various control signals from an acoustic device or the like on which the acoustic signal processing device X is mounted. In3 is provided.

The two A / D converters 11 and 12 respectively convert the analog input audio signals of the respective channels constituting the analog stereo sound signal into digital L channel input signal XL (t) and R channel input signal XR at a predetermined sampling period. is converted to (t).
The two input buffers 13 and 14 are buffers that temporarily store the digitized L channel input signal XL (t) and R channel input signal XR (t), respectively, by a predetermined number of samples.
The memory 19 is a readable / writable storage means such as a RAM or a flash memory, and stores data that is read and written by the arithmetic unit 10 in the course of the processing.
The calculation unit 10 is a calculation unit such as a DSP (Digital Signal Processor) or MPU (Micro Processor Unit), for example, and realizes various functions by executing a program stored in a ROM (not shown) in advance. .
The arithmetic unit 10 executes a predetermined acoustic signal processing program, thereby digitizing input acoustic signals XL (t) and XR (t) of two channels (L channel and R channel) constituting a stereo acoustic signal. From each of them, a vocal canceling process for reducing the signal component of the main vocal sound included in both signals almost equally is executed. At that time, the calculation unit 10 performs processing to generate spectrum data (frequency domain acoustic signal) from the time domain acoustic signal by executing the FFT program, and also executes the IFFT program to execute the spectrum. Processing to generate a time domain acoustic signal from the data is also performed.

Next, the procedure of the vocal cancellation process executed by the calculation unit 10 will be described with reference to the flowchart shown in FIG. 2 and FIG. Hereinafter, S1, S2,... Represent identification codes of processing procedures (steps). FIG. 3 is a diagram illustrating an example of data obtained in the course of the voice cancellation process.
[Steps S1, S2]
First, the calculation unit 10 determines whether or not the control signal input through the third input terminal In3 is an execution command for voice cancel processing (whether or not an execution command is being input) (S1). stand by.
Here, the execution command is, for example, when an acoustic device or the like mounted with or connected to the acoustic signal processing device X is in accordance with an operation status (operation input information) of a predetermined operation input unit (operation key or the like). This signal is output to the acoustic signal processing device X.
[Step S2]
When the calculation unit 10 determines that the execution command is being input, the arithmetic input unit 10 inputs the two-channel input sound signals XL (t) and XR (t) (the digitized time-domain sound signals) constituting the stereo sound signal. ) Is input through the input buffers 13 and 14 for a predetermined time length (one frame) (S2).
FIG. 3A shows an image diagram of the input acoustic signals XL (t) and XR (t) input in step S2.
Next, the calculation unit 10 executes a predetermined program to execute the following corrected signal generation processing (S3 to S6) and difference signal generation processing (S7) in parallel or sequentially.

[Signal generation after correction]
[Step S3]
In the post-correction signal generation process, first, the arithmetic unit 10 generates the spectrum data XL (f) and XR (f) of the L and R two-channel input acoustic signals XL (t) and XR (t), The generated spectrum data XL (f) and XR (f) are stored in the memory 19 (S3, an example of a spectrum generation procedure). In addition, the calculating part 10 which performs step S3 is an example of a spectrum production | generation means.
In step S3, the calculation unit 10 executes an FFT (Fourier transform) program, and inputs the time-domain input acoustic signals XL (t) and XR (t) to the frequency-domain acoustic signals XL (f) and XR ( f) Convert to (ie spectral data). As a result, for each of the L and R channels, spectral data that is a set of FFT values for each frequency bin (a predetermined unit frequency band divided by a predetermined frequency width) is obtained. Here, the bandwidth of the frequency bin is, for example, 21.5 Hz when the sampling frequency of the acoustic signal is 44.1 kHz and the FFT size (number of unit samples of the FFT processing) is 2048 samples, and the FFT size is also the same. In the case of 4096 samples, it is appropriate that the frequency is 10.77 Hz.
FIG. 3B shows an image diagram representing the spectrum data XL (f) and XR (f) obtained in step S3.

