JP2009060467A - System and method for reproducing sound field - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reproduce an original sound field in which a sound field can not be specified, in a space different from that of the original sound field extremely faithfully. <P>SOLUTION: A covariance matrix R1 of acoustic signals collected by microphones M4, M5 installed at target sound field points T1, T2 in an original sound field 10 is calculated and in a regenerative sound field 20 which is a space different from that of the original sound field 10, a covariance matrix R2 of acoustic signals collected by microphones M6, M7 installed at reproduction points P1, P2 reflected with a geometrical positional relation of target sound field points T1, T2 is calculated. Such a coefficient K as to make the covariance matrix R2 approximately equal to the covariance matrix R1 is estimated, and acoustic signals collected by microphones M1-M3 installed in the original sound field 10 are multiplied by the estimated coefficient K, and output from speakers S1-S3 installed in the regenerative sound field 20. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a sound field reproduction system and a sound field reproduction method for reproducing a sound field of one environment in another environment.

Conventionally, various methods for reproducing a sound field of one environment in another environment have been devised (for example, see Patent Document 1). As described in Patent Document 1, an acoustic signal output from a measurement speaker, which is a sound source placed in a measurement environment such as a hall, is collected by a measurement microphone, and based on the acoustic measurement result. By using the obtained transfer function to reproduce sound in a space different from the measurement environment, it is possible to reproduce a reverberation feeling specific to the measurement environment, a sense of localization of a virtual sound image, and the like.
JP 2007-124023 A

  However, since it is assumed that a sound source is specified in a measurement environment (original sound field) in order to use a transfer function as described in Patent Document 1, an original sound field where a sound source cannot be specified is assumed. There is a problem that cannot be reproduced in another space.

  Therefore, the present invention has been proposed in view of the above-described circumstances, and a sound field reproduction that can reproduce an original sound field in which a sound source cannot be specified very faithfully in a different space from the original sound field. It is an object to provide a system and a sound field reproduction method.

  The present invention calculates a covariance matrix (hereinafter referred to as a first covariance matrix) of an acoustic signal picked up at a target sound field point of the original sound field, and reproduces the sound as a different space from the original sound field. In the sound field, a covariance matrix (hereinafter referred to as a second covariance matrix) of the acoustic signal collected at the reproduction point reflecting the geometric positional relationship between the target sound field points is calculated. A coefficient K is estimated such that the second covariance matrix is substantially equal to the first covariance matrix, and an acoustic signal obtained by multiplying the acoustic signal collected in the original sound field by the estimated coefficient K is reproduced in the reproduction sound field. Output to.

  According to the present invention, the first covariance matrix of the acoustic signal collected at the target sound field point of the original sound field and the geometrical relationship between the target sound field points in the reproduction sound field that is a different space from the original sound field. Since the sound signal collected at the playback point that reflects the positional relationship is equal to the second covariance matrix of the acoustic signal and output to the playback sound field, the original sound field where the sound source cannot be specified, It can be reproduced very faithfully in a different space from the original sound field.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. First, the sound field reproduction system shown as an embodiment of the present invention will be described with reference to FIG. FIG. 1 is a diagram showing a conceptual configuration of a sound field reproduction system.

  As shown in FIG. 1, target sound field points T1 and T2 to be reproduced in the reproduction sound field 20 are set in the original sound field 10, and the target sound field points T1 and T2 are set to the target sound field points T1 and T2. The microphones M4 and M5 for collecting the acoustic signal of the original sound field 10 are installed. The acoustic signals collected by the microphones M4 and M5 are input to the calculation unit 11.

  Here, only the target sound field points T1 and T2 are set as the target sound field points, but two or more, that is, a plurality of target sound field points are set so that mutual spatial acoustic features can be extracted. Is possible. Increasing the target sound field point leads to a cost increase due to an increase in the number of microphones installed to collect the sound signal, but the spatial acoustic features of the target sound field point can be reproduced more faithfully.

