JPH09243448A - Method and apparatus for estimating acoustic transmission characteristics and acoustic device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To highly accurately make estimation even if a distance from a sound receiver is long. SOLUTION: Transmission characteristics of a sound source 13 and in a plurality of sound receivers 14-n are measured in an acoustic system 11 (32), a common pole is estimated from the transmission characteristics (71), residues are calculated (72), a residue function with a distance from a reference point as a variable is obtained from the residues (73), a transmission characteristic position 24 to be estimated is inputted, the residue value is obtained by the residue function according to its shift from the reference point (74), and transmission characteristics between the sound source 13 and the position 24 are obtained from the residue values and the pole (75).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a sound field reproducing device,
The present invention relates to an acoustic device such as a sound image simulator and a noise control device, a method of predicting an acoustic transfer characteristic applied to the acoustic signal processing method and the device thereof, and an acoustic device utilizing the prediction of the transfer characteristic.

[0002]

2. Description of the Related Art An acoustic transfer characteristic prediction method is a method of receiving a sound transfer characteristic from a sound source placed in a target acoustic system (for example, a room sound field) to an arbitrary sound receiving point at a sound receiving point position. It means a method of prediction without installing a device (for example, a microphone). Although the following description will be made assuming that the signal is a discrete signal, exactly the same argument holds when the signal is a continuous signal. For discrete signals, the time representation of the signal is
An integer parameter k representing time is expressed as x (k), and its frequency expression is expressed as X (z) using z transformation.

Sound transmission characteristics in a room Fig. 7A
This will be described with reference to FIG. A sound source (for example, a speaker) 13 and a sound receiver (for example, a microphone) 14 in the indoor space 11 are arranged. The signal X (z) is input to the input terminal 12 and the sound source 13
When the signal X (z) is output from the above, the signal X (z) reaches the sound receiver 14 under the influence of the acoustic transfer characteristic H (z).
The signal Y (z) received by the sound receiver 14 is output from the output terminal 15
Output from The acoustic transfer characteristic H (z) describes the input / output relationship of the output signal Y (z) at the output end 15 with respect to the input signal X (z) input at the input end 12, and H (z) = Y ( It is expressed as z) / X (z) (1). This H (z) has different characteristics even in the same indoor space 11 if the spatial arrangements of the sound source 13 and the sound receiver 14 are different. As can be seen from the above description, normally, in order to know the acoustic transfer characteristics at an arbitrary sound receiving point, it is necessary to install a sound receiver (for example, a microphone) at the sound receiving point and measure the input / output characteristics between them. .

(1) A conventional simple method for predicting the indoor transfer characteristic.
As a conventional simple method of predicting, there are two methods that sandwich the sound receiving point.
A sound receiver is installed at the point of 2 points from the sound source to both sound receivers.
Consider a method of measuring acoustic transfer characteristics and averaging both measured characteristics.
available. FIG. 7B shows a diagram for explaining this method. Figure
Elements common to 7A are assigned the same numbers. Transfer characteristics
The sound receivers 14-1 and 14-2 are provided on both sides of the position 24 to be predicted.
Is provided. In this method, first,
2 from each input / output characteristic between each of the sound generators 14-1 and 14-2
Two acoustic transfer characteristics H₁(z) and H_TwoMeasure (z). Sound transmission
As a method of measuring the characteristic, the signal input to the sound source 13 is X
(z), when the signal received by the sound receiver is Y (z),
For example, H (z) = (Y (z) X (z)) / (X (z) X (z)) can be obtained. Next, the two measured
Acoustic transfer characteristics H₁(z) and H _TwoAcoustic transfer characteristics based on (z)
Predicted value H '(z) is H' (z) = (H₁(z) + H_Two(z)) / 2 Calculate as (2).

However, according to this method, H '(z) becomes true as the frequency increases when the distance between the two sound receivers 14-1 and 14-2 is far from the sound receiving point 24 whose transfer characteristic is to be predicted. However, there is a problem in that it does not match the acoustic transfer characteristic H (z). (2) Example of conventional sound field reproducing device A sound field reproducing device is a device for reproducing a sound field in another indoor space in one indoor space. The principle will be described with reference to FIG. The same elements are designated by the same reference numerals. The sound field reproduction device 11 includes an inverse filter calculation circuit 33, an indoor transfer characteristic inverse filter convolution circuit 34, a simulated sound field convolution circuit 35, and an original sound reproduction circuit (for example, a CD player).
Consists of. First, the microphone 14 is installed near the ear of the listener 37 in the indoor space 11, and the indoor transfer characteristic H (z) from the speaker 13 to the microphone 14 is measured using the indoor transfer characteristic measuring circuit 32. Here, normally, the room transfer characteristics corresponding to both the right ear and the left ear are measured, but here, for simplification of description, only the right ear will be described, but the same applies to the left ear. The indoor transfer characteristic H (z) measured by the indoor transfer characteristic measuring circuit 32 is input to the inverse filter calculation circuit 33, and the inverse filter is calculated. The inverse filter is a filter having the inverse characteristic of the indoor transfer characteristic H (z), and in this case has the characteristic of 1 / H (z). Inverse filter characteristic 1 / H calculated by the inverse filter calculation circuit 33
(z) is sent to the inverse filter convolution circuit 34.