[Step S4]
Next, the arithmetic unit 10 obtains a predetermined approximation between the data (FFT value) between the two channels for each of a plurality of frequency bins belonging to the predetermined set frequency band W0 in the spectrum data generated in step S3. Data approximation determination processing is performed to determine whether or not the condition is satisfied, and the determination result for each frequency bin is stored in the memory 19 (S4, an example of an approximation determination procedure). Hereinafter, the data of the frequency bin determined to satisfy the approximate condition by the process of step S4 is referred to as approximate data. In addition, the calculating part 10 which performs the process of step S4 is an example of an approximation determination means.
Here, the set frequency band W0 is a frequency band corresponding to a vocal voice signal (an example of a specific acoustic signal), that is, a human voice frequency band, for example, a band from 200 Hz to 5 kHz.
The approximation condition may be, for example, a ratio (P _L / _L ) between the power P _L of the frequency bin data (FFT value) in the _L channel spectrum data and the power P _R of the frequency bin data in the R channel spectrum data. P _R ) is within a predetermined range (for example, 1 ± α, α is a constant of 0 <α <1). Alternatively, a condition that the power difference (| P _L −P _R |) of the frequency bin data (FFT values) of both channels is within a predetermined range may be considered as the approximate condition.
The powers P _L and P _R of the data of each frequency bin are obtained from the square sum (R ² + I ² ) of the real part R and the imaginary part I of each data (FFT value) and the square root thereof.
If the constant α is set to a value close to 0, the signal component of the vocal sound included in the stereo sound signal is included almost equally in each of the L channel and the R channel, so that the frequency bin corresponding to the vocal sound The data satisfies the approximation condition. On the other hand, since the signal component of the sound other than the vocal sound is included only in one of the two channels or mainly in one of them, the frequency bin data corresponding to the sound other than the vocal sound does not satisfy the approximation condition.
Here, if α is too close to 0 (the approximation condition is too strict), the signal component of the vocal sound is not reduced, and if α is too large, the signal component is reduced to an acoustic signal component other than the vocal sound. Keep in mind that For example, the constant α is set to a value such as 0.2.

[Step S5]
Further, the calculation unit 10 performs a process of reducing and correcting the power of the approximate data (frequency bin data determined to satisfy the approximate condition by the approximate determination process in step S4) in the spectrum data of the L channel and the R channel. (S5, an example of a spectral correction procedure). As a result, the spectrum data XL (f) and XR (f) stored in the memory 19 by the process of step S3 are updated to the corrected spectrum data XL ′ (f) and XR ′ (f). In addition, the calculating part 10 which performs step S5 is an example of a spectrum correction means.
For example, the arithmetic unit 10 calculates the correction coefficient k0 (0 ≦ k0 <1) specified by the control signal input through the third input terminal In3 as the real part R and the imaginary part of the approximate data (FFT value) of both channels. Reduction correction is performed by multiplying each part I.
In this case, the correction coefficient k0 is determined according to the operation status (operation input information) of a predetermined operation input unit (operation key or the like), for example, by an acoustic device or the like mounted with or connected to the acoustic signal processing device X. This is a coefficient output to the acoustic signal processing device X. Thereby, the user of the apparatus can completely remove the vocal sound (k0 = 0) or suppress the volume of the vocal sound (0 <k0 <1). Of course, it is also conceivable that the arithmetic unit 10 performs the correction using a correction coefficient k0 (for example, k0 = 0) stored in advance in a ROM or the like.
FIG. 3C shows an image diagram representing the corrected spectrum data XL ′ (f) and XR ′ (f) obtained in step S5 when the correction coefficient k0 = 0. Of the frequency bins belonging to the range of the set frequency band W0, for the frequency bins (frequency bands W1 and W2) satisfying the approximate condition, the power of the spectrum data is corrected to 0 (zero).

[Step S6]
Further, based on the spectrum data XL ′ (f) and XR ′ (f) corrected by the process of step S5, the calculation unit 10 corrects the time-domain corrected acoustic signals XL ′ (t) and XR for each of the two channels. '(t) is generated and stored in the memory 19 (S6, an example of the corrected acoustic signal generation procedure). In addition, the calculating part 10 which performs the process of step S6 is an example of the corrected acoustic signal generation means.
In this step S6, the arithmetic unit 10 executes an IFFT (Inverse Fourier Transform) program, and each of the corrected spectrum data XL ′ (f) and XR ′ (f) representing the frequency domain acoustic signal is obtained in the time domain. The corrected acoustic signals XL ′ (t) and XR ′ (t) are converted. Thereby, for each of the two channels L and R, the corrected acoustic signals XL ′ (t) and XR ′ (t) for one frame (for a predetermined time length) in which the intensity of the signal component of the vocal sound is reduced and corrected are obtained. can get.

[Step S7: Difference Signal Generation Processing]
On the other hand, in the difference signal generation process, the arithmetic unit 10 calculates a difference signal (ΔXL (t) = XL (t) −) for each of the two channels L and R, which is the result of calculating the difference of the input acoustic signal with respect to the other channel. XR (t) and ΔXR (t) = XR (t) −XL (t)) are generated (S7, an example of a difference signal generation procedure). In addition, the calculating part 10 which performs step S7 is an example of a difference signal production | generation means.