  In the original sound field 10, microphones M <b> 1 to M <b> 3 that collect the acoustic signal of the original sound field 10 are installed at arbitrary positions other than the target sound field points T <b> 1 and T <b> 2. The number of microphones installed at any position other than the target sound field point in the original sound field 10 may be at least two. The acoustic signals collected by the microphones M <b> 1 to M <b> 3 are input to the calculation unit 12.

  Next, the reproduction sound field 20 will be described. As shown in FIG. 2, the reproduction sound field 20 reflects the geometrical positional relationship between the target sound field points T1 and T2, and reproduces the spatial acoustic features of the target sound field points T1 and T2. P1 and P2 are set, and microphones M6 and M7 for collecting sound signals of the reproduction sound field 20 at the reproduction points P1 and P2 are installed at the reproduction points P1 and P2. The acoustic signals collected by the microphones M6 and M7 are input to the calculation unit 11.

  Reflecting the geometric positional relationship between the target sound field points T1 and T2 means that the positional relationship between the target sound field points T1 and T2 is also maintained at the reproduction points P1 and P2. If the distance between the sound field points T1 and T2 is 50 cm, the distance between the reproduction points P1 and P2 is also 50 cm. When there are two or more target sound field points, for example, when three points are provided, the geometrical positional relationship between the target sound field points formed by the three target sound field points is reflected at the playback point. There is a need. Therefore, the number of target sound field points provided in the original sound field 10 and the number of reproduction points provided in the reproduction sound field 20 are necessarily the same.

  The reproduction points P1 and P2 set in the reproduction sound field 20 should be set at arbitrary positions in the reproduction sound field 20 as long as the geometric positional relationship between the target sound field points T1 and T2 is reflected in this way. Can do.

  Further, the reproduction sound field 20 corresponds to each of the microphones M1 to M3 installed in the original sound field 10, and the spatial sound of the target sound field points T1 and T2 of the original sound field 10 is controlled by the output control of the calculation unit 12. Speakers S1 to S3 are provided that can output acoustic signals whose characteristics are reproduced at reproduction points P1 and P2 of the reproduction sound field 20. The speakers S <b> 1 to S <b> 3 are installed so as to be independent from each other by providing an interval that is at least half the wavelength of the lowest frequency of the acoustic signal to be reproduced. The speakers S1 to S3 are also provided at intervals such that the reproduction points P1 and P2 are at least half the wavelength of the lowest frequency of the acoustic signal to be reproduced.

  The computing unit 11 is a covariance matrix (hereinafter referred to as covariance matrix) representing the mutual covariance and frequency characteristics of the acoustic signals collected by the microphones M4 and M5 installed at the target sound field points T1 and T2 of the original sound field 10. The matrix is a covariance matrix R1). Similarly, the computing unit 11 similarly uses a covariance matrix (representing the mutual covariance and frequency characteristics of acoustic signals collected by the microphones M6 and M7 installed at the reproduction points P1 and P2 of the reproduction sound field 20). Hereinafter, this covariance matrix is referred to as a covariance matrix R2.) The calculation unit 11 inputs the calculated two covariance matrices R1 and R2 to the calculation unit 12.

  In addition, the calculation unit 11 outputs the acoustic signals collected by the microphones M4 and M5 for each predetermined frequency band (for each octave), for example, for each 1/3 oct (octave) or 1/4 oct (octave). Can also be calculated. In this case, the calculation unit 11 calculates the covariance matrix R2 for the same predetermined frequency band for the acoustic signals collected by the microphones M6 and M7.

  The calculation unit 12 collects sound at the covariance matrix R1 of the acoustic signal collected at the target sound field points T1 and T2 of the original sound field 10 received from the calculation unit 11 and the reproduction points P1 and P2 of the reproduction sound field 20. And a coefficient such that the covariance matrix R2 is substantially equal to the covariance matrix R1 (R1 = R2). Further, the calculation unit 12 performs control so that the estimated coefficient is multiplied by the acoustic signal collected by the microphones M1 to M3 installed in the original sound field 10 and output.

