JP5683140B2 - Noise-to-noise direct ratio estimation device, interference noise elimination device, perspective determination device, sound source distance measurement device, method of each device, and device program - Google Patents

Description

  The present invention can be applied to, for example, a voice call, a hands-free method of operating a device by voice input, and the like, and is used when collecting sound by emphasizing only the sound of a sound source located within a specific distance range from a microphone. The present invention relates to a noise-to-noise ratio estimation device, an interference noise removal device, a perspective determination device, a sound source distance measurement device, a method of each device, and a device program.

  Conventionally, for the purpose of identifying the distance from a sound source and enhancing or suppressing only the sound from the sound source within a specific distance range, the power of the direct sound and reverberant sound is estimated from the signal received from the microphone, There is an idea of obtaining the ratio (for example, Non-Patent Document 1). The idea that the conventional direct ratio estimation apparatus obtains the direct ratio will be described with reference to the drawings.

  FIG. 1 illustrates a scene in which the direct ratio estimation apparatus is used. Assume that a small microphone array 11 is surrounded by, for example, four speakers 12 to 14 for a conference. It is assumed that a television 16, a telephone 17, and a speaker 18 for broadcasting in the hall are arranged in the conference room. In such a scene, the speakers 12 to 14 located within a predetermined distance range (within a circle indicated by a broken line) around the small microphone array 11 without picking up the voice of the in-house broadcast or the sound of the telephone. I want to collect only the utterances.

  In order to distinguish the distance from the microphone array to the sound source, attention is paid to the ratio of direct sound and indirect sound (reverberation sound) included in the received sound (hereinafter referred to as direct ratio). FIG. 2 shows a sound propagation path from the sound source 21 to the microphone 22 when a microphone is placed indoors and a sound is recorded. The direct sound is a sound wave indicated by a thick solid line that directly reaches from the sound source 21 to the microphone. One reverberant sound is a sound wave indicated by a broken line that reaches the microphone 22 after the sound emitted from the sound source 21 is reflected by a wall, floor, ceiling, or the like.

  FIG. 3 shows the relationship between the direct ratio and the distance between the microphones. The horizontal axis in FIG. 3 is the distance from the microphone to the sound source, and the vertical axis is the direct ratio. In general, the indirect sound has a certain magnitude that does not depend on the distance from the microphone. In contrast to the indirect sound, the direct sound exhibits a characteristic that monotonously decreases as the distance from the microphone increases. The direct ratio obtained by dividing the direct sound by the indirect sound has a characteristic that decreases monotonously as the distance increases, as in the case of the direct sound.

  A conventional direct ratio estimation device can estimate the direct ratio from the received sound and estimate the distance of the sound source from the microphone array.

Y.Hioka, K.Niwa, S.Sakauchi, K.Furuya, and Y.Haneda. Estimating direct-to-reverberant energy ratio based on spatial correlation model segregating direct sound and reverberation.Proceedings of 2010 IEEE International Conference of Acoustics, Speech and Signal Processing (ICASSP2010), pages 149-152, 2010. Yusuke Hioka, Sumio Hannai, Kenichi Furuya, Yoichi Haneda, “Examination of sound collection by distance based on direct ratio of received signal”, Acoustical Society of Japan, 2009 Autumn Meeting, pp.633-634

  However, in general, in addition to direct sound and reverberation sound, noise specific to each microphone is superimposed on a signal collected by a microphone. In the conventional method described above, noise is not taken into account, and there is a problem that the accuracy of the direct ratio is deteriorated when the noise level is large.

  The present invention has been made in view of such problems, and prevents noise-to-direct ratio estimation accuracy from being deteriorated due to noise, and is capable of estimating the direct ratio with high accuracy even when noise is present. It is an object of the present invention to provide a direct ratio estimation device, an interference noise removal device, a perspective determination device, a sound source distance measurement device, a method of each device, and a device program using the same.

  The noise-to-noise direct ratio estimation device of the present invention includes a plurality of frequency domain conversion units and a direct ratio estimation unit. The plurality of frequency domain conversion units convert the received sound signals received by the plurality of microphones into frequency domain signals. The direct ratio estimating unit includes a spatial correlation matrix calculating means, a signal power estimating means, and a direct ratio calculating means. The spatial correlation matrix calculation means calculates the spatial correlation matrix by vectorizing the frequency domain signals with the frequency domain signals output from the plurality of frequency domain transform units as inputs. The signal power estimation means obtains a vector composed of direct sound power, reverberant power and noise power from the microphone arrangement information given in advance and the spatial correlation matrix calculated from the received sound, and the vector Outputs the power of the direct sound and reverberant sound of the element. The direct ratio calculation means calculates the direct ratio obtained by dividing the power of the direct sound by the power of the reverberant sound.