[Step S8]
Then, after performing the corrected signal generation process (S3 to S6) and the difference signal generation process (S7), the arithmetic unit 10 performs the corrected acoustic signal XL ′ in the time domain for each of the two channels L and R. (t) and XR ′ (t) and the difference signals ΔXL (t) and ΔXR (t) are combined to output two-channel output acoustic signals YL (t) and YR (t Are stored in the output buffers 15 and 16 (S8, an example of an output acoustic signal synthesis procedure). Thereafter, the arithmetic unit 10 returns the process to step S1 described above. In addition, the calculating part 10 which performs step S8 is an example of an output acoustic signal synthetic | combination means.
In this step S8, for example, the calculation unit 10 uses the corrected acoustic signals XL ′ (t) and XR ′ (t) and the difference signals ΔXL (t) and ΔXR (t) as the following (1). As shown in the equation, they are synthesized by their weighted addition. However, k1 and k2 are predetermined weighting factors (0 <k1 ≦ 1, 0 <k2 ≦ 1).
YL (t) = k1 · XL ′ (t) + k2 · ΔXL (t)
YR (t) = k1 · XR ′ (t) + k2 · ΔXR (t) (1)
By repeating the processes in steps S1 to S8, the output acoustic signals YL (t) and YR (t) are stored in the output buffers 15 and 16 one frame at a time, and the output acoustic signal YL (t). , YR (t) (digital signal) is taken out by the D / A converters 17 and 18 sample by sample at the same period as the sampling period, and sequentially converted into analog signals and output through the output terminals Ot1 and Ot2. Thereby, an analog output sound signal (stereo sound signal) having a slight time lag with respect to the sound signals sequentially input through the input terminals In1 and In2 is output in real time through the output terminals Ot1 and Ot2.

As described above, according to the acoustic signal processing device X, for each of the two channels in the stereo acoustic signal, the correction in which the stereo sound is maintained by performing the vocal sound signal correction for each channel individually. A signal obtained by synthesizing the post-acoustic signals XL ′ (t) and XR ′ (t) and the difference signals ΔXL (t) and ΔXR (t) in which sound continuity is maintained is an output acoustic signal YL (t). , YR (t).
In addition, since the frequency band for performing the approximate determination of the signal between channels and the reduction process based on the determination result is limited to the frequency band corresponding to the vocal sound to be reduced, the frequency band is different from the vocal sound for which reduction is desired. The component of the acoustic signal is not reduced.
As a result of the above, according to the acoustic signal processing device X, the intensity of the signal component of the vocal sound that is almost uniformly included in the signals of the respective channels is limited without impairing the stereo sense and continuity of the stereo acoustic signal. Reduced and corrected.

FIG. 4 synthesizes an example of the spectrum YL (f) of the output acoustic signal of one channel obtained by the acoustic signal processing apparatus X and the difference signal ΔXL (t) that is the output acoustic signal obtained by the conventional voice cancellation processing. 6 is a graph showing an example of a spectrum XL ″ (f) of the corrected acoustic signal X ′ (t) before the correction. With respect to the spectrum YL (f) of the output acoustic signal shown in FIG. The weighting factor for combining t) was k1 = k2 = 0.5.
As can be seen from FIG. 4, in the 1000 Hz peripheral band W1 and the 2000 Hz peripheral band W2 (frequency bin) where the approximate condition is satisfied, a drop in power of the spectrum XL ″ (t) of the corrected acoustic signal due to the reduction correction, It can be seen that this is compensated by the synthesis of the difference signal ΔXL (t), which is not originally reduced (removed) among the signal components reduced (removed in this case) by the processing of step S5 described above. This means that the remaining signal components (signal components necessary for maintaining the continuity of the signal) are complemented, so that the continuity of the sound of the output acoustic signal YL (t) is maintained and the sound is not It is possible to avoid the occurrence of noise caused by being continuous.
Further, in the acoustic signal processing apparatus X, the calculation target is low because the frequency band to be processed such as the determination of the approximate condition is limited to the range of the set frequency band W0. For example, if the sampling frequency of the acoustic signal is 44.1 kHz and the FFT size is 4096 samples (frequency bin bandwidth is 10.77 Hz), the frequency band to be processed is a band from 200 Hz to 5 kHz (the set frequency band W0). ) (The calculation processing amount of the acoustic signal processing device X) is limited to 1/5 or less of the calculation processing amount when the limitation is not performed, as shown in Patent Document 1. .
It should be noted that if the weighting factor k2 used in the synthesis process (S8) described above is too small (close to 0), sufficient supplementation for maintaining the continuity of sound cannot be performed.

In the embodiment described above, an example is shown in which the component of a specific acoustic signal (vocal speech signal) is reduced by setting the correction coefficient k0 to “0 ≦ k0 <1”. However, the sound signal processing apparatus X corrects (reduces or enhances) the intensity of a component of a specific sound signal (vocabulary sound signal) to an arbitrary strength by allowing the correction coefficient k0 to be larger than 1. It is also conceivable to configure it as a device.
As a result, for example, a vocal sound signal in a stereophonic audio signal of music can be limitedly enhanced (emphasized) without hindering stereo sense or continuity.