  Further, when the calculation unit 11 calculates the covariance matrices R1 and R2 for each predetermined frequency band by the calculation unit 11, the calculation unit 12 uses the covariance matrix R2 calculated for each predetermined frequency band as the covariance matrix R1. A coefficient that is approximately equal to is estimated.

  Next, the processing operation of the sound field reproduction system of FIG. 1 will be described using the flowchart shown in FIG.

  First, in step ST1, sound is simultaneously generated by microphones M4 and M5 installed at target sound field points T1 and T2 of original sound field 10 and microphones M1 and M3 installed at arbitrary positions other than target sound field points T1 and T2. Pick up the signal. The acoustic signals picked up by the microphones M1 to M5 are respectively input to the calculation unit 11 and recorded.

In step ST <b> 2, the calculation unit 11 calculates a covariance matrix R <b> 1 of acoustic signals collected by microphones M <b> 4 and M <b> 5 installed in the original sound field 10. The covariance matrix R1 indicating the covariance between the acoustic signals collected at the target sound field points T1 and T2 and the frequency characteristics is shown below as p = {1,2} and q = {1,2} ( 1) It can be expressed as: The covariance matrix R1 calculated by the calculation unit 11 is input to the calculation unit 12.

  In step ST <b> 3, the arithmetic unit 12 causes the acoustic signals collected by the microphones M <b> 1 to M <b> 3 installed in the original sound field 10 to be output from the speakers S <b> 1 to S <b> 3 installed in the reproduction sound field 20.

  In step ST4, the microphones M6 and M7 installed at the reproduction points P1 and P2 of the reproduction sound field 20 collect sound signals output from the speakers S1 to S3. The acoustic signals picked up by the microphones M6 and M7 are input to the calculation unit 11 and recorded.

In step ST5, the calculating part 11 calculates the covariance matrix R2 of the acoustic signal collected by the microphones M6 and M7 installed in the reproduction sound field 20. The covariance matrix R2 indicating the covariance between the acoustic signals collected at the reproduction points P1 and P2 and the frequency characteristics is represented by p = {1,2} and q = {1,2} (2) It can be expressed as: The covariance matrix R2 calculated by the calculation unit 11 is input to the calculation unit 12.

  In step ST <b> 6, the calculation unit 12 receives the covariance matrix R <b> 1 of the acoustic signal collected from the target sound field points T <b> 1 and T <b> 2 of the original sound field 10 received from the calculation unit 11 and the reproduction points P <b> 1 and P <b> 2 of the reproduction sound field 20. Is compared with the covariance matrix R2 of the acoustic signal collected in step (2), and a coefficient K is estimated such that the covariance matrix R2 is substantially equal to the covariance matrix R1 (R1 = R2).

  For example, the calculation unit 12 estimates the coefficient K using the steepest descent method so that the similarity between the covariance matrix R1 and the covariance matrix R2 is minimized.

Here, the steepest descent method will be briefly described. First, consider a function f (x) in which a variable is represented by an n-dimensional vector x = (x ₁ , x ₂ ,..., X _n ). The steepest descent method is one of calculation procedures which considers this function f (x) as an error function and derives the function value to the minimum.

FIG. 3 is a diagram schematically showing a curved surface of the error function. Here, if the initial value of the variable is set to x ⁽⁰⁾ , the error curved surface can be lowered by changing the variable from this point in the direction opposite to the gradient of the function. This procedure is performed until the error converges. In addition, the arrow Y1 in FIG. 3 has shown a mode that it moves to the reverse direction of a gradient, and the arrow Y2 has shown a mode that the step width is optimized.