Next, the original sound X (z) output from the original sound reproduction circuit 36 is sent to the simulated sound field convolution circuit 35. The transfer characteristic G (z) of the sound field to be reproduced is inputted to the simulated sound field convolution circuit 35 in advance, and the convolution of G (z) and X (z) is executed in the simulated sound field convolution circuit 35. It The convolved signal G (z) X (z) is sent to the inverse filter convolution circuit 34, and the inverse filter convolution circuit 34 further convolves G (z) X (z) with 1 / H (z). . Eventually, the signal output from the inverse filter convolution circuit 34 becomes G (z) X (z) / H (z). This signal is input to the speaker 13. The signal G (z) X (z) / H (z) output from the speaker is further convolved in space with its true transfer characteristic H (z) in the indoor space 11 and reaches the microphone position 14. Therefore, the signal reaching the microphone position 14 is H (z) G (z) X (z) / H (z).
That is, at the position of the microphone 14, the signal is G (z) X
(z), and in the space with the indoor transfer characteristic H (z),
The listener 37 can obtain the impression as if he or she is in a room having the room transfer characteristic of G (z).

However, when the listener 37 moves to another listening position having the indoor transfer characteristic H ₁ (z), the signal at the listening position is H ₁ (z) G (z) X (z). / H (z) and G
(z) You cannot hear the X (z) signal. In such a case, it is usually necessary to move the microphone 14 and perform remeasurement. (3) Example of a conventional sound image simulator A sound image simulator is
This device realizes sound image localization at an arbitrary position. The principle will be described with reference to FIG. In FIG. 9, the same elements as those described above are designated by the same reference numerals. Now, the speaker 13 is the listener 3
It is installed at an angle of θ from the front direction of 7.

In FIG. 9A, when the signal X (z) is supplied from the original sound reproducing device 41 to the speaker 13, the acoustic transfer characteristics from the speaker to the ears are transmitted to the respective ears of the listener 37.
The signal arrives via HR (z, θ) and HL (z, θ). That is, the listener 37 receives the signal HR (z, θ) X in the right ear.
(Z), the signal HL (z, θ) X (z) is heard with the left ear. This acoustic transfer characteristic, HR (z, θ), HL
(Z, θ) is called a head-related transfer characteristic, and the difference between left and right hearing, that is, the difference between HR and HL is an important factor for humans to perceive the direction of the sound source. .

From the above, the conventional sound image simulator is configured as shown in FIG. 9B. Sound image simulator 5
Reference numeral 1 is a convolution circuit for a right ear head transfer characteristic simulating filter, 5
Reference numeral 2 is a convolution circuit for the left ear head transfer characteristic simulating filter.
3 comprises a head-related transfer characteristic storage circuit 54 for each direction. The head-related transfer characteristic storage circuit 54 for each direction stores the head-related transfer characteristics for each direction θ measured in advance. In the sound image simulator 51, the angle θ of the head-related transfer characteristic to be simulated is input from the desired direction input circuit 56. According to this input, the direction-dependent head-related transfer characteristic storage circuit 54 transfers the stored left and right head-related transfer characteristics H′R (z, θ) and H′L (z, θ) to the respective head-related transfer characteristics. It is given to the characteristic simulation filter convolution circuits 52 and 53. This filter convolution circuit 5
Numerals 2 and 53 are H'R (z, θ) and H'L (z, θ) convolved with the reproduction signal X (z) from the original sound reproduction device 41, respectively, and these convolution outputs are received by the headphone 5 of the listener 37.
Supply to the right and left input terminals of 5. At this time, the listener 37 receives the signal H′R (z, θ) X (z) in the right ear and the signal H′L in the left ear.
You will hear (z, θ) X (z). Assuming that the transfer characteristics are simulated with sufficient accuracy, H′R≈HR,
H′L≈HL, and this result is the same as the listening condition described with reference to FIG. 9A, and the listener 37 listening to the headphones 55 perceives that a sound source exists in the θ direction.
The above is the principle of the sound image simulator.

Naturally, the above-described head-related transfer characteristics greatly change depending on the direction θ of the sound source. Therefore, in order to localize the sound image in various directions, it is necessary to measure and accumulate the head-related transmission characteristics in many directions, and this kind of device is often used because there is a large amount of data to be accumulated. It is a problem in using it. Furthermore, it also has a problem that a sound image cannot be localized in a direction other than the accumulated direction.

(4) Example of Noise Control Device A noise control device is a device that eliminates the noise emitted from the noise source in the indoor space with a sound having the opposite phase of the noise. This principle will be described with reference to FIG. The same elements are designated by the same reference numerals. In this description, the speaker 1
Although only the noise control at the position of the microphone 14 will be described with reference to FIG. 3, the same argument holds when the number of control points is increased or the number of control speakers is increased.

