JP2870440B2 - Three-dimensional sound field reproduction method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a three-dimensional sound field reproducing system for reproducing a three-dimensional sound field using a two-channel speaker.

[0002]

2. Description of the Related Art Conventionally, this type of three-dimensional sound field reproduction system employs a signal transmitted from a right speaker to a listener's left ear and a signal transmitted from a left speaker to a listener's right ear (hereinafter referred to as a crosstalk path signal). ) Is canceled, and only the signal transmitted from the right speaker to the listener's right ear and the signal transmitted from the left speaker to the listener's left ear (hereinafter referred to as a main path signal) are transmitted to the listener. This is to reproduce a sound field having a sense of reality compared to a sound field reproduced by the audio signal of FIG.

FIG. 3 is a diagram for explaining a conventional three-dimensional sound field reproduction system.

The filter configuration used in the conventional three-dimensional sound field reproduction system has the following configuration composed of four FIR filters (digital filters whose impulse response is represented by a finite number of series). FIR filters _XLL 101 and _XLR 102 each receiving an audio digital signal L for the left speaker 109 and a right speaker 110
For receiving audio digital signal R for input
R filters X _RL 103 and X _RR 104,
FIR filter X _LL 101 and FIR filter X _RL 1
03 are output from the D / A converter 105 and the amplifier 10
7, and is input to the left speaker 109 via the FIR filter X _LR 102 and the FIR filter X
Each output of the _RR 104 is input to the right speaker 110 via the D / A converter 106 and the amplifier 108 in order.

In this three-dimensional sound field reproduction system, when the audio signals L and R are respectively input, each FIR filter X
_{The LL} 101, X _LR 102, X _RL 103, and X _RR 104 cancel the crosstalk path signal when the signals are output from the left and right speakers 109 and 110, and transmit only the main path signal to the listener. Each FI at this time
R filters _XLL 101, _XLR 102, _XRL 103, _XRR
The filter coefficient of 104 is determined as follows.

Each FIR filter X_LL101, X_LR10
2, X_RL103, X_RRThe filter coefficients of 104
Left and right ears of the listener from speakers 109 and 110, respectively
Can be expressed using a transfer function transmitted to Sand
From the right speaker 109 and the left speaker 110
The transfer function transmitted to the left ear is H_RL, H_LLAnd then on the right
From the speaker 109 and the left speaker 110 to the right ear
The transfer functions _RR, H_LRThen, each FIR
Filter X_LL101, X_LR102, X_RL103, X _RR1
The filter coefficient of 04 is represented by the following equation.

[0007] X_LL= H_RR÷ (H_LL× H_RR-H_LR× H_RL) (1) X_LR= -H_LR÷ (H_LL× H_RR-H_LR× H_RL) (2) X_RR= H_LL÷ (H_LL× H_RR-H_LR× H_RL) (3) X_RL= -H_RL÷ (H_LL× H_RR-H_LR× H_RL(4) Each filter coefficient X_LL, X_LR, X_RL, X_RRDetermine each biography
Function H_LL, H_RR, H _LR, H_RLIs desired in an anechoic room
A signal (reference signal) that has a strong correlation with the wave is generated in the system,
Input from the dummy head microphone installed at the specified listening position
The difference between the input signal (signal corresponding to sound pressure) and the reference signal is 2
It is obtained by minimizing the root mean square error. This nothing
Based on each transfer function obtained in the sound room, etc., FFT (Fast
 Performs convolution operation processing by FourieTransfor)
Thus, each FIR filter in the above equations (1) to (4) is
Ruta X_LL101, X_LR102, X_RL103, X_RR104
Are determined.

[0008]

However, the above-mentioned conventional three-dimensional sound field reproduction system has the following problems.

(1) When calculating the transfer function, a delay error occurs in the measurement of sound pressure using a dummy head microphone due to insufficient convergence of the filter. Therefore, the transfer function obtained by the conventional example includes an error corresponding to the delay error with respect to the ideal transfer function. In the conventional method in which the filter coefficient of the FIR filter is determined using the transfer function including the delay error as described above, the crosstalk path signal cannot be canceled accurately, and the main path signal cannot be accurately determined by the listener. There is a problem that can not be communicated.

(2) In the conventional system, the filter coefficient of each FIR filter is determined based on a transfer function obtained by measurement in an anechoic room or the like. However, since the room acoustic characteristics vary depending on the installation environment, if the filter coefficient is determined based on the transfer function obtained by measurement in an anechoic room, etc. However, there is a problem that the crosstalk path signal cannot be sufficiently canceled, and a desired realism cannot be obtained.