  Further, the interference noise removing apparatus of the present invention includes the noise-to-noise direct ratio estimating apparatus of the present invention, and in addition, one microphone array, a processing target signal generating unit, a target signal adjusting unit, And an inverse frequency domain transform unit.

  The noise-resistant direct ratio estimation apparatus of the present invention newly adds a cross-correlation model of noise to information on the cross-correlation between microphones used when performing direct ratio estimation, and obtains signal power. Thereby, it is possible to estimate the powers of the three components of the direct sound, the reverberant sound, and the noise separately, and it is possible to improve the estimation accuracy of the direct ratio.

  Moreover, the interference noise removal apparatus of the present invention can enhance the sound of a sound source close to the microphone even in a noisy environment, and can remove the sound from a far sound source.

The figure which shows an example of the scene using the conventional direct ratio estimation apparatus. The figure which shows the propagation path of the sound indoors. The figure which shows the relationship between direct ratio and the distance between microphones. The figure which shows the function structural example of the noise-resistant direct ratio estimation apparatus 100 of this invention The figure which shows the operation | movement flow of the noise-resistant direct ratio estimation apparatus 100. The figure which shows the function structural example of the interference noise removal apparatus 200 of this invention. The figure which shows the operation | movement flow of the interference noise removal apparatus 200. The figure which shows the function structural example of the process target signal production | generation part 43. FIG. The figure which shows the function structural example of the perspective determination apparatus 300 of this invention. The figure which shows the function structural example of the sound source distance estimation apparatus 400 of this invention. The figure which shows the function structural example of the interference noise removal apparatus 500 of this invention. The figure which shows the example of the microphone array of equal intervals arrangement | positioning. The figure which shows the outline | summary which moves a small array. The figure which shows the experimental condition of an effect confirmation experiment. The figure which shows the direct ratio calculated | required by the conventional method and the method of this invention when SNR: 10dB. The figure which shows the direct ratio calculated | required with the method of this invention and the method of this invention when SNR: 20dB.

  Embodiments of the present invention will be described below with reference to the drawings. The same reference numerals are given to the same components in a plurality of drawings, and the description will not be repeated. In the following description, the symbols “￣”, “^”, etc. used in the text should be written directly above the immediately preceding character, but immediately after the character due to restrictions on the text notation. It describes. In the formula, these symbols are written in their original positions.

FIG. 4 shows an example of the functional configuration of the noise-resistant direct ratio estimation device 100 of the present invention. The operation flow is shown in FIG. The noise-to-noise direct ratio estimation device 100 includes a plurality of frequency domain conversion units 42 _{1 to} 42 _M and a direct ratio estimation unit 44. Each of the plurality of frequency domain transform unit 42 ₁ through 42 _M, the received sound signals received sound by a plurality of microphones m ₁ ~m _M constituting the microphone array 41 is input. The direct ratio estimating unit 44 includes a spatial correlation matrix calculating unit 441, a signal power estimating unit 442, and a direct ratio calculating unit 443. Each unit and each means of the noise-to-noise ratio estimation apparatus 100 are realized by a predetermined program being read into a computer composed of, for example, a ROM, a RAM, a CPU, etc., and executed by the CPU. It is.

A plurality of frequency domain transform section 42 _1, ..., 42 _M converts a plurality of microphones m _1, ... received sound signals are received sound in m _M x _m (n) to a frequency domain signal (step S42). The frequency domain converters 42 ₁ ,..., 42 _M sample the received sound signal x _m (n), for example, at a sampling frequency of 16 kHz and convert it into a digital signal, for example, 256 samples as one frame. In Step S42, discrete Fourier transform is performed to output a frequency component X _m (ω, l) (step S42). ω is a frequency, and l is a frame number. An A / D converter that converts the received sound signal x _m (n) into a digital signal is omitted.

Spatial correlation matrix calculating unit 441, a plurality of frequency domain transform means 42 _1, ..., 42 signal X _{1 in} the frequency domain _M is output (ω, l), ..., X M (ω, l) as an input, the frequency The region signals X ₁ (ω, l),..., X _M (ω, l) are vectorized, and the spatial correlation matrix R (ω, l) shown in equation (1) is calculated using the input signals (step) S441).

