Sound collection device, sound collection program, and sound collection method

Abstract

Provided are a sound collection device, program, and method for suppressing distortion of sound collected when performing area sound collection processing for collecting sound in a target area. [Solution] The area sound collection device calculates a null former spectrum for each microphone array based on input spectra supplied from multiple microphone arrays, updates the set target area direction based on the input spectrum and the set target area direction, calculates a null correction gain value for each null former spectrum that makes the frequency characteristics in the set target area direction closer to flatter, acquires a corrected null former spectrum using the null correction gain value, selects a corrected null former spectrum with the smallest output from the multiple corrected null former spectra for each microphone array, acquires the selected corrected null former spectrum as a beamformer output, and extracts and acquires the target area sound whose sound source is the target area using the acquired beamformer output for each microphone array.

Description

[Technical field]

[0001] The present invention relates to a sound collection device, a program, and a method, and can be applied to, for example, a sound collection device that emphasizes only sounds emitted from a specific area and suppresses sounds from other areas. [Background technology]

[0002] When using a speech recognition system in a noisy environment, ambient noise that is mixed in with the desired target sound is a nuisance that reduces the speech recognition rate of the recorded voice.

[0003] Conventionally, in an environment where multiple sound sources exist, beamforming using a microphone array has been used as a technology for obtaining the necessary target sound by collecting only sounds from a specific direction and avoiding the inclusion of unnecessary sounds. Beamforming is a technology for forming directivity by utilizing the time difference between signals arriving at each microphone (see Non-Patent Document 1). A microphone array that performs beamforming is also called a beamformer.

[0004] On the other hand, since a beamformer collects all sounds from a specific direction, if you want to collect only a specific area (also called a target area), the beamformer will also collect sounds outside the target area in the same direction as the beamformer. Therefore, Patent Document 1 proposes a method of collecting target sounds by using multiple beamformers, directing their directivities from different directions toward the target area, and having the directivities intersect at the target area (hereinafter referred to as "area sound collection").

[0005] Area pickup requires the beamformer to have sharp directionality in order to form the target area. It is desirable for the beamformer's beam width to be approximately 60 degrees or less. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2013-183358 A [Non-patent literature]

[0007] [Non-Patent Document 1] Futoshi Asano, "Acoustic Technology Series 16: Array Signal Processing of Sound - Localization, Tracking and Separation of Sound Sources", edited by the Acoustical Society of Japan, Corona Publishing, published on February 25, 2011 Summary of the Invention [Problem to be solved by the invention]

[0008] In both additive and subtractive beamformers, a large number of microphones are required to form a sharp directivity. For example, the Minimum Variance Distortionless Response (MVDR) requires one more microphone than the number of noise sources to be suppressed. Although area recording with such a large number of microphones can be experimentally achieved in a laboratory, it is not suitable for a commercially available product in terms of component costs and installation space.

[0009] On the other hand, the technology described in Patent Document 1 uses spectral subtraction in the frequency domain (a method of subtracting amplitude spectra from each other, but replacing negative results with zero or a small number) to achieve sharp directivity with a small number of microphones. A null former that forms a blind spot in the direction of the target area is created, and sharp directivity is formed in the direction of the target area by spectrally subtracting the null former amplitude spectrum from the microphone input amplitude spectrum. This method can form sharp directivity with just two microphones, but there is a risk of distortion such as musical noise and missing area sound components.

[0010] Therefore, with conventional technologies, there were issues such as the need for a large number of microphones to achieve area sound collection, which resulted in the equipment becoming large in size, and the use of spectral subtraction with a small number of microphones resulted in distortion of the extracted area sound.