It is an object of the present invention to provide a three-dimensional sound field reproduction method capable of obtaining a filter coefficient of an FIR filter based on a transfer function taking a delay error into consideration.

It is still another object of the present invention to provide a three-dimensional sound field reproduction system capable of accurately transmitting a main path signal to a listener by setting the above-mentioned filter characteristics to be optimal for room acoustic characteristics which vary depending on an installation environment.

[0013]

In order to achieve the above object, a three-dimensional sound field reproducing method according to the present invention comprises a first and a second audio signal for generating a three-dimensional sound field, and the first audio signal. First and second FIR filters as inputs, and third and second FIR filters as inputs of the second audio signal .
And a fourth FIR filter and the first and third FIR filters.
A first speaker operated by an output of the IR filter;
A second speaker that is operated by outputs of the second and fourth FIR filters, and the first to fourth speakers are configured to reproduce a three-dimensional sound field by each output of the first and second speakers. A three-dimensional sound field reproduction system for setting a filter coefficient of an FIR filter, wherein a dummy head microphone installed at a predetermined listening position and a reference signal having a strong correlation with a desired wave are generated , and based on the reference signal, 1st and 1st
2 speakers are operated respectively, out of the dummy head microphone measures the output of the first and second speaker
Transfer function measuring means for determining a transfer function relating to a main path signal and a transfer function relating to a crosstalk path signal such that an error between a force and the reference signal is minimized; and Using the transfer function relating to the transfer function and the crosstalk path signal and the inverse function of the transfer function relating to the main path signal as parameters, the filter coefficients of the first to fourth FIR filters are set using convolution operation processing by FFT. And a filter coefficient calculating means.

In the above case, the dummy head microphone is
First to measure the sound pressure applied to the left and right ears of the listener
And a second dummy head microphone,
A number measuring means for generating a reference signal;
And a reference signal generated by the reference signal generating means.
First and second adaptive filters as inputs;
The output of the first dummy head microphone is connected to one input
And outputs the output of the first adaptive filter to the other input.
First adding means for outputting an error between these inputs as a force
And the output of the second dummy head microphone to one input
And the output of the second adaptive filter is
Second adding means for inputting and outputting an error between these inputs
And the square mean of the error output from the first adding means
And calculating said error so that said root mean square error is minimized.
Update the filter coefficient of the first adaptive filter
1 from the least squares processing means and the second addition means.
The root-mean-square error of the input error is obtained, and the root-mean-square error is
The filter of the second adaptive filter is minimized so as to be minimum.
A second least squares processing means for updating the filter coefficient,
Wherein said filter coefficient calculating means comprises said first and second
2 based on the filter coefficient of the adaptive filter
Set the filter coefficients of the first to fourth FIR filters
May be configured.

[0015]

In the conventional three-dimensional sound field reproduction method, since the delay transfer function relating to the main path signal is not used as a parameter when performing the convolution operation processing by FFT, the filter coefficient of the FIR filter calculated by this is It had a delay error.

As a result of studying the above-mentioned delay error, it has been found that the delay error can be corrected by using the reciprocal of the transfer function relating to the main path signal, as will be described in detail later.

In the three-dimensional sound field reproduction method of the present invention, in addition to the transfer function relating to the main path signal and the transfer function relating to the crosstalk path signal, the transfer function relating to the main path signal is used as a parameter when performing the convolution operation processing by FFT. The inverse function is used. By adding the inverse function of the transfer function relating to the main path signal to the parameter, the delay error in obtaining the filter coefficient is corrected,
A filter coefficient in which a delay error has been corrected is set as a filter coefficient of the FIR filter. In the method of the present invention in which the filter coefficient in which the delay error is corrected is set as the filter coefficient of the FIR filter, the crosstalk signal can be canceled more accurately and the path signal can be more accurately compared to the conventional method. Can be communicated to the listener.

Further, in the present invention, the transfer function relating to the main path signal and the transfer function relating to the crosstalk path signal determined by the transfer function measuring means are determined so as to be optimal for room acoustic characteristics which vary depending on the installation environment. Therefore, the filter coefficient of the FIR filter can be set to a value according to the installation environment. Therefore, the filter characteristics can be optimized for the room acoustic characteristics that change depending on the installation environment, and the main path signal can be accurately transmitted to the listener.