First, the indoor transfer characteristic measuring circuit 32 transmits the indoor transfer characteristic H (z) between the speaker 14 and the microphone 13.
Is measured. The measured H (z) is passed to the inverse filter calculation circuit 33, and the inverse filter calculation circuit 33 calculates the inverse characteristic 1 / H (z) of the indoor transfer characteristic H (z), and the result is inverse filtered. It is passed to the convolution circuit 34. The noise transfer characteristic measuring circuit 61 detects the noise N (z) of the noise source 64 picked up by the noise source pickup microphone 62 and the noise N (z) to the microphone 14 through the indoor transfer characteristic F (z). The indoor transfer characteristic (noise transfer characteristic) F (z) through which the noise has passed is measured using the signal F (z) N (z) picked up by the sound. Measured F (z)
N (z) picked up is the noise transfer characteristic convolution device 63.
Passed to. Here, the noise transfer characteristic convolution device 63
Is the arrival time of the sound that reaches the sound receiver 14 from the noise source 64, and the control signal of the opposite phase in which the output of the inverse filter convolution circuit 34 reaches the sound receiver 14 through the indoor transfer characteristic H (z). To make the arrival time the same, the noise transfer characteristic F (z)
Then, the time enough to compensate for the time difference is removed, and this is designated as F '(z). Generally, to guarantee this time,
It is necessary to make the distance from the noise source 64 to the sound receiver 14 longer than the distance from the control speaker 13 to the sound receiver 14. Next, the noise transfer characteristic convolution device 63 convolves the time-compensated pseudo noise transfer characteristic F ′ (z) and the noise N (z), and passes it to the inverse filter convolution circuit 34. The inverse filter convolution circuit 34 receives the pseudo noise signal F ′.
(z) N (z) is convolved with the inverse filter 1 / H (z) to generate the signal -F '(z) N (z) / H (z), which is passed to the speaker 13. Signal output from speaker 13 -F '(z) N (z)
/ H (z) passes through the indoor transfer characteristic H (z) and passes through the sound receiver 14
Leads to. After all, the signal observed by the sound receiver 14 is F (z)
N (z) -F '(z) N (z) H (z) / H (z) and F' (z)
H (z) / H (z) is the time-compensated result F (z), the signal at the sound receiver 14 becomes 0, and the noise is eliminated at that position.

However, in this noise control device, when it is desired to move the noise control point, the sound receiver 14 must be moved to the noise control point, and the sound receiver cannot be installed at a location where it cannot be installed. Then, there is a problem that noise control is basically impossible.

[0014]

As described above, the conventional acoustic transfer characteristic predicting method has a problem that the predictive accuracy is deteriorated particularly on the high frequency side when the position where the transfer characteristic is desired to be predicted is far from the sound receiver. Also, in sound field reproduction, if the listener's position moves, it is necessary to re-measure the acoustic transfer characteristics between the speaker and the listener, and it is not necessary to re-measure such acoustic transfer characteristics. It is very convenient if you can predict. If the sound image simulator can also predict head-related transfer characteristics in various directions, it is not necessary to measure and store a large number of head-related transfer characteristics. Further, even in the noise control device, when the sound receiver cannot be installed at the noise control point, the noise control cannot be performed, but the control can be performed if the acoustic transfer characteristics can be predicted.

An object of the present invention is to provide a method and an apparatus for accurately predicting acoustic transfer characteristics.

[0016]

SUMMARY OF THE INVENTION According to the present invention, a physical pole of a target acoustic system is estimated from a plurality of acoustic transfer characteristics in the acoustic system, and a plurality of acoustic transfer characteristics for the estimated pole are estimated. By taking advantage of the fact that each residue is a function with the sound source / sound receiving point arrangement as a variable, the sound source / sound receiving point position of the place to be predicted is input to the residue change function. The acoustic transfer characteristic predicting method is characterized in that the acoustic transfer characteristic is predicted using the calculated residual and the estimated pole.

The present invention can predict the acoustic transfer characteristics at a position where a sound receiving point or a sound source cannot be arranged, and can be used for a sound field reproducing device, a sound image simulator, a noise control device and the like. The principle of the present invention will be described in detail below. According to the acoustic transfer characteristic prediction method of the present invention, the acoustic transfer characteristics in the same acoustic system can be expressed by the resonance frequency peculiar to the acoustic system and the residue that is a function of the sound source / sound receiving point arrangement. Is based. That is, acoustically, the acoustic transfer characteristic is represented by the following equation.

H (ω, R) = Σ _n (A _n (R) ω) / (ω ² −ω ² _n ) (3) where ω is the angular frequency and ω _n is the nth angular resonance frequency ( Complex number), A _n is a residue (complex number) for the n-th angular resonance frequency, and H (ω, R) is an acoustic transfer characteristic function that represents the response when the angular frequency ω is input. Further, R represents the arrangement of the sound source rs and the sound receiving point ro, and r = (x, y, z). Generally Tomesu A _n, the room size (length, width, height) by a unique function P (r) determined by reflectance between the _{wall, A n (R) = P} n (r s) P n (r _o ) It is expressed as (4). Also, each eigenfunction is approximately
It can be separated into eigenfunctions in the x, y, and z directions, and can be expressed as P _n (r) = P _nx (x) P _ny (y) P _nz (z) (5). Further, the separated eigenfunction is, for example, in the x-axis direction, P _nx (x) = C cos (k _x x + φ) (6) k _x = (n _x π / L _x ) + iγ _x (7) Is known to be. The same applies to the y and z directions,
It is the same at both the sound source and the sound receiving point. Here, L _x is the length of the room in the x-axis direction, n _x is a positive integer, γ is an amount determined by the reflectance of the wall, C is a constant, and φ is a phase.

In summary, the transfer characteristic H (ω, R) between the positions of arbitrary sound sources and sound receiving points is given by the resonance frequency ω _n ,
It is determined only by the residue which is a function of the sound source and the sound receiving point (position), and the residue A _n is a damping sine function with respect to the position of the sound source or the sound receiving point, and the length L in the axial direction of the wall It has an amount γ corresponding to the reflection coefficient, a phase φ, and a gain constant C as constants. That is, in order to predict the acoustic transfer characteristics at an arbitrary sound source / sound receiving point arrangement, the resonance frequency ω _n , the axial length, the amount γ corresponding to the reflection coefficient of the wall, the phase φ, and the gain constant C are calculated. It can be seen that the residue A _n (R) that has as a parameter may be determined by some method.