[0011] In view of the above problems, there is a need for a sound collection device, a sound collection program, and a sound collection method that suppress distortion of sounds collected when performing area sound collection processing to collect sounds in a target area. [Means for solving the problem]

[0012] The first sound collection device of the present invention comprises: a null former forming means for forming, for each of a plurality of microphone arrays, a plurality of null formers each having a blind spot facing outside a target area based on an input spectrum of an acoustic signal supplied from the plurality of microphone arrays, and calculating a null former spectrum; a target area direction estimating means for performing, for each of the microphone arrays, a target area direction estimating process for estimating a target area direction based on the input spectrum and a currently set target area direction, and setting and updating the set target area direction based on a result of the target area direction estimation process; and a frequency characteristic of the set target area direction set and updated by the target area direction estimating means for each of the null former spectra. The system is characterized by having a null-correction gain value calculation means that calculates and holds a null-correction gain value that brings the signal closer to flat; a null characteristic correction means that acquires a corrected null former spectrum corrected using the corresponding null-correction gain value for each of the null former spectra; an acoustic enhancement means that selects, for each of the microphone arrays, the corrected null former spectrum with the smallest output from the corresponding plurality of corrected null former spectra and acquires the selected corrected null former spectrum as a beamformer output; and a target area sound acquisition means that extracts and acquires a target area sound having a sound source in the target area using the beamformer output for each of the microphone arrays acquired by the acoustic enhancement means.

[0013] A second sound collection program of the present invention includes a computer, comprising: a null former forming means for forming, for each microphone array, a plurality of null formers each having a blind spot facing outside a target area based on an input spectrum of an acoustic signal supplied from the plurality of microphone arrays, and calculating a null former spectrum; a target area direction estimating means for performing, for each of the microphone arrays, a target area direction estimating process for estimating a target area direction based on the input spectrum and a currently set set target area direction, and setting and updating the set target area direction based on a result of the target area direction estimation process; and a frequency characteristic of the set target area direction set and updated by the target area direction estimating means for each of the null former spectra. the null correction gain value calculation means calculating and holding a null correction gain value that brings the characteristic closer to flatter; a null characteristic correction means acquiring a corrected null former spectrum corrected using the corresponding null correction gain value for each of the null former spectra; an acoustic enhancement means selecting, for each of the microphone arrays, the corrected null former spectrum having the smallest output from a plurality of the corrected null former spectra corresponding thereto, and acquiring the selected corrected null former spectrum as a beamformer output; and a target area sound acquisition means extracting and acquiring a target area sound having a sound source in the target area, using the beamformer output for each of the microphone arrays acquired by the acoustic enhancement means.

[0014] The third aspect of the present invention is a sound collection method performed by a sound collection device, the sound collection device having a null former forming means, a target area direction estimating means, a null correction gain value calculating means, a null characteristic correcting means, an audio enhancement means and a target area sound acquiring means, the null former forming means forming a plurality of null formers with blind spots facing outside a target area for each of the microphone arrays based on an input spectrum of an audio signal supplied from the plurality of microphone arrays to calculate a null former spectrum, the target area direction estimating means performing a target area direction estimation process for estimating a target area direction for each of the microphone arrays based on the input spectrum and a currently set target area direction, and updating the set target area direction based on a result of the target area direction estimation process, the null correction gain value calculating means calculating a null former spectrum for each of the null formers. and calculating and storing a null correction gain value for a beamformer spectrum that makes the frequency characteristics in the set target area direction set and updated by the target area direction estimation means closer to flatter, the null characteristic correction means acquiring a corrected null former spectrum corrected using the corresponding null correction gain value for each of the null former spectra, the acoustic enhancement means selecting the corrected null former spectrum having the smallest output from the corresponding plurality of corrected null former spectra for each of the microphone arrays and acquiring the selected corrected null former spectrum as a beamformer output, and the target area sound acquisition means extracting and acquiring a target area sound having the target area as a sound source using the beamformer output for each of the microphone arrays acquired by the acoustic enhancement means. Effect of the Invention

[0015] According to the present invention, it is possible to provide a sound collection device, a sound collection program, and a sound collection method that suppress distortion of sound collected when performing area sound collection processing to collect sound in a target area. [Brief description of the drawings]