[0019]

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a diagram for explaining a three-dimensional sound field reproduction system according to one embodiment of the present invention.

The three-dimensional sound field reproduction method of the present embodiment is a two-channel speaker three-dimensional sound field reproduction method, which is capable of correcting a delay error generated when measuring a transfer function. Further, even if the room acoustic characteristic changes due to the installation environment, the optimum filter coefficient of the FIR filter can be determined according to the change.
Hereinafter, the system configuration of this system will be described.

In FIG. 1, _HLL is a transfer function transmitted from the left speaker 10 to the left ear, _HRL is a transfer function transmitted from the right speaker 9 to the left ear, _HRR is a transfer function transmitted from the right speaker 9 to the right ear, H _LR indicates a transfer function transmitted from the left speaker 10 to the right ear.

The filter configuration is the same as that of the conventional system.
FIR filters X _LL 1,1 each having an input transfer function (audio digital signal) L for the left speaker 9 as an input.
X _LR 2, and FIR filters X _RL 3 and X _RR 4 each having an input transfer function (audio digital signal) R for the right speaker 10 as inputs. The FIR filters X _LL 1 and X _RL 3 Each output is D / A
The input is input to the left speaker 9 via the converter 5 and the amplifier 7 in order, and the FIR filter X _LR 2 and the FIR filter
Each output of the filter X _RR 4 is input to the right speaker 10 via the D / A converter 6 and the amplifier 8 in this order.

As means for determining the filter coefficients of the FIR filters X _LL 1, X _LR 2, X _RL 3, and X _RR 4 in the above-mentioned filter configuration section, a signal detection section 12, a transfer function measurement section 13 and a filter calculation section A part 14 is provided. The signal detection unit 12 measures sound pressures P _L and P _R applied to the left and right ears of the listener at a predetermined listening position. The transfer function measuring unit 13 calculates a transfer function based on the sound pressure detected by the signal detecting unit 12. The filter calculation unit 14 determines each of the FIR filters X _LL 1, X _LR 2, X _RL 3, based on the transfer function obtained by the transfer function measurement unit 13.
This is for calculating the filter coefficient of X _RR 4.

In the three-dimensional sound field reproducing system of the present embodiment having the above system configuration, the FIR filters X _LL 1 and X _LR are used by the signal detector 12, transfer function measuring unit 13 and filter calculator 14 at the installation location. 2, X _RL 3, X _RR
Four filter coefficients are determined.

Hereinafter, each FIR filter X _LL 1, X _LR 2,
A method for _obtaining the filter coefficients of X _RL 3 and X _RR 4 will be described in detail.

P _L and P _R applied to the left and right ears of the listener are the coefficients by which the filter coefficients X _LL , X _LR , X _RL and X _RR can cancel the crosstalk path signal and transmit the main path signal. Then, it can be expressed by the following equation using the input transfer functions L and R, and the transfer functions H _LL , H _RL , H _RR and H _LR .

[0028] _{P L = (X LL × H} LL + X LR × H RL) L + (X RR × H RL + X RL × H LL) R (5) P R = (X LL × H LR + X LR × H RR ) L + ( _XRR × _HRR + _XRL × _HLR ) R (6) Here, the coefficient terms of the above equations (5) and (6) are changed so that only the main path signal is transmitted to both ears. Put on.

_XLL × _HLL + _XLR × _HRL = 1 (7) _XLL × _HLR + _XLR × _HRR = 0 (8) _XRR × _HRL + _XRL × _HLL = 0 (9) _XRR × H _RR + X _RL × H _LR = 1 (10) The left side “1” of the equations (7) and (10) is an ideal transfer function of the main path signal transmitted to the listener, and the left side of the equations (8) and (9). "0" is an ideal transfer function of the crosstalk path signal.
Since X _LL and X _LR and X _RL and X _RR are obtained in the same manner, calculation of X _LL and X _LR will be described below.

When the above equations (7) and (8) are represented by a matrix,

[0031]

(Equation 1) Becomes The above equation (11) is put as the following equation.

HX = 1 (12) H ⁻¹ HX = H ⁻¹ (13) is obtained by multiplying both sides of the above equation (12) by the inverse matrix H ⁻¹ of the matrix H. From this equation (13), it can be seen that the delay parameter H ⁻¹ H has occurred. Therefore, X _LL and X _LR can be determined by the following equations (X _RL and X _RR can be similarly determined).