Now, in order to determine the above parameters, the above-mentioned acoustic transfer characteristics are rewritten as a discrete system model, which is as follows. H (z, R) = ΣA _n (R) / (1-p _n z ⁻¹ ) (8) Σ is from n = 0 to p, where H (z, R) is the acoustic transfer characteristic in a discrete system. Where z is the z-transform and the frequency ω and z = exp (−j
ω) are tied together. Here, _pn is a pole corresponding to the resonance frequency, and the residue A _n (R) is the same as that in the equation (3). By the way, the above formula and the previously proposed common pole / zero model Japanese Patent Publication No. 7-39968 (Japanese Patent Laid-Open No. 4-295)
728) publication,

[0021]

[Equation 1]

Comparing with, it can be seen that p _n is the same. Here, p _n is a common pole that does not depend on the sound source / sound receiving point arrangement, q _n is a zero point that depends on the sound source / sound receiving point arrangement, and a _n
Is an AR (auto regressive) coefficient corresponding to the common pole, b _n
Is the MA (moving average) coefficient corresponding to the zero point, P, Q
Are the orders of the poles and zeros, respectively, and C ₂ is a constant. Also,
In this specification, the common pole p _n having the same equation (8) and (9)
Therefore, the equation (8) which is the basis of the present invention is called a common pole / residue model.

According to the common pole / zero model, the common pole p _n corresponding to the resonance frequency ω _n can be obtained from a plurality of transfer characteristics. That is, the resonance frequency ω _n is immediately determined by the common pole / zero modeling method. Next, a method of determining the parameters of the function A _n (R) will be described. A _n (R) is A _n (R) = Px (x _s ) Px (x _o ) Py (y _s ) Py ( _yo ) Pz (z _s ) Pz (z _o ) ( 10) which is the residue of each axis A _ni (R _i ), A _ni (R _i ) = P i (i _s ) P i (i _o ) (i = x, y, z) (11 ), It is clear that it is sufficient to calculate A _n for each axis. Now, since the residue of each axis is the same as the residue of the one-dimensional sound field, the one-dimensional sound field will be described here.

Now, the parameters of the residue A _nx (Rx) = Px (x _s ) Px (x _o ) (12) in the one-dimensional sound field are estimated. The case where is fixed will be described. The same argument holds when the position of the sound receiving point is fixed.
Now, when fixing the sound source position, Px (x _s) is constant. Now, since the A _nx (Rx) does not depend only on the receiving point position x _o, the receiving point position x _o x, k a k _n, A _nx the (Rx) A
_When rewritten as _n (x), A _n (x) = D cos (k _n x + φ) (13). Here, D is a gain constant and φ is a phase. Further, k _n is a complex number including the amount γ corresponding to the reflectance of the wall, but here, it is assumed that γ is small and is ignored. That is, k _{n is set} to k _n = nπ / Lx (14). Now, the above-mentioned k _n and the actual frequency of the angular resonance frequency ω _{n have} a relationship of k _n = ω _n × f _s / c (15). Where f _s is the sampling frequency and c
Is about 340 m / s in sound velocity. Therefore, the value of k _n is
It is obtained from the real frequency part of the frequency ω _n of the common pole p _n obtained by the common pole / zero model.

Now, the remaining parameters D of the function A _n (x)
And φ are A _n (x) at the actual sound receiving point position
Estimate using the value of. A at position x with sound receiver
_n (x) is the acoustic transfer characteristic H (z,
x) can be obtained as A _n (x) = (1-p _n z ^-1 ) H (z, x) | _z = p _n (16). Since the remaining parameters of the function A _n (x) are D and φ, it is considered sufficient if there are at least two simultaneous equations, that is, if A _n (x) at two measurement points is known. If the spatial period k _n of A _n (x) and the measurement interval match, the equation becomes the same and it becomes impossible to solve. Therefore, when measuring,
It is necessary to make sure that the number is one or more and that the positions are not evenly spaced.

Now, A_n(x) is the position of three sound receiving points
x₁, X_Two, X_ThreeIf we could measure about_n(X₁) = D cos (k_nx₁+ Φ) A_n(X_Two) = D cos (k_nx_Two+ Φ) (17) A_n(X_Three) = D cos (k_nx_Three+ Φ) simultaneous equations are obtained. Now k_nx₁+ Φ is k_nx
Replaced with x ₁, X_TwoThe interval of d₁, X₁When
x_ThreeThe interval of d_TwoThen, the above simultaneous equations are_n(X₁) = D cos (k_nx) A_n(X_Two) = D cos (k_nx) cos (k_nd₁) −D sin (k_nx) sin (k_nd₁) (18) A_n(X_Three) = D cos (k_nx) cos (k_nd_Two) −D sin (k_nx) sin (k_nd_Two) Can be rewritten as When this is expressed in matrix,Becomes Well, k_n, D₁, D_Two, A_n(X₁), A_n
(X_Two), A_n(X_Three) Is known, this matrix
When the equation is expressed as Ax = b (20), x = (A^TA)^-1A^TThe least squares solution can be obtained by b (21) and D cos (k_nx)
And D sin (k_nThe value of x) is obtained. Using these values
Then, the reference point x₁By giving the distance d from
You can know the value of the residue at the desired position. That is, A_n
(X₁+ D) is A_n(X₁+ D) = D cos (k_nx) cos (k_nd) −D sin (k_nx) sin (k_nd) It can be calculated as (22). This method is an absolute position
Relative distance only to a certain reference point without knowing
If you know the distance, you can determine the residue at that position.
You can see that they also have points.