  Here, T is a matrix transposition, H is a conjugate transposition, and L is the number of frames for which an average is obtained. The spatial correlation matrix R (ω, l) is input to the signal power estimation unit 442.

The signal power estimation unit 442 includes each component R _ij (ω, l) of the spatial correlation matrix R (ω, l) output from the spatial correlation matrix calculation unit 441, a microphone arrangement of a microphone array given in advance, and a sound source. Components R _d (ω) (formula (3)), matrix R _r (ω) (formula (4)), and matrix R _n (ω) (formula (5)), d _ij (Ω), r _ij (ω), and n _ij (ω) are respectively used as a matrix A (ω) shown in Expression (6) and B (ω) shown in Expression (7). . The matrix R _n (ω) (formula (5)) is an M × M unit matrix.

Here, D _mn is the distance between the m-th microphone and the n-th microphone, and θ is the direction of the sound source viewed from the front of the microphone array. Here, the shape of the microphone array is a linear arrangement, and the front of the microphone array means the normal direction of a straight line in which the microphones are arranged.

Then, simultaneous equations _Pd (ω, l), reverberant power P _r (ω, l), and noise power P _n (ω, l) are established by solving the simultaneous equations shown in equation (8). The vector P (ω, l) (equation (9)) composed of l) is obtained, and the direct sound power P _d (ω, l) and the reverberation power P _r (ω, l) are output.

Note that the matrix R _d (ω) in the case where the arrangement of the microphone array is other than a straight line can be expressed in the form shown in the more general expression (10).

Here, D _mn (θ) ￣ represents a distance difference between the m-th microphone and the n-th microphone when viewed from the direction of the angle θ °. Further, the derivation of the solution of the simultaneous equations of Expression (8) is performed by, for example, converting the pseudo inverse matrix A ⁺ (ω) (Expression (11)) of A (ω) to B (ω, l as shown in Expression (12). ) From the left side.

The direct ratio calculation means 443 calculates the direct ratio E _R (l) from the direct sound power P _d (ω, l) and the reverberant sound power P _r (ω, l) according to the equation (13) and outputs it. .

  In the method according to the first embodiment, since the direct ratio is directly obtained without the noise power, it is possible to accurately estimate the direct ratio. This noise-to-noise direct ratio estimation device 100 can be used for an interference noise removal device. FIG. 6 shows a functional configuration example of the interference noise removal apparatus 200 including the noise-resistant direct ratio estimation apparatus 100. The operation flow is shown in FIG.

The interference noise removal apparatus 200 includes a single microphone array 41, a noise-to-noise ratio ratio estimation apparatus 100, a processing target signal generation unit 43, a target signal adjustment unit 45, and an inverse frequency domain conversion unit 46. To do. The noise-to-noise direct ratio estimation apparatus 100 is the same as that already described with reference to FIG. 4 including a plurality of frequency domain conversion units 42 ₁ ,..., 42 _M and a direct ratio estimation unit 44. Each functional configuration unit excluding the microphone array 41 is realized by a predetermined program being read into a computer including, for example, a ROM, a RAM, and a CPU, and the CPU executing the program.

Microphone array 41 is a plurality of microphones m _1, consisting of ... m _M. A plurality of frequency domain transform section 42 _1, ..., 42 _M, a plurality of microphones m _1, ... m _M received sound has been received sound signal x _m (n) are inputted, respectively, each received sound signal in the frequency domain It converts into a signal (step S42).

Processing signal generating unit 43, a plurality of frequency domain transform section 42 _1, ..., generates a 42 _M signals X _m of frequency domain output (omega, l) synthesizing the processing object signal Y (ω, l) (Step S43). The noise-to-noise direct ratio estimation apparatus 100 performs the same operation as described above, and calculates and outputs the direct ratio E _R (ω, l) (step S44). However, in the interference noise elimination apparatus described here, the denominator of the equation (13) and the equation (14) excluding the sum calculation in the numerator are used as the direct ratio.

The target signal adjustment unit 45 receives the processing target signal Y (ω, l) and the direct ratio E _R (ω, l) as input and adjusts the amplitude of the processing target signal Y (ω, l) according to the values. A post-processing signal Z (ω, l) is generated (step S45).

The inverse frequency domain transform unit 46 transforms the processed signal Z (ω, l) into a time domain signal z (n) (step S46). The operations from step S42 to step S46 are continued until all sound reception signals x _m (n) are completed.

Here, adjustment according to the value of the direct ratio E _R (ω, l) means threshold processing of E _R (ω, l) and the amplitude of the post-processing signal Z (ω, l) as the value increases. And processing such as decreasing the amplitude of the post-processing signal Z (ω, l) as the value increases. Details will be described later.