X _LL = (1 ÷ det H) × (H _RR ÷ (H _LL × H _RR )) (14) X _LR = (1 ÷ det H) × (−H _LR ÷ (H _LL × H _RR )) (15) det H = ( _HLL × _HRR ) ÷ ( _HLL × _HRR ) − ( _HLR × _HRL ) ÷ ( _HLL × _HRR ) (16) The above equations (14) to (16) are converted into the above equations ( 7)
A transfer function of the main path signal from the speaker to the listener is obtained.

Det H ÷ det H = 1 (17) Here, since the inverse function of each transfer function is a least-squares solution, the above equation (7) can be expressed as follows from the above equation (17).

X _LL × H _LL + X _LR × H _RL = det H ÷ det H (18) X _LL and X _LR can be obtained from the above equation (18) and the above equation (8) (X _RL and X _RR can be similarly obtained).

Each filter coefficient X _LL , X _LR , X _RL , X
_RR is a transfer function _{H LL, H RL, H RR} , H LR transfer function H _LL, inverse function of H _RR (H _{LL) -1,} by performing arithmetic processing convolution by FFT to (H _{RR) -1} Since it is obtained, it is expressed by the following equation.

X_LL= X × (X^Two)_-1× (H_LL× H_RR)_-1× H_RR (19) X_LR= X × (X^Two)_-1× (H_LL× H_RR)_-1× (−H_LR) (20) X_RR= X × (X^Two)_-1× (H_LL× H_RR)_-1× H_LL (21) X_RL= X × (X^Two)_-1× (H_LL× H_RR)_-1× (−H_RL(22) X = H_LL× H_RR× (H_LL× H_RR)_-1 -H_LR× H_RL× (H_LL× H_RR)_-1 (23) (X^Two)_-1= ｛[H_LL× H_RR× (H_LL× H_RR)_-1 -H_LR× H_RL× (H_LL× H_RR)_-1]^Two｝_-1 (24) Equations (19) to (24) obtained as described above are used.
And each filter coefficient X _LL, X_LR, X_RL, X_RRSought
For example, the crosstalk path signal is attenuated compared to the conventional method.
And transmit the main path signal accurately.
You.

The inverse function (X ² ) ₋₁ , (H _LL × H _RR )
_-1 can be obtained by the inverse filter theorem using a Levinson algorithm or the like as follows. When the relationship between the ideal impulse response Y, the transfer function H, and the inverse function Z to be obtained is expressed by a matrix,

[0039]

(Equation 2) become that way. When the transposed matrix is applied to both sides of the above equation (25) and the least square solution of the inverse function Z is obtained, H ^t Y = H ^t HZ (26) When the cross-correlation vector P = H ^t Y and the auto-correlation matrix R = H ^t H can be expressed as P = RZ, and the inverse function Z can be obtained by the following equation.

Z = R ⁻¹ P (27) The least square solution of this inverse function Z is the inverse function (X ² ) ₋₁ , (H _LL
× H _RR ) ₋₁ , and the transfer function H corresponds to X ² , H _LL × H _RR .

Next, the signal detecting section 12 and the transfer function measuring section 13 are means for determining each of the filter coefficients X _LL , X _LR , X _RL , and X _RR using the above equations (19) to (24). The filter calculator 14 will be described in detail.

FIG. 2 is a diagram showing the internal configuration of the signal detection unit 12, transfer function measurement unit 13, and filter calculation unit 14 shown in FIG.

The signal detecting section 12 is provided at a predetermined listening position.
Pressure P on the left and right ears of the listener _L, P_RMeasure
Dummy head microphone 15_L, 15_RAnd these dummy
Head microphone 15_L, 15_RAnalog detected by
A / D converters 16 and 17 for A / D conversion of signals
It is composed of

The transfer function measuring unit 13 is an M sequence (short for Maximum Period sequence) for generating a reference signal in the system.
A noise unit 22, adaptive filters 18, 20;
It is composed of least square method processing units 19 and 21 and adders 23 and 24. Adaptive filters 18, 20
Output the output signal o using the reference signal from the M-sequence noise unit 22 as an input. The adders 23, 24
Each of the outputs of the / D converters 16 and 17 is set as one input as an ideal output d.
20 each output signal o is used as the other input to output an error e between the ideal output d and the output signal o . Least square processor 19 and 21, respectively updates the filter coefficient of the adaptive filter 18 using the least squares method to the mean square error of the error e output from the adder 23 becomes minimum, respectively I do.