  The present invention can be used for an acoustic signal processing device.

1 is a block diagram illustrating a schematic configuration of an acoustic signal processing device X according to an embodiment of the present invention. The flowchart showing the procedure of the voice cancellation process by the acoustic signal processing apparatus X. The figure showing an example of the data obtained in the process of the voice cancellation process by the acoustic signal processing apparatus X. The figure showing an example of the spectrum of the output acoustic signal obtained by the acoustic signal processing apparatus X, and an example of the spectrum of the output acoustic signal obtained by the conventional voice cancellation process.

Explanation of symbols

X ... acoustic signal processing device 10 ... arithmetic units 11, 12 ... A / D converters 13, 14 ... input buffers 15, 16 ... output buffers 17, 18 ... D / A converter 19 ... memory In1-In3 ... input terminals Ot1, Ot2 ... Output terminals S1, S2, ... Processing procedure (step)

Claims (5)

An acoustic signal processing device that corrects a component of a specific acoustic signal that is substantially equally contained in both signals from each of two-channel input acoustic signals constituting a stereo acoustic signal,
Spectrum generating means for generating spectrum data of each of the two-channel input acoustic signals;
Approximation determination means for determining whether data of a plurality of frequency bins belonging to a predetermined frequency band corresponding to the specific acoustic signal in the spectrum data satisfies a predetermined approximation condition between the two channels; ,
Spectrum correction means for correcting the power of the data of the frequency bin determined to satisfy the approximation condition by the approximation determination means in the spectrum data of each of the two channels;
A corrected acoustic signal generating means for generating a time-domain corrected acoustic signal for each of the two channels based on the spectral data corrected by the spectrum correcting means;
Difference signal generating means for generating a difference signal of the input acoustic signal with respect to the other channel for each of the two channels;
Output acoustic signal combining means for generating a two-channel output acoustic signal constituting a stereo acoustic signal by combining the corrected acoustic signal and the difference signal for each of the two channels;
An acoustic signal processing apparatus comprising:   The acoustic signal processing apparatus according to claim 1, wherein the approximate condition is a condition in which a power ratio or difference of frequency bin data in the spectrum data of each of the two channels is within a predetermined range.   The acoustic signal processing apparatus according to claim 1, wherein the output acoustic signal synthesizing unit synthesizes the corrected acoustic signal and the difference signal by weighted addition. An acoustic signal processing program for causing a predetermined processor to execute a process of correcting a component of a specific acoustic signal substantially equally contained in both signals from each of two channel input acoustic signals constituting a stereo acoustic signal,
A given processor,
Spectrum generating means for generating spectrum data of each of the two-channel input acoustic signals;
Approximation determination means for determining whether data of a plurality of frequency bins belonging to a predetermined frequency band corresponding to the specific acoustic signal in the spectrum data satisfies a predetermined approximation condition between the two channels; ,
Spectrum correction means for correcting the power of the data of the frequency bin determined to satisfy the approximation condition by the approximation determination means in the spectrum data of each of the two channels;
A corrected acoustic signal generating means for generating a time-domain corrected acoustic signal for each of the two channels based on the spectral data corrected by the spectrum correcting means;
Difference signal generating means for generating a difference signal of the input acoustic signal with respect to the other channel for each of the two channels;
Output acoustic signal combining means for generating a two-channel output acoustic signal constituting a stereo acoustic signal by combining the corrected acoustic signal and the difference signal for each of the two channels;
An acoustic signal processing program for functioning as each means. An acoustic signal processing method for executing a process of correcting a component of a specific acoustic signal substantially equally contained in both signals from each of two channel input acoustic signals constituting a stereo acoustic signal,
A spectrum generation procedure for generating spectrum data of each of the two-channel input acoustic signals;
An approximation determination procedure for determining whether data of each of a plurality of frequency bins belonging to a predetermined frequency band corresponding to the specific acoustic signal in the spectrum data satisfies a predetermined approximation condition between the two channels; ,
A spectrum correction procedure for correcting the power of the frequency bin data determined to satisfy the approximation condition by the approximation determination procedure in the spectrum data of each of the two channels;
A corrected acoustic signal generation procedure for generating a time domain corrected acoustic signal for each of the two channels based on the spectral data corrected by the spectral correction procedure;
A difference signal generation procedure for generating a difference signal of the input acoustic signal with respect to the other channel for each of the two channels;
An output acoustic signal synthesis procedure for generating a two-channel output acoustic signal constituting a stereo acoustic signal by synthesizing the corrected acoustic signal and the difference signal for each of the two channels;
An acoustic signal processing method comprising:

Images