If A _n (x) obtained in the above description and the common pole are used, the transfer characteristic at an arbitrary position x is H (z, x) = Σ (A _n (x) / ( 1-p _n z ^-1 )) (23) n can be predicted as 0 to p. In addition, since A _n and p _n are usually complex numbers and have conjugation, they can be expressed in the following form. H (z) = Σ ((b _0n −b _1n z ⁻¹ ) / (1−a _1n z ⁻¹ + a _2n z ⁻² )) (24) Σ is from n = 0 to p / 2 It can be directly realized as a parallel type digital filter.

Further, the transfer characteristics of the obtained common pole / residue model are expanded to show the following commonly used ARMA (Auto regressive-moving a).
verage) model can be converted. H (z) = Σb _n z ^-1 / (1 + Σa _n z ^-1 ) (25) The denominator Σ is n = 1 to p, and the numerator Σ is n = 0 to p.
This can be realized as an IIR (Infinite Impulse response) filter. When developed in the form of the ARMA model, the impulse response h (n) of the transfer characteristic can be obtained according to the following equation.

[0029] _{^{h (k) = Σ p n}} = 1 a n h (k-n) + Σb n δ (k-n) (26) where, [delta] (k) is a [delta] function, when k = 0 It is a function that becomes 1 only and becomes 0 at other k. The method of predicting the transfer characteristic according to the present invention has been summarized above and will be described with reference to FIG. The previously mentioned elements are numbered identically. The indoor transfer characteristic prediction device 78 includes a common pole estimation circuit 71, a residue calculation circuit 72, a residue function determination circuit 73, a residue numerical value calculation circuit 74 at a predicted position, and an indoor transfer characteristic calculation circuit 75.

First, in the room transfer characteristic measuring circuit 32, n
The room transfer characteristics of each of the sound receivers 14-n are measured. Next, the common pole estimation circuit 71 estimates the common pole peculiar to the room from the plurality of measured transfer characteristics. The residue value calculation circuit 72 calculates the residue value for the known point of the sound receiver position according to the equation (16) using the measured indoor transfer characteristic and the estimated common pole. next,
The residue function determination circuit 73 determines the function of the residue according to the procedure from Expression (17) to Expression (22). Next, the position of the sound receiving point 24 to be predicted is input from the predicted position input circuit 77. The input predicted position is passed to the predicted position residue numerical value calculation circuit 74 via the position input terminal 76. The predicted position residue numerical value calculation circuit 74 inputs the predicted position to the residue function determined by the residue function determination circuit 73, and calculates the residue value at the sound receiving point position 24 to be predicted. The indoor transfer characteristic calculation circuit 75 calculates the indoor transfer characteristic according to the equation (23) from the residual value calculated by the predicted position residual value calculation circuit 74 and the common pole estimated by the common pole estimation circuit 71. Since the functions of the common pole and the residue are unique to the room acoustic system, these are obtained in advance, and these are stored in a memory or the like, and according to the predicted position to be input, this and the acoustic transfer characteristics in the sound source. May be requested. In that case, the sound receiver 14-n, the room transfer characteristic measuring circuit 32, the common pole estimating circuit 71, the residue numerical value calculating circuit 72, and the residue function determining circuit 73 may be omitted from FIG.

Hereinafter, application examples of the present invention will be described based on three embodiments.

[0032]

BEST MODE FOR CARRYING OUT THE INVENTION Prediction Example of Acoustic Transfer Characteristics An example of predicting the acoustic transfer characteristics when the walls in the y and z axis directions are completely absorbing in the three-dimensional sound field shown in FIG. 2 will be described. In FIG. 2, the same elements as those described above are given the same numbers.
The reflectance of the wall 81 is 0.8, the interval between the walls 81 in the x-axis direction is 6.8 m, and the sampling frequency is 1 kHz.

First, 5 measurable positions, 3.4 m,
The sound receivers 14-1 to 14-5 are arranged at positions of 3.5 m, 3.65 m, 3.7 m, and 4.2 m, and the indoor transfer characteristic measuring circuit 32 measures a plurality of indoor transfer characteristics. Next, in the indoor transfer characteristic prediction device 78, the residues at the common pole and at the 3.65 m, 3.7 m, and 4.2 m positions were obtained by the method described above, and the parameters of the residue function were determined. Next, the position of 3.95 m is input to the position input device 77, passed to the indoor transfer characteristic prediction device 78, and the residue value at the 3.95 m position is obtained, and combined with the common pole, the common pole / residue model The transfer characteristics were obtained in the form. Next, the obtained form of the indoor transfer characteristic was converted into the ARMA (recursive moving average type) model of the equation (25), and then the impulse response was obtained from the equation (26).
The result is shown in FIG. 3A together with the actually measured impulse response at the 3.95 m position. It can be seen that the waveforms are almost indistinguishable. Next, the frequency-amplitude characteristic of this transfer characteristic is shown in FIG. 3B. For reference, FIG. 3B also shows a result (conventional method) obtained by simply averaging the transfer characteristics at the 3.7 m and 4.2 m positions. From the figure, it can be seen that the high-frequency components are greatly different in the case of the conventional method in which simple averaging is performed. However, it can be seen that the use of the present invention predicts an indoor transfer characteristic that is substantially coincident with the actually measured impulse response. Sound Field Reproducing Device An example of applying the present invention to a sound field reproducing device will be described with reference to FIG. The previously mentioned elements are numbered identically. The listening position input circuit 111 has the same function as the position input circuit 77. However, since the listener 37 is targeted here, the listening position input circuit 111 is used as the listening position input circuit. Listen position input circuit 111
Has a function of detecting a listening position by using, for example, a TV monitor.