  With the above operation, noise removal is performed by, for example, emphasizing only sounds within a specific distance range and suppressing and collecting sounds outside the range by one microphone array. Hereinafter, the present invention will be described in more detail by showing more specific functional configuration examples of the respective units.

[Processing signal generator]
FIG. 8 shows a more specific functional configuration example of the processing target signal generation unit 43. The processing target signal generation unit 43 includes a plurality of weight multiplication units 431 _{1 to} 431 _M and an addition unit 432. The plurality of weight multiplying means 431 _{1 to} 431 _M are respectively frequency components X ₁ (ω, l),..., X _M (ω) of the plurality of received signals x _m (n) received by M microphones. , L) is multiplied by a weighting factor w _m (ω).

For the weights used in the weight multiplication means 431 _{1 to} 431 _M , for example, when M microphones are omnidirectional, w _m = 1 / M so that all frequency components X ₁ (ω, l), .., X _M (ω, l) is averaged to stabilize the processing target signal Y (ω, l). Also, when M microphones have directivity, use only a specific microphone signal by setting w ₁ = 1, w _m = 0 (m = {2,..., M}). Can do. For example, using a method such as that described in the reference “Oga, Yamazaki, Kanada,“ Sound System and Digital Signal Processing ”published by the Institute of Electronics, Information and Communication Engineers”, using filter coefficients for weighted beamforming, Arbitrary directivity can be formed with a microphone array.

The adding means 432 adds all the frequency components X ₁ (ω, l),..., X _M (ω, l) multiplied by the weights and outputs the processing target signal Y (ω, l).

[Target signal adjustment section]
The target signal adjustment unit 45 can be configured by, for example, a filter coefficient calculation unit 451 and a multiplication unit 452 (FIG. 6). The filter coefficient calculation unit 451 calculates and outputs a filter coefficient G (ω, l) with the direct ratio E _R (ω, l) as an input. For the calculation of the filter coefficient G (ω, l), for example, a binary filter using a threshold value as shown in Expression (15) is used.

The threshold value Th can be set to any value between the minimum value and the maximum value of the direct ratio E _R (ω, l). The sound quality is improved when the threshold Th is brought close to the minimum value (0). On the contrary, when the threshold value Th is brought close to the maximum value, the noise suppression effect is enhanced, but the distortion of the received sound signal is increased and the sound quality is deteriorated.

  Thus, the threshold Th has a trade-off relationship between the sound quality and the noise suppression. Therefore, the threshold Th is determined empirically in accordance with the purpose of use in consideration of this trade-off relationship.

Further, when calculating the filter coefficient G (ω, l), as shown in the equation (16), if a time frequency band in which the value of the direct ratio falls below the threshold Th ₂ is emphasized, it is farther than a specific distance range. The sound source can be emphasized.

  In addition, although the binary filter of 0 or 1 was mentioned as an example of filter coefficient G ((omega), l), filter coefficient G ((omega), l) does not necessarily need to be 0 and 1, for example, 0.1 And a sufficiently different value such as 0.9.

  Further, a real number of 1 or more may be set for the filter coefficient G (ω, l). That is, the processing target signal Y (ω, l) may be amplified. Alternatively, the processing target signal Y (ω, l) may be greatly suppressed by setting the value to 0.1 or less.

The multiplication coefficient 452 multiplies the processing target signal Y (ω, l) by the filter coefficient G (ω, l) obtained in this way, and the processed signal Z (ω, l) = G (ω, l) · Y (ω, l) is generated. Therefore, the post-processing signal Z (ω, l) can be composed of only the processing target signal Y (ω, l) having a large direct ratio E _R (ω, l). That is, only the direct sound can be extracted.

As a second embodiment of the present invention, a perspective determination apparatus 300 that determines the perspective of a sound source using the direct ratio E _R (l) described in the first embodiment will be described. FIG. 9 shows a functional configuration example of the perspective determination device 300. The perspective determination device 300 includes a microphone array 41, a noise-to-noise direct ratio estimation device 100, and a perspective determination unit 121. The microphone array 41 and the noise-to-noise ratio estimation device 100 are the same as those of the interference noise removal device 200. The perspective determination device 300 is also realized by a predetermined program being read into a computer including, for example, a ROM, a RAM, and a CPU, and the CPU executing the program.

  The perspective determination device 300 determines whether a sound source of a sound received at a certain time is far or near when sound sources at a plurality of different distances sound at different times. The perspective determination unit 121 included in the perspective determination device 300 includes an accumulation unit 1211 and a determination unit 1212.