The filter calculating section 14 is adapted to update the adaptive filter 1 updated by the least squares processing sections 19 and 21.
Each transfer function (H _LL , H _RL ,
H _RR , H _LR ), convolution operation processing by FFT, and FI by an algorithm for obtaining an inverse filter.
Calculation of filter coefficients to be set for the R filters _XLL1 , _XLR2 , _XRL3 , and _XRR4 is performed using the above-described equations (19) to (24).

It should be noted that the input of each speaker 9, 10 in the figure is transmitted via a D / A converter 5 and an amplifier 7.
_{Alternatively} , when the transfer functions H _LL and H _LR are obtained via the D / A converter 6 and the amplifier 8,
When the transfer functions H _RR and H _RL are obtained, the right speaker 10 is used.
It is configured to be input to. Also, the signal detection unit 1
2. Dummy head microphones 1 corresponding to both ears of listener
By providing 5 _L and 15 _R, and providing the transfer function measuring unit 13 with the adaptive filters 18 and 20, the least square method processing units 19 and 21, and the adders 23 and 24, the transfer functions H _RR and H
_RL or the transfer functions H _LL and H _LR can be obtained at the same time.

Hereinafter, the signal detecting unit 12 and the transfer function measuring unit 1
3 and the specific operation of the filter calculation unit 14
I will tell. Here, the transfer function H_RRAnd H_RL, And transmission
Number H _LLAnd H_LRIs similarly calculated, so here
Is the transfer function H_LLAnd H_LRExplain when to ask for
You.

To determine the transfer functions _HLL and _HLR , a reference signal is input from the M-sequence noise section 22 to the left speaker 9 via the D / A converter 5 and the amplifier 7, and the adaptive filters 18, 20 Is input to each of.

When the reference signal is input to the right speaker 9, a signal is output from the left speaker 9 and detected by the dummy head microphones 15 _L and 15 _{R of the} signal detection unit 12. That is, the dummy head microphone 15
_L signal transmitted to the left ear of the listener from the left speaker 9 is detected by the signal transmitted to the right ear of the listener from left speaker 9 by the dummy head microphone 15 _R is detected.

When the signals from the right speaker 9 are detected by the dummy head microphones 15 _L and 15 _R , respectively, the output analog signals of the dummy head microphones 15 _L and 15 _R become A
A / D conversion is performed by the A / D converters 16 and 17 and the adders 23 and 24
Are input respectively.

When the ideal output d from each of the A / D converters 16 and 17 is input to the adders 23 and 24, the adder 23 outputs the ideal output d output from the A / D converter 16 and the M-sequence noise. An output signal o output from the adaptive filter 18 with the reference signal from the
Is obtained as the error e, which is input to the least square method processing unit 19. Similarly, in the adder 24,
The difference between the ideal output d from the A / D converter 16 and the output signal o from the adaptive filter 18 is obtained, and this difference is input to the least squares processing unit 21 as an error e.

As described above, when the error e is input from each of the adders 23 and 24 to the least squares processing units 19 and 21,
The least-squares method processing unit 19 updates the filter coefficient of the adaptive filter 18 using the least-squares method so that the mean square error of the input error e is minimized. Similarly, the least-squares method processing unit 21 updates the filter coefficient of the adaptive filter 20 using the least-squares method so that the input error e is minimized. Thus, the transfer function H _LR is obtained from the updated filter coefficient of the adaptive filter 18, and the transfer function H _LL is obtained from the updated filter coefficient of the adaptive filter 20.

The above processing is performed by using the transfer function H_RRAnd H_RLTo
The transfer function H_LL, H_LR, H_RR, H_RLIs all
When calculated, the filter calculation unit 14 calculates the calculated
Transfer function H_LL, H_LR, H_RR, H_RLBased on FFT
Convolution operation processing and an algorithm for finding the inverse filter
FIR filter X by algorithm_LL1, X_LR2, X _RL
3, X_RRThe above equation is used to calculate the filter coefficient to be set to 4.
This is performed using (19) to (24). Each FIR filter X
_LL1, X_LR2, X_RL3, X_RRCalculate the filter coefficient of 4
Then, the filter calculation unit 14 calculates the calculated filter
Each FIR filter X shown in FIG._LL1, X
_LR2, X_RL3, X_RRTransfer to 4 and set.

As described above, each FIR filter X
By the filter coefficients of the _{LL 1, X LR 2, X} RL 3, X RR 4 is set to attenuate the crosstalk path signal as compared with the conventional method, it is possible to accurately transmit the main path signal .