First, the indoor transfer characteristic measuring circuit 32 measures a plurality of indoor transfer characteristics corresponding to a plurality of sound receiver positions 14-n installed at positions where the listener 37 is not disturbed. Next, the acoustic transfer characteristic prediction circuit 78 of the present invention estimates the residue function and the common pole by using the plurality of transfer characteristics measured by the indoor transfer characteristic measuring circuit 32. Next, the acoustic transfer characteristic prediction circuit 78, based on the listening position input from the listening position input circuit 111, the room transfer characteristic H at the listening position.
Predict (z). In the inverse filter calculation circuit 33, the inverse filter 1 / H (z) of the predicted transfer characteristic is calculated and passed to the inverse filter convolution device 34. In the inverse filter convolution circuit 34, the original sound X reproduced by the original sound reproduction circuit 36 is reproduced.
(z) is a signal G (z) X (z) convolved with the simulated sound field G (z) in the simulated sound field convolution circuit 35, and the inverse filter calculation circuit 3
Convolution with the inverse filter 1 / H (z) calculated in 3,
The signal G (z) X (z) / H (z) is generated and input to the speaker 13. Signal G (z) X (z) / input to the speaker 13
Since H (z) reaches the listener via the room transfer characteristic H (z), the listener eventually finds that H (z) G (z) X (z) / H (z) =
You will hear G (z) X (z). That is, the listener 37 can hear the signal G (z) X (z) that is not affected by the indoor transfer characteristic H (z) of the room 11 in which the listener 37 is present. When the listener moves, the listening position input circuit 111 sets the listening position again, and the acoustic transfer characteristic prediction circuit 78
By predicting the transfer characteristic of the moved destination and calculating the inverse filter again by the inverse filter calculation circuit 33, it is possible to make the listener hear the signal G (z) X (z). As described above, according to the sound field reproducing apparatus using the present invention, it is possible to reproduce the sound field only by using the indoor transfer characteristic measured by the indoor transfer characteristic measuring circuit 31 at the beginning, and Even if the position moves, it is not necessary to measure the indoor transfer characteristic again. Sound Image Simulator FIG. 5 shows a sound image simulator device to which the present invention is applied. In order to operate this sound image simulator, first, head-related transfer characteristics in several directions are measured, and the common pole / residue modeling device 1
It is necessary to perform common pole / residue modeling at 27. In the common pole / residue modeling device 127, the common pole calculation circuit 71 uses the direction-dependent head-related transfer characteristics of the direction-dependent head-related transfer characteristic measured value storage circuit 54 in which some measured values are stored. To calculate. Next, the residue numerical value of the head-related transfer characteristic for each direction stored in the residue numerical value calculation circuit 72 is calculated. The residue function determination circuit 73 determines the residue function based on the result of the residue value calculation circuit.

Next, in the sound image simulator device 51, the common pole determined by the common pole / residue modeling device 127 is stored in the common pole storage circuit 126, and the residue function is stored in the residue function storage circuit 125. . In the sound image simulator apparatus 51, first, the listener 37 inputs a desired input circuit 56 in a direction in which the sound is desired to be localized. The predicted position residue numerical value calculation circuit 74 calculates the residue value in the direction input by the desired direction input circuit 56 for the left ear and the right ear. That is, the positional relationship similar to the positional relationship between the sound source and the left ear and the right ear corresponding to the head-related transfer characteristics in each direction in the holding circuit 54 is obtained for the input direction, and the deviation from the reference point is used. Right ear,
Calculate the residue value for the left ear. Alternatively, the residual value may be obtained by regarding the deviation of the input direction with respect to the reference direction as the deviation from the reference point. The calculated residue value is converted into a MA (moving avalanche) coefficient corresponding to the numerator in the last expression of the expression (25) in the residue value MA conversion circuit 124, and the MA coefficient convolution circuit 12 for the right ear is respectively converted.
2, passed to the left ear MA coefficient convolution circuit 123. The common pole storage circuit 126 passes the AR (auto-regressive) coefficient corresponding to the common pole to the common pole convolution circuit 121.
The signal reproduced by the original sound reproduction device 41 passes through the common pole convolution circuit 121 and then the MA coefficient convolution circuit 1 for the right ear.
22, passed to the left ear MA coefficient convolution circuit 123.
The right-ear MA coefficient convolution circuit 122 and the left-ear MA coefficient convolution circuit 123 convolve the signals passed through the common-pole convolution circuit 121 with the respective MA coefficients to generate a right-ear signal and a left-ear signal. , To the headphones 55. When the listener 37 wants to change the direction, he / she inputs the direction again in the desired direction input circuit 56, and the residue numerical value calculation circuit 74 recalculates the residue value and realizes localization in the desired direction. To do.

At this time, if it is attempted to simulate the head-related transfer characteristics in an arbitrary direction, the conventional sound image simulator shown in FIG. 9B needs to accumulate the head-related transfer characteristics in an enormous direction. However, when the present invention shown in FIG. 5 is applied, the residue depending on the direction is described by a function obtained from some small head-related transfer characteristics, and thus the head-related transfer characteristics to be accumulated are described. This means that the amount of data has been greatly reduced. Also, since the common pole is a direction-independent quantity,
The amount of this data is also very small.