The accumulating unit 1211 accumulates the direct ratio E _R for the past L time frames, and outputs the comparison direct ratio E ^. Compared Chokkan ratio E ^ in, for example stored Chokkan ratio average value E of _{E R (l) ^ = 1} / LΣ l L E R (l) and the average value E of the minimum and maximum values ^ = 1/2 (max E _R (l) + min E _R (l)) or the like is used.

The determination unit 1212 compares the direct ratio E _R (l) with the comparison target direct ratio E ^, and indicates that the distance is close to the perspective determination result Y _l when E _R (l)> E ^. 1, when E _R (l) <E ^ outputs 0 for example indicating that distance is long in the distance determination result Y _l. The distance determination result Y _l is nearest received sound signals of the past L time period is either a sound from relatively close sound source or those indicating which sounds from a relatively distant sound source.

The distance determination result by using a Y _l, the received sound signal inputted sequentially, it is possible to isolate the distance between the between the microphone and the sound source. That is, sounds from a plurality of sound sources can be selected according to the distance from the microphone.

FIG. 10 shows a functional configuration example of the sound source distance measuring apparatus 400 of the present invention. Sound source distance measuring apparatus 400, and one microphone array 41, the noise immunity Chokkan ratio estimation device 100, the distance - Chokkan ratio database (hereinafter, the distance - referred Chokkan ratio DB) 4 and 7, the distance determination unit 4 8, comprising a. The noise-to-noise direct ratio estimation apparatus 100 includes a plurality of frequency domain conversion units 42 ₁ ,..., 42 _M and a direct ratio estimation unit 44. Each functional configuration unit excluding the microphone array 41 is realized by a predetermined program being read into a computer including, for example, a ROM, a RAM, and a CPU, and the CPU executing the program.

Microphone array 41 is a plurality of microphones m _1, consisting of ... m _M. A plurality of frequency domain transform section 42 _1, ..., 42 _M, a plurality of microphones m _1, ... m _M received sound has been received sound signal x _m (n) are inputted, respectively, each received sound signal in the frequency domain Convert to signal. The frequency domain transform units 42 ₁ ,..., 42 _M perform discrete Fourier transform for each frame and output a frequency component X _m (ω, l).

Chokkan ratio estimation unit 44, a plurality of frequency domain transform section 42 _1, ..., 42 signal _in the frequency domain _m outputs X _m (ω, l) to estimate the Chokkan ratio E _R of the received sound signals as input .

Distance - Chokkan ratio DB4 7 records the relationship between the distance between the Chokkan ratio E _R and the microphone array and sound source. Distance determining section 4-8, the distance as an input Chokkan ratio - Chokkan ratio DB4 7 reference to estimate the sound source distance estimate d ^ corresponding to its Chokkan ratio.

Some received signals have components concentrated in a specific frequency band. When the direct ratio E _R of such a sound reception signal is calculated by the direct ratio calculation means 443, the estimation accuracy of the direct ratio E _R deteriorates.

  Therefore, as shown in the equation (17), by using the direct ratio calculation means 443 ′ (FIG. 4) for calculating the direct ratio E in a specific frequency region Ω, the accuracy of the direct ratio is improved. I can do it.

Here, the frequency region Ω is determined, for example, by selecting a frequency band in which signal components are concentrated. For example, among the outputs X _m (ω, l) of the frequency domain converter 42 _m connected to an arbitrary m-th microphone, the absolute value of X _m (ω, l) is preliminarily set as shown in the equation (18). It is determined by selecting the frequency ω having a value larger than the set threshold value P _th or by selecting the frequency ω from the largest absolute value of X _m (ω, l) to the Kth.

Here, P _th is, for example _{| X m (ω, l)} | of an average value of all the frequency used.

  FIG. 11 shows a functional configuration example of the interference noise removing apparatus 500 of the present invention. The interference noise removal apparatus 500 includes the noise-to-noise direct ratio estimation apparatus 100 described in the first embodiment, a processing target signal generation unit 72, a target signal adjustment unit 73, and an inverse frequency domain conversion unit 74.

The processing target signal generation unit 72 receives the frequency domain signals X _m (ω, l) output from the plurality of frequency domain conversion units 42 _{1 to} 42 _M in the noise-to-noise direct ratio estimation apparatus 100 as input. (Ω, l) is output. The processing target signal X (ω, l) is a signal obtained by synthesizing the frequency domain signal X _m (ω, l) with, for example, an adding means (not shown). Before the addition, the signal X _m (ω, l) in each frequency domain may be multiplied by a weight.