[0016] [Figure 1]1 is a block diagram showing a functional configuration of an area sound collecting device according to an embodiment. [Diagram 2] 1 is a diagram showing an example of an arrangement configuration of microphones (microphone arrays) that supply acoustic signals (input signals) to an area sound collection device according to an embodiment. FIG. [Diagram 3] 1 is a block diagram showing an example of a hardware configuration of an area sound collecting device according to an embodiment. [Figure 4] FIG. 13 is a diagram showing an example of the direction of null formers in an arbitrary microphone array according to an embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0017] (A) Main embodiment Hereinafter, an embodiment of a sound collection device, a sound collection program, and a sound collection method according to the present invention will be described in detail with reference to the drawings. In this embodiment, an example in which the sound collection device, the sound collection program, and the sound collection method of the present invention are applied to an area sound collection device will be described.

[0018] (A-1) Configuration of the embodiment FIG. 1 is a block diagram showing the functional configuration of an area sound collecting device 1 according to the first embodiment.

[0019] The area sound collection device 1 suppresses sounds emitted outside a target area and collects sounds emitted within the target area based on acoustic signals captured by two microphone arrays having two microphones.

[0020] FIG. 2 is a diagram showing an example of the arrangement of microphones (microphone array) that supply acoustic signals (input signals) to the area sound collection device 1. As shown in FIG.

[0021] In FIG. 2, the target area TA to be collected is hatched (with a diagonal line pattern).

[0022] As shown in FIG. 2, the area sound collection device 1 of this embodiment performs a process of collecting sound having a sound source in a target area TA based on acoustic signals (input signals) supplied from four microphones M11, M12, M21, and M22.

[0023] Here, it is assumed that a first microphone array MA1 is made up of two microphones M11 and M12, and a second microphone array MA2 is made up of microphones M21 and M22. Note that in this embodiment, the number of microphones constituting each microphone array is two (2ch), and the number of microphone arrays is two (2-microphone array), but the number of microphones and microphone arrays and the number of microphones (number of channels) per microphone array may be increased as long as each is two or more.

[0024] Next, the internal configuration of the area sound collecting device 1 will be described with reference to FIG.

[0025] As shown in FIG. 1, the area sound collection device 1 has a frequency analysis unit 101, a null former forming unit 102, a null characteristic correction unit 103, an audio enhancement unit 104, a beamformer selection unit 105, a null correction gain calculation unit 106, and a target area direction estimation unit 107.

[0026] The frequency analysis unit 101 converts the acoustic signals (input signals) captured by the two microphone arrays MA1 and MA2 into frequency domain signals to obtain an input spectrum.

[0027] The null former forming unit 102 forms a plurality of null formers, each of which has a blind spot directed outside the target area, based on the input spectrum obtained by the frequency analysis unit 101, and obtains a plurality of null former spectra.

[0028] The destination area direction estimation unit 107 uses the input spectrum for each microphone array obtained by the frequency analysis unit 101 to perform a process of estimating the direction of the destination area for each microphone array (hereinafter referred to as the "destination area direction estimation process") and obtains the resulting direction.

[0029] The null correction gain calculation unit 106 calculates a correction value (hereinafter referred to as a “null correction gain value”) for each of the multiple null formers formed by the null former formation unit 102, which corrects the frequency characteristics in the direction obtained as a result of the destination area direction estimation process by the destination area direction estimation unit 107 to be flat, and supplies the null correction gain value to the null characteristic correction unit 103.

[0030] The null characteristic correction unit 103 multiplies each of the multiple null formers calculated by the null former formation unit 102 by multiple null correction gain values ​​(null correction gain values ​​corresponding to each null former spectrum) supplied from the null correction gain calculation unit 106 to obtain a "corrected null former spectrum."

[0031] The acoustic enhancement unit 104 obtains, for each microphone array, a beamformer output in which directivity is formed in the direction of the target area (beamformer output) based on the corrected null former spectrum calculated by the null characteristic correction unit 103.

[0032] The beamformer selection unit 105 extracts and obtains a spectrum of a target area sound (hereinafter referred to as an "area sound spectrum") whose sound source is the target area, based on the beamformer output sound calculated for each microphone array by the acoustic enhancement unit 104. The means and data format by which the beamformer selection unit 105 outputs the area sound spectrum (target area sound) are not limited, and various means and data formats can be applied.