In equation (3), α (> 0) is the step width. The step width is not a constant value, and an appropriate value needs to be set every time the variable is updated. In determining α, an initial value is first given, and α ^(m) is updated so as to satisfy the following expression (4).

  This operation itself is also a problem of minimizing one variable. For example, if the initial value at m = 0 is set appropriately large and the error continues to be reduced by 1/2, the step can be easily performed. The width can be determined.

By applying the steepest descent method, the coefficient K is estimated so that the covariance matrix R2 is substantially equal to the covariance matrix R1 (R1 = R2) if the following equation (5) is substantially satisfied. be able to. N represents the length of “row” and “column” of the covariance matrix R1 or R2.

  In step ST <b> 7, the calculation unit 12 performs control so that the estimated coefficient K is multiplied by the acoustic signal collected by the microphones M <b> 1 to M <b> 3 installed in the original sound field 10 and output.

  As described above, the sound signals output from the speakers S <b> 1 to S <b> 3 by the output control of the sound signal using the coefficient estimated by the calculation unit 12 are generated at the reproduction points P <b> 1 and P <b> 2 of the reproduction sound field 20. The spatial acoustic features of the target sound field points T1 and T2 can be faithfully reproduced.

  In addition, when the covariance matrices R1 and R2 are calculated for each predetermined frequency band and the coefficient K is estimated, the coefficients K are divided into predetermined frequency bands in steps ST2 and ST5 of the flowchart shown in FIG. The processing steps of calculating the dispersion matrices R1 and R2 and estimating the coefficient K such that the covariance matrix R2 is substantially equal to the covariance matrix R1 in step ST6 may be repeated for each predetermined frequency band.

  As described above, the sound signals output from the speakers S1 to S3 by the output control of the sound signals using the coefficients for the predetermined frequency bands estimated by the calculation unit 12 are the reproduction points P1 and P2 of the reproduction sound field 20. The spatial acoustic features of the target sound field points T1 and T2 of the original sound field 10 can be reproduced more faithfully, and the reproducibility can be improved.

  Next, an embodiment in which the above-described sound field reproduction system is actually implemented will be described. Specifically, a lobby with a colonnade in a large facility with heavy traffic and a lot of environmental noise is set as the original sound field 10, and an anechoic room is set as the reproduction sound field 20, and the sound field reproduction system as described above is set. To reproduce the spatial acoustic feature of the target sound field point of the original sound field 10 at the playback point of the reproduced sound field 20.

  With the sound field reproduction system constructed as described above, a sound field reproduction including the collected environmental noise is performed, and a series of processing procedures for evaluating the execution result will be described with reference to the flowchart shown in FIG.

  First, in step ST11, a covariance matrix is prepared and environmental noise is collected.

Specifically, in the lobby with a colonnade in the large-scale facility where the above-mentioned traffic of the above-mentioned sound source 10 is intense and there is a lot of environmental noise, an interval of 2 m is provided from the central position C1 shown in FIG. Six microphones M _{1 to} M ₆ for measuring directivity of all fingers (hereinafter simply referred to as microphones M _{1 to} M ₆ ) are installed at separated positions.

As shown in FIG. 5, target sound field points T <b> 3 to T <b> ₆ that are desired to be reproduced in the reproduction sound field 20 are located in the central area shown in FIG. 5 in the region surrounded by the microphones M _{1 to} M ₆ . They are set at 0.15 m intervals on the xy axis with the position C1 as the origin. Then, the target sound field point T3 to T6, placing the microphone _M 7 _{~M 10} for picking up the acoustic signal of the original sound field 10 in the target sound field point T3 to T6. The microphones M _{7 to} M ₁₀ are the same as the microphones M _{1 to} M ₆ .

Although not shown, the microphones M _{1 to} M ₆ are respectively connected to an MTR (Multi Track Recorder) that records acoustic signals m _{1 to} m ₆ that are environmental noises picked up by the microphones M _{1 to} M _6. ing. This MTR is assumed to be incorporated in the arithmetic unit 12 shown in FIG. The acoustic signals m _{1 to} m ₆ collected by the microphones M _{1 to} M ₆ are synchronously recorded by the MTR.