  The target signal adjustment unit 73 receives the direct ratio E (ω) output from the noise-to-noise direct ratio estimation device 100 and the processing target signal X (ω, l) output from the processing target signal generation unit 72 as inputs. A post-processing signal Y (ω, l) in which the amplitude of the target signal X (ω, l) is adjusted is generated. The inverse frequency domain transform unit 74 transforms the processed signal Y (ω, l) into a time domain signal y (n).

The target signal adjustment unit 73 includes, for example, a distance calculation unit 7 3 1, a filter formation unit 7 3 2, and a multiplication unit 7 3 3. The distance calculation means 7 3 1 obtains a functional expression d = f (E _R (ω, l)) indicating the relationship between the distance between the microphone array 41 and the sound source and the direct ratio E _R (ω, l). A sound source distance estimated value d ^ corresponding to the input direct ratio E is calculated.

Filter forming means 7 3 2, as shown in equation (19), the sound source distance estimate d ^ is, emphasize the two time frequency components takes a value between the different sizes threshold d _f and d _n so And a filter for emphasizing only the sound source in the band-like region in the two distance sections is formed.

  Here, l and ω of G (ω, l) are the L frames obtained by averaging the equation (1) in the spatial correlation matrix calculation means 431 and the direct values of the processing of the direct ratio estimation unit 43 described above. The same G (ω, l) is multiplied to all the frequencies included in the frequency Ω (equation (17)) averaged by the interval ratio calculation means 443. Further, in the equation (19), the value of G (ω, l) is not necessarily 1 and 0, and may be a value that is sufficiently different, for example, 0.9 and 0.1.

The multiplying unit 7 3 3 multiplies the processing target signal X (ω, l) by the filter G (ω, l) to generate a processed signal Y (ω, l). Therefore, the post-processing signal Y (ω, l) is obtained by enhancing or suppressing the sound of the sound source located in a specific distance range from the microphone array 41 in two distance sections. The post-process signal Y (ω, l) is converted into a time domain signal y (n) by the inverse frequency domain converter 73.

  The spatial correlation matrix R (ω, l) in the above-described embodiment is based on the average value of the number of frames L, as is apparent from the equation (1). Therefore, when the position of the sound source moves, the direct ratio cannot be obtained accurately. Therefore, a noise-resistant direct ratio estimation apparatus 600 that can accurately determine the direct ratio even when the position of the sound source moves will be described.

  In this embodiment, for example, a microphone array 130 having an equal interval as shown in FIG. 12 is used. The functional configuration of the direct ratio estimation unit 44 according to this embodiment is the same as that shown in FIG.

The signal power estimation means 442 ′ (FIG. 4) gives in advance each component R ′ _{i, j} (ω, l) of the small spatial correlation matrix R ′ (ω, l) output from the spatial correlation matrix calculation means 441 ′. And the matrix R _d (ω) given by the direction of the sound source (formula (3)), the matrix R _r (ω) (formula (4)) and the matrix R _n (ω) ( A matrix A (ω) shown in Expression (20), which is composed of each component of Expression (5), d _{i, j} (ω), r _{i, j} (ω), and n _{i, j} (ω). ) And B (ω, l) shown in equation (21). Here, the small spatial correlation matrix R ′ (ω, l) is a matrix obtained by the sum of the spatial correlation matrices obtained for each small microphone array (formula (22)).

However, R ₁₁ ′ (ω, l), R ₁₂ ′ (ω, l), R ₂₁ ′ (ω, l), R ₂₂ ′ (ω, l), which are the components of B (ω, l), It is obtained by R ′ (ω, l) shown in the equation (22).

However,

  Expressions (22) and (23) calculate a small spatial correlation matrix obtained by the sum of the spatial correlation matrices when adjacent microphones are moved as two small arrays as shown in FIG. That is, a small array including two adjacent microphones is moved (130a → 130b →... → 130g) to obtain the sum of spatial correlation matrices. When the number of microphones is M ′, Expression (20) can be expressed by Expression (24), Expression (21) can be expressed by Expression (25), and Expression (22) can be expressed by Expression (26).

Then, simultaneous equations _Pd (ω, l), reverberant power P _r (ω, l), and noise power P _n (ω, l) are established by solving the simultaneous equations shown in equation (8). The vector P (ω, l) (equation (9)) composed of l) is obtained, and the direct sound power P _d (ω, l) and the reverberation power P _r (ω, l) are output.