[0033] FIG. 3 is a block diagram showing an example of the hardware configuration of the area sound collecting device 1. As shown in FIG.

[0034] FIG. 3 shows an example of a hardware configuration when the area sound collecting device 1 is configured using software (computer).

[0035] 3 includes, as a hardware component, a computer 300 in which a program (including the sound collection program of the embodiment) is installed. The computer 300 may be a computer dedicated to the sound collection program, or may be configured to be shared with programs of other functions.

[0036] The computer 300 shown in FIG. 3 includes a processor 301, a primary storage unit 302, and a secondary storage unit 303. The primary storage unit 302 is a storage unit that functions as a working memory (work memory) for the processor 301, and can be, for example, a high-speed memory such as a dynamic random access memory (DRAM). The secondary storage unit 303 is a storage unit that records various data such as an operating system (OS) and program data (including data of a sound collection program according to the embodiment), and can be, for example, a non-volatile memory such as a FLASH (registered trademark) memory, HDD, or SSD. In the computer 300 of this embodiment, when the processor 301 starts up, the OS and programs (including the sound collection program according to the embodiment) recorded in the secondary storage unit 303 are read, deployed on the primary storage unit 302, and executed.

[0037] Note that the specific configuration of the computer 300 is not limited to the configuration in Fig. 3, and various configurations can be applied. For example, if the primary storage unit 302 is a non-volatile memory (e.g., a flash memory, etc.), the secondary storage unit 303 may be excluded.

[0038] (A-2) Operation of the embodiment Next, the operation of the area sound collecting device 1 of this embodiment having the above-mentioned configuration will be described.

[0039] The microphone arrays MA1 and MA2 convert the input acoustic signals from analog to digital signals and supply them to the frequency analysis unit 101 as input signals.

[0040] The frequency analysis unit 101 performs an arbitrary frequency analysis on a given input signal in the time domain, and supplies the obtained input spectrum to the null former forming unit 102 and the destination area direction estimating unit 107. In the frequency analysis unit 101, there is no limitation on the frequency analysis method, and for example, a fast Fourier transform or a wavelet transform may be used.

[0041] The null former forming unit 102 calculates multiple null formers that form blind spots in predetermined directions other than the target area for each microphone array based on the supplied input spectrum, and supplies the obtained null former spectrum to the null characteristic correcting unit 103. The operation of the null former forming unit 102 will be described in detail. Since the target area is not a point but has a range, the target area direction also has a range.

[0042] Next, the positional relationship between each microphone array and the target area TA will be described.

[0043] FIG. 4 is a diagram showing an example of the direction of a null former in an arbitrary microphone array MAa.

[0044] In the following, "a" in "MAa" etc. indicates the number (identifier; ID) of the microphone array. Here, the number of the microphone array MA1 is 1, and the number of the microphone array MA2 is 2. In other words, the microphone array MAa shown in FIG. 4 indicates either the microphone array MA1 or MA2 (any microphone array). FIG. 4 illustrates that the microphone array MAa includes microphones ML and MR. For example, when the microphone array MAa is the microphone array MA1, the microphones ML and MR are microphones M11 and M12, respectively, and when the microphone array MAa is the microphone array MA2, the microphones ML and MR are microphones M21 and M22, respectively. In the image diagram of FIG. 4, the shape of the target area TA is illustrated as a circle for convenience of illustration, but in this embodiment, the actual shape of the target area TA is as shown in FIG. 2.

[0045] Here, the front direction (0 degree direction) of the microphone array MAa is defined as the direction that is perpendicular to the straight line connecting the positions (center positions) of the two microphones ML and MR and that forms the smallest angle with the direction from the center point of the two microphones ML and MR toward the center point (which can also be the center of gravity) of the target area TA (for example, toward the center point of the target area).