By applying the present invention to the sound image simulator as described above, not only the problem of the accumulated data amount of the acoustic transfer characteristics, which is the problem of the conventional apparatus of this kind, can be solved, but also the direction can be continuously expressed. By handling possible physical quantities inside the simulator, a more natural sound image localization feeling can be given. Noise Control Device An example in which the present invention is applied to a noise control device will be described with reference to FIG. The same elements are designated by the same reference numerals. The noise transfer characteristic prediction circuit 131 has the same function as the indoor transfer characteristic prediction circuit 78, but here, different names are used for distinction. In this description, since the function of the noise transfer characteristic convolution circuit shown in FIG. 7 is the same as that of the conventional technology, the compensation of the arrival time difference will not be described in detail.

First, the position 133 at which noise is desired to be controlled is input by the control point position input circuit 132. The desired position of the input control point is passed to the indoor transfer characteristic prediction circuit 78 and the noise transfer characteristic prediction circuit 131. Noise transfer characteristic measurement circuit 6
1 measures a plurality of noise transfer characteristics corresponding to a plurality of sound receiver positions 14-n. Next, the plurality of measured noise transfer characteristics are passed to the noise transfer characteristic prediction circuit 131. In the noise transfer characteristic prediction circuit 131, the indoor transfer characteristic prediction circuit 78
The noise transfer characteristic F (z) at the position input from the control point position input circuit 132 is predicted by the same procedure as. The predicted noise transfer characteristic F (z) is the noise transfer characteristic convolution circuit 63.
And is convolved with the noise N (z) picked up by the noise pickup device 62 to generate a signal F (z) N (z), which is passed to the inverse filter convolution circuit 34.

Next, the indoor transfer characteristic measuring circuit 32 measures a plurality of indoor transfer characteristics corresponding to the plurality of sound receiver positions 14-n, and transfers this to the indoor transfer characteristic predicting circuit 78. In the indoor transfer characteristic prediction circuit 78, the control point position input circuit 132
The indoor transfer characteristic H (z) at the position input from is predicted. The predicted indoor transfer characteristic H (z) is passed to the inverse filter calculation circuit 33, and the inverse filter 1 / H (z) is calculated,
It is passed to the inverse filter convolution circuit 34.

In the inverse filter convolution circuit 34, the inverse filter 1 / H (z) calculated in the inverse filter calculation circuit 33 and the signal F (z) generated in the noise transfer characteristic convolution circuit 63.
The signal is convolved with N (z) to generate a signal -F (z) N (z) / H (z), which is input to the speaker 13. The signal F (z) N (z) / H (z) reproduced by the speaker 13 is the indoor transfer characteristic H (z).
Arrive at control point 133 through
Signals from noise sources F (z) N (z) and -H (z) F (z) N (z)
Since the sum of / H (z) is observed, the observation signal at the control point position becomes 0, and the noise control at the control point 133 is realized. If you want to move the control point,
Input the position to be controlled by the control point position input circuit and recalculate both the noise transfer characteristic and the room transfer characteristic.

As described above, in the noise control device according to the present invention, it is not necessary to measure the transfer characteristic at the control point position. Therefore, noise control at a place where a sound receiver cannot be installed at the control point is also possible. It will be possible. Further, even when the control point is desired to be moved, it is not necessary to re-measure the transfer characteristic. In the embodiments shown in FIGS. 4, 5, and 6, respectively, poles and residue functions are obtained in advance and held in holding means such as a memory, and the corresponding acoustic transmission is performed according to the input position or the input direction information. You may make it calculate a function. In this case, the sound receiver 14-n, the acoustic characteristic measuring circuits 32 and 61, the common pole estimating circuit 71, the residue numerical value calculating circuit 72, and the residue function determining circuit 73 can be omitted.

[0042]

According to the present invention, a common pole common to the indoor transfer characteristics of a target acoustic system and a residue function which is a function of a sound source / sound receiving point position are calculated from a plurality of acoustic transfer characteristics in the acoustic system. Acoustic transfer characterized in that it is possible to predict the indoor transfer characteristics at any position simply by estimating and expressing the indoor transfer characteristics with this common pole and residue function and inputting the positions of the sound source and the sound receiving point. This is a characteristic prediction method.

If the present invention is used to predict the transfer characteristic of a place where the sound receiver cannot be installed, it is possible to know the acoustic transfer characteristic at the position where the sound receiver is not present. Further, when the present invention is applied to the sound field reproducing apparatus, the inverse filter can be calculated without measuring the indoor transfer characteristic at the listener's position, and it is possible to follow the listener's movement. .

If the present invention is applied to a device such as a sound image simulator that stores and simulates a plurality of known acoustic transfer characteristics, the number of parameters required for storage can be reduced, and as a result, storage can be performed. It is possible to reduce the amount of data to be processed. Further, in the present invention, since the residue function is a continuous function with respect to the arrival direction of sound, it has a unique feature that the arrival direction of sound can be continuously changed.

If this invention is applied to noise control,
It is possible to control the noise at the position where the sound receiver cannot be installed.
Furthermore, the acoustic transfer characteristics simulated (represented) by the common pole / residue model, which is the basis of the present invention, can be applied to various acoustic signal processing devices used in the process of calculation,
It is possible to improve the calculation amount and the data storage amount.

[Brief description of drawings]

FIG. 1 is a diagram illustrating an indoor transfer characteristic prediction device of the present invention.

FIG. 2 is a diagram showing a concrete indoor transfer characteristic prediction using the indoor transfer characteristic prediction device of the present invention.

FIG. 3A is a diagram comparing the original impulse response and the waveform of the impulse response predicted using the indoor transfer characteristic predicting apparatus of the present invention, and B is the original indoor transfer characteristic and the indoor transfer characteristic predicting apparatus of the present invention. It is the figure which compared the waveform of the frequency amplitude characteristic of three persons of the indoor transfer characteristic predicted using it and the indoor transfer characteristic predicted using the conventional prediction method.