  The direct ratio calculation means 443 performs exactly the same processing as in the first embodiment. As described above, by calculating the small spatial correlation matrix obtained by the sum of the spatial correlation matrices obtained for each small array as in the spatial correlation matrix calculating means 441 ′, it is possible to obtain an accurate straight line for a moving sound source. The ratio can be determined.

  Although the number of microphones constituting the small array has been described with two examples, the number may be any number. Further, the arrangement of the microphones is not limited to a linear array arranged linearly at equal intervals. Any microphone array may be used as long as a microphone is provided at a position overlapping by translation of a small microphone array composed of a plurality of microphones arranged in a regular rule, such as a rectangular planar array, a triangular planar array, and a rectangular parallelepiped array.

〔Experimental result〕
For the purpose of confirming the effect of the present invention, an experiment was performed in which the direct ratio was estimated using signals received by the microphone array when white noise was emitted from the sound source 21 and compared with the conventional method.

  FIG. 14 shows the simulation conditions. A room having a plane size of 4 × 6 m and a height of 2.7 m was assumed. FIG. 14 is a view of the room as viewed from above. A microphone array in which eight microphones were arranged in a circle with a radius of 6 cm was used. The microphone array was placed at a height of 1.5m from the floor. The sound source 21 was arranged at a height of 1.5 m in the direction of an angle of 10 ° from the center axis of the circle. The reverberation time of the room is about 550 ms, the sampling frequency is 16 kHz, and the length of one frame in the processing is 512 samples.

  FIG. 15 shows the time when SNR: 10 dB, and FIG. 16 shows the time when SNR: 20 dB. (A) is the conventional method, and (b) is the present invention. The horizontal axis is the distance (cm), and the vertical axis is the direct ratio (dB). ○ The solid line indicates the estimated direct ratio, and ● the broken line indicates the correct direct ratio.

  Comparing FIGS. 15 and 16 (a) and (b), it can be seen that the present invention has less error. For example, the estimation error of the direct ratio with respect to a sound source with a SNR of 10 dB (FIG. 15) at a distance of 10 cm is about 5 dB in the noise-resistant direct ratio estimation apparatus according to the present invention, while the conventional technique has 10 dB. Further, comparing FIG. 15 and FIG. 16 where the noise power is 1/10, it can be seen that there is almost no difference between FIG. 15 (b) and FIG. 16 (b). This result shows that the method of the present invention is not easily affected by the magnitude of the superimposed noise. As described above, according to the noise-resistant direct ratio estimation method according to the present invention, there is an effect of accurately estimating the direct ratio even when uncorrelated noise is superimposed on the microphone.

  Note that the processes described in the above method and apparatus are not only executed in time series according to the order of description, but may also be executed in parallel or individually as required by the processing capability of the apparatus that executes the processes. good. Further, an interference noise removal system or the like may be constructed by using two or more microphone arrays on which the above apparatus and method are mounted.

  Further, when the processing means in the above apparatus is realized by a computer, the processing contents of functions that each apparatus should have are described by a program. Then, by executing this program on the computer, the processing means in each apparatus is realized on the computer.

  The program describing the processing contents can be recorded on a computer-readable recording medium. As the computer-readable recording medium, for example, any recording medium such as a magnetic recording device, an optical disk, a magneto-optical recording medium, and a semiconductor memory may be used. Specifically, for example, as a magnetic recording device, a hard disk device, a flexible disk, a magnetic tape or the like, and as an optical disk, a DVD (Digital Versatile Disc), a DVD-RAM (Random Access Memory), a CD-ROM (Compact Disc Read Only). Memory), CD-R (Recordable) / RW (ReWritable), etc., magneto-optical recording medium, MO (Magneto Optical disc), etc., semiconductor memory, EEP-ROM (Electronically Erasable and Programmable-Read Only Memory), etc. Can be used.

  The program is distributed by selling, transferring, or lending a portable recording medium such as a DVD or CD-ROM in which the program is recorded. Further, the program may be distributed by storing the program in a recording device of a server computer and transferring the program from the server computer to another computer via a network.

  Each means may be configured by executing a predetermined program on a computer, or at least a part of these processing contents may be realized by hardware.