In the above-described embodiment, the signal detection unit 1
2. Dummy head microphones 1 corresponding to both ears of listener
By providing 5 _L and 15 _R, and providing the transfer function measuring unit 13 with the adaptive filters 18 and 20, the least square method processing units 19 and 21, and the adders 23 and 24, the transfer functions H _RR and H
_{Although the configuration in which the RL} or the transfer functions _HLL and _HLR are obtained at the same time has been described, the number of the dummy head microphone, the adaptive filter, the least-squares method processing unit, and the adder may be one. In this case, each transfer function is obtained separately.

[0056]

Since the present invention is constructed as described above, it has the following effects.

In the three-dimensional sound field reproduction system of the present invention, the filter coefficient of the set FIR filter is obtained by correcting the delay error, so that the crosstalk signal can be more accurately generated than in the conventional system. It is possible to cancel, and there is an effect that a sense of reality is improved.

Furthermore, since the filter characteristics can be optimized for the room acoustic characteristics that vary depending on the installation environment, even when the room acoustic characteristics are significantly different,
There is an effect that a desired realism can be obtained.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining a three-dimensional sound field reproduction method according to one embodiment of the present invention.

FIG. 2 is a diagram showing an internal configuration of a signal detection unit 12, a transfer function measurement unit 13, and a filter calculation unit 14 shown in FIG.

FIG. 3 is a diagram for explaining a conventional three-dimensional sound field reproduction method.

[Explanation of symbols]

1-4 FIR filter 5, 6 D / A converter 7, 8 Amplifier 9, 10 Speaker 12 Signal detection unit 13 Transfer function measurement unit 14 Filter calculation unit 15 _L, 15 _R Dummy head microphone 16, 17 A / D converter 18, Reference Signs List 20 adaptive filter 19, 21 least squares processing unit 22 M-sequence noise unit

──────────────────────────────────────────────────の Continued on front page (51) Int.Cl. ⁶ Identification code FI H04R 5/033 G10K 15/00 L

Claims (2)

(57) [Claims] Input [1 claim: a first and second audio signals for generating a three-dimensional sound field, the first and second FIR filters to the first audio signal input, said second audio signal Third and fourth FIR filters, a first speaker operated by the output of the first and third FIR filters, and a second and fourth FIR filter.
And a second speaker that is operated by the output of the FIR filter of the first through fourth FIR filters so that a three-dimensional sound field is reproduced by each output of the first and second speakers.
A three-dimensional sound field reproduction system for setting a filter coefficient of the IR filter, a dummy head microphone is installed at a predetermined listening position to generate a strong reference signal correlated with the desired wave, based on the reference signal
Operating the first and second speakers respectively.
So, to measure the output of the first and second speaker said
The error between the output of the dummy head microphone and the reference signal is minimized.
Transfer function measuring means for obtaining a transfer function relating to the main path signal and a transfer function relating to the crosstalk path signal , such as to be small; and a transfer relating to the transfer function relating to the main path signal and the cross talk path signal determined by the transfer function measuring means. Filter coefficient calculating means for setting filter coefficients of the first to fourth FIR filters using convolution operation processing by FFT, using a function and an inverse function of a transfer function relating to the main path signal as parameters. Characteristic three-dimensional sound field reproduction method. 2. The three-dimensional sound field reproduction system according to claim 1,
And wherein the dummy head microphones measure sound pressure applied to left and right ears of a listener, respectively.
And a second dummy head microphone, wherein the transfer function measuring means inputs a reference signal generating means for generating the reference signal, and a reference signal generated from the reference signal generating means.
A first and a second adaptive filter, and an output of the first dummy head microphone to one input.
And the other input the output of said first Adapute Ibufiruta
First adding means for outputting an error between these inputs as a force
And the output of the second dummy head microphone as one input
And outputs the output of the second adaptive filter to the other input.
Second adding means for outputting an error between these inputs as a force
And the mean square error of the error output from the first adding means.
And calculate the first value so that the root-mean-square error is minimized.
A first method for updating the filter coefficient of the adaptive filter
Mean square error of the error output from the least square method processing means and the second addition means
And calculate the second value so that the root mean square error is minimized.
Second to update the filter coefficients of the adaptive filter
A least-squares method processing means, wherein the filter coefficient calculating means comprises:
Based on the filter coefficient of the adaptive filter, the first
To setting the filter coefficients of the fourth to fourth FIR filters.
A three-dimensional sound field reproduction method characterized by the following.