[0046] 2, since the target area TA is not a point but has a range, the apparent area direction from each microphone array also has a range. Therefore, here, the target area direction (range of the target area direction) in microphone array MA1 is θ1L~θ1R (rad) (-π / 2<θ1L<0<θ1R<π / 2), and the target area direction (range of the target area direction) in microphone array MA2 is θ2L~θ2R (rad) (-π / 2<θ2L<0<θ2R<π / 2).

[0047] As described above, the predetermined blind spot direction applied to each null former of the null former forming unit 102 must be outside the target area direction (a direction that does not include the range of the target area direction). Therefore, the n-th blind spot direction φan(rad) (n=1,...,N) of the microphone array MAa (a=1,2) must satisfy -π-θaL<φan<θaL or θaR<φan<π-θaR.

[0048] Here, since the microphone array MAa includes two microphones, when a blind spot is formed in φan, a blind spot is also formed in π-φan (=-π-φan). In addition, since it is difficult to know in advance from which direction noise will come, the predetermined blind spot direction may be determined in advance by design (applying a predesigned fixed value).

[0049] From the above, it is preferable that the predetermined blind angle directions applied to each null former of the null former forming unit 102 are determined in the same number from the two ranges of -π / 2≦φan<θaL and θaR<φan≦π / 2. The number of predetermined blind angle directions set for each microphone array in the null former forming unit 102 (the number of null formers set for each microphone array MAa in the null former forming unit 102) and the combination of blind angle directions are not limited. For example, when the number N of the predetermined blind angle directions is 6 (when N=6), φa1=-π / 2, φa2=-π / 3, φa3=-π / 6, φa4=π / 6, φa5=π / 3, and φa6=π / 2 can be selected, and this selection method is preferable. Note that N is not limited to 6, and may be designed to be increased or decreased according to the target area TA and the assumption of the interfering sound source (the number of interfering sound sources and the direction of arrival), etc.

[0050] Here, the input spectrum of microphone array a is X aL (ω), X aR (ω), the null former generating unit 102 generates a plurality of null former spectra Y an Calculate (ω), where ω is the angular frequency, i is the imaginary unit, d is the microphone spacing, and c is the speed of sound.

number

[0051] Next, a specific example of the destination area direction estimation process performed by the destination area direction estimating unit 107 will be described.

[0052] The destination area direction estimation unit 107 determines a new destination area center point direction for each microphone array based on the input spectrum supplied from the frequency analysis unit 101 and the current destination area center point direction θa (rad) (hereinafter, this θa will also be referred to as the "set destination area direction") stored in the destination area direction estimation unit 107, and updates (sets and updates) the direction as the new destination area center point direction θa (rad). Various methods can be used to update the destination area center point direction θa (rad) in the destination area direction estimation unit 107, but for example, the update method described below is preferable. Note that the initial value of θa (rad) is not limited, and a value stored in advance may be set. For example, the initial value of θa (rad) may be set to the front direction (0 degree direction) of each microphone array.

[0053] First, as shown in the following formulas (2) and (3), the destination area direction estimation unit 107 estimates three directions, θa, θa+δθ, and θa-δθ, for each microphone array, based on the direction of the center point of the current destination area stored in the destination area direction estimation unit 107 (here, these three directions are generalized as θ ak ) for the null former spectrum ^Y ak Here, δθ is a preset shift value, and can be set to any value. In this embodiment, the destination area direction estimation unit 107 calculates the null former spectrum ^Y in three directions, θa, θa+δθ, and θa-δθ, with the center point direction as the reference. ak (ω) is calculated, but any other combination of directions (for example, any four or more directions based on θa) may be applied. In other words, the null former spectrum ^Y ak It is desirable that the directions when calculating (ω) include at least a clockwise direction and a counterclockwise direction with θa as the reference (that is, three or more directions including θa).

number

[0054] Next, the destination area direction estimation unit 107 calculates each null former spectrum ^Y ak For (ω), the sum P ak Take.

number

[0055] Then, the destination area direction estimation unit 107 calculates the sum P ak The direction in which the value of is the smallest is updated as the new center point direction of the destination area. For example, when the sum value of the null former spectrum for θa+δθ is the smallest, the destination area direction estimation unit 107 updates θa+δθ as the new center point direction of the destination area. Then, the destination area direction estimation unit 107 supplies the updated center point direction of the destination area to the null correction gain calculation unit 106.