FIG. 4 is a diagram showing a configuration example in which the present invention is applied to a sound field reproducing device.

FIG. 5 is a diagram showing a configuration example in which the present invention is applied to a sound image simulator.

FIG. 6 is a diagram showing a configuration example in which the present invention is applied to a noise control device.

7A is a diagram illustrating an acoustic transfer characteristic H (z), and FIG. 7B is a diagram illustrating a conventional indoor transfer characteristic predicting apparatus.

FIG. 8 is a diagram illustrating a conventional sound field reproducing device.

9A is a diagram for explaining head-related transfer characteristics, and FIG. 9B is a diagram for explaining a conventional sound image simulator device.

FIG. 10 is a diagram illustrating a conventional noise control device.

Continued Front Page (72) Inventor Junji Kojima 3-19-2 Nishishinjuku, Shinjuku-ku, Tokyo Inside Nippon Telegraph and Telephone Corporation

Claims (6)

[Claims] 1. A sound transfer characteristic prediction method for predicting a sound transfer characteristic between a sound source placed in an acoustic system and a sound receiving point, comprising: a sound transfer characteristic between a plurality of sound sources and a sound receiver in the acoustic system. Is measured, the poles of the acoustic system are estimated from the measured plurality of acoustic transfer characteristics, and a residue for the estimated poles is obtained for the measured plurality of acoustic transfer characteristics, and the obtained plurality is determined. The residue function is determined from the residue of, the residue function is input to the position of the sound source and the sound receiving point to determine the residue at these sound source / sound receiving point positions, and the determined residue and the The sound source from the estimated poles
A method for predicting acoustic transfer characteristics, comprising predicting acoustic transfer characteristics between sound receiving point positions. 2. The residue function is assumed to be a damping sine function whose variables are the positions of a sound source and a sound receiver whose acoustic transfer characteristics are measured, and has parameters such as a period, an amplitude, a phase, and a damping constant, The period of the damped sine function is determined from the frequency of the corresponding estimated pole, and the amplitude and phase of the damped sine function and the damping constant are the residuals determined for each of the plurality of acoustic transfer characteristics. The acoustic transfer characteristic prediction method according to claim 1, wherein the method is determined as an amount that most matches the value of the number. 3. An acoustic transfer characteristic predicting device for predicting an acoustic transfer characteristic between a sound source placed in an acoustic system and a sound receiving point, comprising: pole holding means for holding a pole of the acoustic system; Residue function holding means for holding a residue function, residue calculation means for obtaining a residue according to the sound source and sound receiving point position of the input acoustic transfer characteristic to be predicted, by means of the residue function, An acoustic transfer characteristic prediction device comprising: a transfer characteristic calculation means for calculating the acoustic transfer characteristic from the obtained residue and the pole. 4. An inverse filter calculation means calculates an inverse filter characteristic of an acoustic transfer characteristic between a listening position in an acoustic system and a sound source, and the inverse filter characteristic is convolved with an acoustic signal by the convolution means, and the convolution is performed. In the acoustic device that reproduces the sound of the acoustic signal from which the influence of the acoustic transfer characteristics is removed by emitting the acoustic signal from the sound source to the listening position, a pole holding unit that holds the pole of the acoustic system. A residue function holding means for holding a residue function having a distance from a listening position in the acoustic system as a variable; a means for obtaining a residue corresponding to an input listening position by the residue function; And a means for calculating a sound transfer characteristic between the sound source and the input listening position from the pole and supplying the calculated sound transfer characteristic to the inverse filter means. 5. The convolution means is used to convolve the respective acoustic transfer characteristics from the sound source direction to the left and right ears to the original sound,
In an acoustic device that presents the convoluted original sound to the left and right ears of the listener with headphones etc. to achieve sound localization, the pole holding means for holding the pole and the distance or direction with respect to the reference position are used as variables. A residue function holding means for holding a residue function, and a residue calculation means for obtaining the left and right ear residues according to the input desired direction by the residue function, the convolution means for the original sound A pole characteristic filter for convolving the auto-regressive coefficients corresponding to the poles, and a left and right ear residue filter for convolving the output of the pole characteristic filter with the moving average coefficients respectively corresponding to the left and right ear residues. And an audio device. 6. A noise transfer characteristic between a noise source and a control point in an acoustic system is convoluted with noise of the noise source by a noise transfer characteristic convolution means, and a control sound between a control sound source in the acoustic system and the control point. By obtaining the inverse filter characteristic of the transfer characteristic by the inverse filter calculating means, convolving the inverse filter characteristic with the convolved noise by the inverse filter convolution means, and emitting the convolved noise from the control sound source. In the acoustic device that gives a noise having a phase opposite to that of the noise at the control point, a pole holding unit that holds a pole in the acoustic system, and a noise residue function having a distance to a reference position in the acoustic system as a variable A noise residue function holding means for holding, a control sound residue function holding means for holding a control sound residue function having a variable with respect to a reference position in the acoustic system, and an input control point position The above A means for obtaining a noise residue by the noise residue function according to a position with respect to a sound source; and a noise transfer characteristic between the noise source and the input control point calculated from the noise residue and the pole Noise transfer characteristic calculation means to be supplied to transfer characteristic convolution means, means for obtaining a control sound residue number by the control sound residue function according to the position of the input control point position with respect to the control sound source, and the control An acoustic device, comprising means for calculating a control sound transfer characteristic between the control sound source and the input control point from a sound number and the pole and supplying the control sound transfer characteristic to the inverse filter calculating means. .