Although not shown, a microphone _M 7 _{~M 10} are respectively connected to the MTR for recording an audio signal _m 7 _{~m 10} is environmental noise picked up by the microphone _M 7 _{~M 10.} This MTR is assumed to be incorporated in the calculation unit 11 shown in FIG.

Calculation unit 11 calculates a covariance matrix R1 from the acoustic signal _m 7 _{~m 10} that was recorded in this MTR. Picked up by the target sound field point T3 to T6, the covariance matrix R1 indicating the covariance and the frequency characteristics of the acoustic signal _m 7 _{~m 10} recorded in the MTR, p = {7,8,9,10}, As q = {7, 8, 9, 10}, it can be expressed as the following equation (6). The covariance matrix R1 calculated by the calculation unit 11 is input to the calculation unit 12.

  Subsequently, in step ST12, the reproduction environment is set and crosstalk is measured.

Specifically, in the above-described anechoic chamber, which is the reproduction sound field 20, a space of 1.5 m from the center position C2 shown in FIG. 6 is provided, and the 6-channel reproduction speakers S are spaced 60 degrees apart from each other. _{1 to} S ₆ (hereinafter simply referred to as speakers S _{1 to} S ₆ ) are installed.

Further, as shown in FIG. 7, the reproduction sound field 20, in the area surrounded by the speaker S ₁ to S _6, to reflect the geometric positional relationship between the target sound field point T3 to T6, a center position Reproduction points P3 to P6 for reproducing the spatial acoustic features of the target sound field points T3 to T6 are set on the xy axis with C2 as the origin at intervals of 0.15 m. Then, microphones M _{11 to} M ₁₄ for collecting the acoustic signals of the reproduction sound field 20 at the reproduction points P 3 to P 6 are installed at the reproduction points P 3 to P 6. The acoustic signals collected by the microphones M _{11 to} M ₁₄ are input to the calculation unit 11.

The arithmetic unit 11 measures the response to the reproduction point P3~P6 output acoustic signals from the speakers _S 1 to S _6. Specifically, the transfer functions from the speakers S _{1 to} S ₆ shown in FIG. ₆ to the microphones M _{11 to} M ₁₄ installed at the reproduction points P 3 to P ₆ shown in FIG. 7 may be measured. However, when the reproduced sound field 20 is an anechoic chamber as in the present embodiment, only a simple time difference and distance attenuation are taken into consideration and output from each speaker S _{1 to} S ₆ by geometrical distance calculation. Responses from the reproduction points P3 to P6 of the acoustic signal thus obtained can be calculated.

The arithmetic unit 11 collects an acoustic signal obtained by collecting reproduced signals s _{1 to} s ₆ , which are acoustic signals output from the 6-channel speakers S _{1 to} S ₆ shown in FIG. 6, by the 4-channel microphones M _{11 to} M _14. Crosstalk (time difference T _{k, p} and distance attenuation g _{k, p} ) when re-collected signals r _{1 to} r ₄ are calculated in advance using equation (7) below.

  Next, in step ST13, the mixing ratio at the time of reproducing environmental noise is optimized. This processing operation will be described with reference to the flowchart shown in FIG.

In step ST21, the arithmetic unit 12, an acoustic signal _m 1 ~m ₆ which was recorded in MTR is picked up by the microphone _M 1 ~M ₆ as an input value, the reproduced signal _{s 1} output from the speaker _S 1 to S ₆ when the ~s ₆ and the output value, providing an appropriate initial value for the mixing coefficient a that constitutes the mixing matrix to determine the ratio of mixing satisfying the shown (8) relationship as equation below.

In step ST22, the arithmetic unit 12 calculates the mix ratio by mixing the matrix, it is outputted from the speaker _S 1 to S ₆ and generates a reproduced signal _s 1 ~s _6.