[0056] The null correction gain calculation unit 106 calculates a null correction gain value that flattens the frequency characteristics of the null former gain (the ratio between the input spectrum and the null former spectrum) in the direction of the center point of the target area supplied from the target area direction estimation unit 107 for each of the multiple null former spectra formed in a predetermined blind spot direction other than the target area, and supplies the null correction gain value to the null characteristic correction unit 103.

[0057] Here, the null-correction gain calculation unit 106 can use various methods to determine the null-correction gain value, but for example, the following determination method is preferable.

[0058] Here, a given blind spot direction φ an Consider the null former of (rad). In this case, the null former gain H an (ω, θ) can be expressed by the following equation (5), where θ is the direction of arrival of the sound source.

number

[0059] From equation (5), we can see that the null former gain differs for each angular frequency and for each direction from which the sound source arrives. Therefore, in order to flatten the frequency characteristics of the null former gain in the direction of the center point of the target area, it is necessary to correct the null former gain, which differs for each frequency, based on the direction of the center point of the target area (θa (rad)). The null correction gain value can be calculated by taking the reciprocal of the null former gain when the target sound arrives from the direction of the center point of the target area, θa (rad), as shown in the following equation (6).

number

[0060] The null characteristic correction unit 103 obtains a corrected null former spectrum by multiplying the multiple null former spectra supplied from the null former formation unit 102 by multiple null-correction gain values ​​supplied from the null-correction gain calculation unit 106. Specifically, the null characteristic correction unit 103 obtains the corrected null former spectrum by using the following equation (7).

number

[0061] The acoustic enhancement unit 104 selects the corrected null former spectrum with the smallest amplitude for each microphone array and for each frequency based on the multiple corrected null former spectra supplied from the null characteristic correction unit 103, and supplies the obtained beamformer spectrum (beamformer output sound) to the beamformer selection unit 105. Specifically, the acoustic enhancement unit 104 calculates the beamformer spectrum Ba(ω) of the microphone array a using the following equations (8) and (9), where ν(ω) is the index number of the blind angle at which the corrected null former spectrum is smallest.

number

[0062] The beamformer selection unit 105 selects the one that minimizes the amplitude of the corrected beamformer spectrum for each frequency based on the two supplied beamformer spectra, and selects it as the spectrum of the target area sound (area sound spectrum) whose sound source is the target area, and outputs the obtained area sound spectrum. Specifically, the beamformer selection unit 105 calculates the area sound spectrum Z(ω) based on the following formulas (10) and (11).

number

[0063] (A-3) Effects of the embodiment According to this embodiment, the following effects can be obtained.

[0064] In the area sound collection device 1 of this embodiment, multiple null formers are calculated in parallel, the characteristics are corrected so that the frequency characteristics in the center point direction of the target area are flat for each null former, and the minimum output is selected from among those outputs to suppress sounds coming from directions other than the target area direction, and a beamformer that collects only the target area direction is formed. In other words, the area sound collection device 1 of this embodiment realizes area sound collection processing based on multiple beamformer outputs without using spectral subtraction. As a result, the area sound collection device 1 of this embodiment can collect only area sounds without causing distortion such as that caused by spectral subtraction.

[0065] (B) Other embodiments The present invention is not limited to the above-described embodiments, and may include modified embodiments such as those exemplified below.