Claims (8)

A plurality of frequency domain converters for receiving received signals received by a plurality of microphones and converting the received signals into frequency domain signals, and direct sound power and reverberation using the frequency domain signals as inputs. A noise-tolerant direct ratio estimation device comprising a direct ratio estimator for calculating a direct ratio obtained by calculating a sound power and a noise power and dividing the direct sound power by the reverberant power,
The direct ratio estimator is
A spatial correlation matrix calculating means for calculating a spatial correlation matrix by vectorizing a signal in the frequency domain with the frequency domain signal output from the plurality of frequency domain transform units as an input;
A vector composed of direct sound power, reverberant sound power and noise power is obtained from the microphone arrangement information given in advance and the spatial correlation matrix, and the direct sound power and reverberation among the vector elements are obtained. Signal power estimation means for outputting the power of sound;
Direct ratio calculation means for calculating the direct ratio obtained by dividing the power of the direct sound by the power of the reverberant sound;
A noise-to-noise direct ratio estimation apparatus comprising: An interference noise elimination apparatus including the noise-to-noise ratio ratio estimation apparatus according to claim 1,
A microphone array composed of a plurality of microphones, each of which receives a received sound signal received by the plurality of frequency domain converters;
A processing target signal generation unit that generates a processing target signal by combining frequency domain signals output from the plurality of frequency domain conversion units;
A target signal adjustment unit that generates a processed signal in which the amplitude of the processing target signal is adjusted to be larger as the direct ratio is larger with the processing target signal and the direct ratio as an input;
An inverse frequency domain transform unit for transforming the processed signal into a time domain signal;
An interference noise removing apparatus comprising: An interference noise elimination apparatus including the noise-to-noise ratio ratio estimation apparatus according to claim 1,
A microphone array composed of a plurality of microphones, each of which receives a received sound signal received by the plurality of frequency domain converters;
A processing target signal generation unit that generates a processing target signal by combining frequency domain signals output from the plurality of frequency domain conversion units;
A target signal adjustment unit that generates a post-processing signal in which the amplitude of the processing target signal is adjusted to be larger as the direct ratio is smaller, with the processing target signal and the direct ratio being input;
An inverse frequency domain transform unit for transforming the processed signal into a time domain signal;
An interference noise removing apparatus comprising: A perspective determination device including the noise-to-noise direct ratio estimation device according to claim 1 and including a perspective determination unit,
The perspective determination unit
Frequency averaging means for averaging the direct ratio in the frequency direction and outputting the frequency average direct ratio;
Accumulating means for accumulating frames for a predetermined period of time in the frequency average direct ratio and outputting the comparative direct ratio;
A determination means for comparing the frequency average direct ratio and the comparison target direct ratio and outputting a perspective determination result;
A perspective determination device comprising: A sound source distance measuring device including the noise-to-noise direct ratio estimating device according to claim 1,
A microphone array composed of a plurality of microphones, each of which receives a received sound signal received by the plurality of frequency domain converters;
A distance-direct ratio database that records the relationship between the direct ratio and distance;
A distance determination unit that estimates the sound source distance estimate corresponding to the direct ratio by referring to the distance-direct ratio database using the direct ratio as an input;
A sound source distance measuring device comprising: An interference noise elimination apparatus including the noise-to-noise ratio ratio estimation apparatus according to claim 1,
A processing target signal generation unit that outputs a processing target signal with the frequency domain signals output by the plurality of frequency domain conversion units;
Processing that emphasizes or suppresses the sound of a sound source located in a specific distance range from the microphone array composed of the plurality of microphones, with the direct ratio output from the noise-tolerant direct ratio estimation device and the processing target signal as inputs. A target signal adjustment unit for generating a post signal;
An inverse frequency domain transform unit for transforming the processed signal into a time domain signal;
An interference noise removing apparatus comprising: A plurality of frequency domain conversion units, a frequency domain conversion process for converting a received signal received by a plurality of microphones into a frequency domain signal;
A spatial correlation matrix calculation unit, wherein the spatial correlation matrix calculation unit calculates a spatial correlation matrix by vectorizing the frequency domain signal by inputting the frequency domain signal output from the plurality of frequency domain transform units;
A signal power estimation unit obtains a vector composed of direct sound power, reverberant sound power and noise power from the microphone arrangement information given in advance and the spatial correlation matrix, and among the vector elements, A signal power estimation process that outputs the power of the direct sound and the power of the reverberant sound,
The direct ratio calculation unit calculates the direct ratio obtained by dividing the power of the direct sound by the power of the reverberant sound; and
A noise-to-noise ratio estimation method including: The noise-to-noise direct ratio estimation device according to claim 1, the interference noise elimination device according to claim 2, claim 3, or claim 6 , or the perspective determination device according to claim 4, or claim An apparatus program for causing a computer to function as the sound source distance measuring apparatus according to Item 5 .

Images