In step ST23, the arithmetic unit 12 calculates the crosstalk using equation (7) in step ST12 described above, picked up by the microphone _M 11 _{~M 14} of four channels disposed in the playback point P3~P6 acoustic The re-collected signals r _{1 to} r ₄ that are signals are synthesized.

In step ST24, the calculation unit 11 calculates the covariance matrix R2 of the acoustic signals collected by the four-channel microphones M _{11 to} M ₁₄ installed at the reproduction points P3 to P6. The covariance matrix R2 can be expressed as in equation (9). The covariance matrix R2 calculated by the calculation unit 11 is input to the calculation unit 12.

In step ST25, the operation unit 12 calculates an error D between the covariance matrix R1 and the covariance matrix R2. The calculation unit 12 calculates the error D by the following equation (10). The error D is always a positive value, and when the covariance matrix R1 and the covariance matrix R2 are equal, “error D = N ² ” is indicated as the minimum value. N represents the length of “row” and “column” of the covariance matrix R1 or R2.

  In step ST <b> 26, the calculation unit 12 determines whether there is an error D. If it is determined that there is almost no error D, the calculation unit 12 proceeds to step ST28. If it is determined that the error D still exists, the calculation unit 12 proceeds to step ST27.

  In step ST27, the arithmetic unit 12 updates the mixing matrix shown in the equation (8) so that the error D is reduced by the steepest descent algorithm described above. When the process of step ST27 ends, the process returns to step ST22.

  In step ST28, the operation unit 12 determines a mixing coefficient in response to determining that there is almost no error D between the covariance matrix R1 and the covariance matrix R2.

  Returning again to the flowchart shown in FIG. 4, the description will be continued.

In step ST14, sound field reproduction of the collected environmental noise and sound field evaluation are executed. The calculation unit 12 multiplies the 6-channel acoustic signals m _{1 to} m ₆ collected by the microphones M _{1 to} M ₆ and recorded by the MTR by the optimized mixing coefficient, thereby reproducing the 6-channel reproduction signal. subjected to mixing in s ₁ ~s ₆ is output from the speaker _S 1 to S _6.

Then, the reproduced signals s _{1 to} s ₆ , which are environmental noises output from the speakers S _{1 to} S ₆ , are collected again by the four-channel microphones M _{11 to} M ₁₄ installed at the reproduction points P 3 to P ₆ , and the original sound The sound field evaluation is executed by comparing the error D between the covariance matrix R1 of the field 10 and the covariance matrix R2 of the reproduced sound field 20.

  FIG. 9 shows an error evaluation result by computer simulation executed based on the above-described embodiment as an example. In FIG. 9, the “x” mark indicates the result of the reproduction without executing the mixing, and the “◯” mark indicates the result of the reproduction performed with the mixing coefficient obtained by the above-described method. Yes. As can be seen from FIG. 9, when the sound field reproduction system shown as the embodiment of the present invention is applied, the environmental noise at the target sound field point of the original sound field 10 is almost faithful to the reproduction point of the reproduction sound field 20. Can be reproduced.

  As described above, the sound signal output in the reproduction sound field 20 by the output control of the sound signal using the mixing coefficient obtained by the calculation unit 12 is the target sound field of the original sound field 10 at the reproduction point of the reproduction sound field 20. The spatial acoustic features of the points can be reproduced very faithfully.

  Therefore, for example, as shown in FIG. 1, the sound source cannot be specified, and the sound field reproduction using the transfer function cannot be performed, so that the sound field is surrounded by a dotted line including the target sound field points T1 and T2 of the original sound field 10 in the environment. The sound field in the region A can be reproduced faithfully to the sound field in the region B including the reproduction points P1 and P2 of the reproduction sound field 20.