[0066] (B-1) In each of the above embodiments, the area sound collection device 1 is configured such that digital signals are supplied from the microphone arrays MA1, MA2. However, the area sound collection device 1 may be configured such that analog signals are supplied from the microphone arrays MA1, MA2, and the analog signals (acoustic signals captured by each microphone of each microphone array) are converted into digital signals on the area sound collection device 1 side. [Explanation of symbols]

[0067] 1...area sound collection device, 101...frequency analysis unit, 102...null former formation unit, 103...filter gain calculation unit, 104...acoustic enhancement unit, 105...beam former selection unit, 106...null correction gain supply unit, 107...target area direction estimation unit, M11, M12, M21, M22, MR, ML...microphone, MA1, MA2, MAa...microphone array, TA...target area

Claims

1. a null former generating means for generating a plurality of null formers each having a blind spot outside a target area based on an input spectrum of an acoustic signal supplied from a plurality of microphone arrays, and calculating a null former spectrum; a destination area direction estimating means for estimating a destination area direction based on the input spectrum and a currently set destination area direction for each of the microphone arrays, and setting and updating the set destination area direction based on a result of the destination area direction estimating means; a null-correction gain value calculation means for calculating and storing a null-correction gain value that makes the frequency characteristic of the set target area direction, which is set and updated by the target area direction estimation means, closer to flat for each of the null former spectra; a null characteristic correction means for acquiring a corrected null former spectrum by correcting each of the null former spectra using the corresponding null correction gain value; an acoustic enhancement means for selecting, for each of the microphone arrays, the corrected null former spectrum having the smallest output from the corresponding plurality of corrected null former spectra, and acquiring the selected corrected null former spectrum as a beamformer output; a target area sound acquisition means for extracting and acquiring a target area sound having a target area as a sound source, using the beamformer output for each of the microphone arrays acquired by the sound enhancement means; A sound collecting device comprising:

2. 2. The sound collection device according to claim 1, wherein the null-correction gain value for each null former spectrum calculated by the null-correction gain value calculation means is obtained by taking the reciprocal of a null former gain in the set target area direction.

3. Computer, a null former generating means for generating a plurality of null formers each having a blind spot outside a target area based on an input spectrum of an acoustic signal supplied from a plurality of microphone arrays, and calculating a null former spectrum; a destination area direction estimating means for estimating a destination area direction based on the input spectrum and a currently set destination area direction for each of the microphone arrays, and setting and updating the set destination area direction based on a result of the destination area direction estimating means; a null-correction gain value calculation means for calculating and storing a null-correction gain value that makes the frequency characteristic of the set target area direction, which is set and updated by the target area direction estimation means, closer to flat for each of the null former spectra; a null characteristic correction means for acquiring a corrected null former spectrum by correcting each of the null former spectra using the corresponding null correction gain value; an acoustic enhancement means for selecting, for each of the microphone arrays, the corrected null former spectrum having the smallest output from the corresponding plurality of corrected null former spectra, and acquiring the selected corrected null former spectrum as a beamformer output; a target area sound acquisition means for extracting and acquiring a target area sound having a target area as a sound source, using the beamformer output for each of the microphone arrays acquired by the sound enhancement means; A sound recording program characterized by functioning as follows.

4. In the sound collection method performed by the sound collection device, The sound collection device includes a null former forming means, a target area direction estimating means, a null correction gain value calculating means, a null characteristic correcting means, an audio enhancing means, and a target area sound acquiring means, The null former generating means generates a plurality of null formers each having a blind spot outside a target area based on an input spectrum of an acoustic signal supplied from a plurality of microphone arrays, and calculates a null former spectrum; The target area direction estimation means performs a target area direction estimation process for estimating a target area direction for each of the microphone arrays based on the input spectrum and a currently set target area direction, and updates the set target area direction based on a result of the target area direction estimation process; The null-correction gain value calculation means calculates and holds a null-correction gain value for each of the null former spectra that makes the frequency characteristic of the set target area direction, which is set and updated by the target area direction estimation means, closer to flat, and the null characteristic correction means obtains a corrected null former spectrum by correcting each of the null former spectra using the corresponding null correction gain value; The acoustic enhancement means selects, for each of the microphone arrays, the corrected null former spectrum having a minimum output from the corresponding plurality of corrected null former spectra, and acquires the selected corrected null former spectrum as a beamformer output; The target area sound acquisition means extracts and acquires a target area sound having a target area as a sound source, using the beamformer output for each of the microphone arrays acquired by the sound enhancement means. A sound collection method comprising:

Images