  In addition, such a sound field reproduction system cannot be used to specify a sound source when applied to a gaming device (game device) that can virtually experience driving of a vehicle, a movie theater, or the like. The sound field in the region including the target sound field point of the original sound field 10 in an environment where the sound field cannot be reproduced using the sound field of the reproduction sound field 20 provided in a pseudo driver's seat of a game device, a movie theater seat, or the like. It is possible to faithfully reproduce the sound field in the area including the reproduction point.

  The above-described embodiment is an example of the present invention. For this reason, the present invention is not limited to the above-described embodiment, and various modifications can be made depending on the design and the like as long as the technical idea according to the present invention is not deviated from this embodiment. Of course, it is possible to change.

It is a figure for demonstrating the notional structure of the sound field reproduction system shown as embodiment of this invention. It is a flowchart for demonstrating the processing operation of the said sound field reproduction system. It is the figure which represented the curved surface of the error function typically. It is a flowchart for demonstrating the implementation method of the Example which actually implemented the said sound field reproduction system, and a series of processing procedures for evaluating an implementation result. It is a figure for demonstrating the sound environment of the original sound field in the Example to which the said sound field reproduction system is applied. It is a figure for demonstrating the reproduction environment of the reproduction | regeneration sound field in the Example to which the said sound field reproduction system is applied. It is a figure for demonstrating the sound collection environment of the reproduction | regeneration sound field in the Example to which the said sound field reproduction system is applied. It is a flowchart for demonstrating the processing method which optimizes the mixing ratio at the time of reproduction | regeneration of environmental noise. It is the figure which showed the error evaluation result by computer simulation.

Explanation of symbols

M1 microphone M2 microphone M3 microphone M4 microphone M5 microphone M6 microphone M7 microphone S1 speaker S2 speaker S3 speaker 10 original sound field 11 calculation unit 12 calculation unit 20 reproduction sound field

Claims (3)

A plurality of first sound collecting means which are respectively installed at a plurality of target sound field points in the original sound field and pick up an acoustic signal;
A second sound collecting means that is installed at an arbitrary position other than the target sound field point of the original sound field and picks up an acoustic signal;
First computing means for calculating a covariance matrix of acoustic signals picked up by the plurality of first sound collecting means;
In a different space from the original sound field, a plurality of third sound pickups that are installed at a plurality of reproduction points reflecting the geometric positional relationship between the plurality of target sound field points and collect an acoustic signal Means,
Second computing means for calculating a covariance matrix of the acoustic signals collected by the plurality of third sound collecting means;
Coefficient estimation means for estimating a coefficient such that the covariance matrix calculated by the second calculation means is substantially equal to the covariance matrix calculated by the first calculation means;
Output means installed in association with the second sound collecting means at an arbitrary position other than the plurality of reproduction points in a different space from the original sound field;
Output control means for controlling the coefficient estimated by the coefficient estimation means to be multiplied by the acoustic signal picked up by the second sound pickup means and output from the output means. Field reproduction system. The first calculation means calculates a covariance matrix for each predetermined frequency band of the acoustic signal collected by the plurality of first sound collection means,
The second calculation means calculates a covariance matrix for each of the predetermined frequency bands of the acoustic signal collected by the plurality of third sound collection means,
The sound field reproduction system according to claim 1, wherein the coefficient estimation unit estimates the coefficient for each of the predetermined frequency bands. Calculate the covariance matrix (hereinafter referred to as the first covariance matrix) of the acoustic signal collected at the target sound field point of the original sound field,
A covariance matrix (hereinafter referred to as the second covariance matrix) of the acoustic signal collected at the playback point that reflects the geometric positional relationship between the target sound field points in the playback sound field, which is a different space from the original sound field. Matrix)
Estimate a coefficient K such that the second covariance matrix is approximately equal to the first covariance matrix;
A sound field reproduction method characterized by outputting an acoustic signal obtained by multiplying an acoustic signal collected in an original sound field by an estimated coefficient K to a reproduction sound field.

Images