JP2001285998A - Out-of-head sound image localization device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a device generally usable by unspecified many listeners without the need of complicated constitution. SOLUTION: An earphone 3 is mounted on the ear of the listener and a microphone 4 gathers sound waves generated from the earphone 3 inside an external auditory canal. An adaptive filter 2 for obtaining the inverse transmission function of an external auditory canal transmission function which is a transmission function from the earphone 3 to the output of the microphone 4 is provided and sound source signals are supplied through a digital filter 1 to the adaptive filter 2. A band filter 5 is constituted so as to turn the product of the external auditory canal transmission function and the inverse transmission function of the adaptive filter 2 to be filter characteristics and the output of the digital filter 1 is inputted. The band filter 5 and the output of the microphone 4 are subtracted in a subtractor 6 and a convergence operation circuit 8 converges the inverse transmission function of the adaptive filter 2 on the basis of the result.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an out-of-head sound image localization apparatus for perceiving a position (sound image) perceived as a sound source outside the head using an audio device in contact with both ears such as stereo headphones and binaural earphones. Things.

[0002]

2. Description of the Related Art Conventionally, various techniques have been considered as an out-of-head sound image localization apparatus. As such a technique, for example, there has been the following technique.

FIG. 2 is a diagram for explaining the principle of realizing out-of-head sound image localization. In FIG. 2, (a) is listening by a speaker,
(B) shows the listening with the binaural earphone. In the figure, 100 is a sound source signal, 101 is a speaker, 102 is a listener, 103 is a digital filter, 104 is an earphone, and 105 is a microphone installed in the ear canal of the listener. The subscripts _L and _R such as SSTF _L and SSTF R
Indicates the left and right sides.

The principle of localization of an out-of-head sound image will be described below with reference to FIG. The sound source in the space (the speaker 10 in FIG.
The same stimulus as the sound stimulus from 1) to the eardrum is applied to the earphone 10
By giving to the eardrum from 4, the sound source can be perceived as being outside. However, it is extremely dangerous to capture a vibration stimulus signal on the eardrum due to a sound wave from a living body as an electric signal. Therefore, the sound source signal 100 in FIG.
The transfer function of the electrical signal from the ear to the eardrum cannot be measured exactly.

Therefore, a very small microphone 105 is attached to the external auditory canal of both ears, and a transfer function from the sound source signal 100 radiated from the speaker 101 to the output of the microphone 105, that is,
Spatial acoustic transfer function (SSTF: Spacia)
l Measure Sound Transfer Function).

Here, literature: Jens Brauelt,
Masayuki Morimoto, Toshiyuki Goto “Spatial Sound” Kashima Publishing p. 20 ~
p. As described in the sound wave propagation in the ear canal as described in Section 2.4 of Section 28, the sound wave in the ear canal is represented by a one-dimensional model. It is self-evident that the electric signal due to the vibration of FIG. Therefore, even if the ultra-small microphone 105 is fixed to the ear canal and the transfer function, that is, the impulse response in the time domain is measured, the result is equivalent.

However, as pointed out in the above document, since the speaker 101 has frequency characteristics, the true transfer function of the electric signal from the input of the speaker 101 to the output of the microphone 105 is If the transfer function is LSTF (Loud Speaker Transfer Function), it is SSTF / LSTF. Just this value is the head related transfer function HRTF (Head Related Transfer Fu
nction). LSTF is the speaker 1
This is a speaker transfer function for correcting the frequency characteristic of No. 01.

HRTF is the ratio of the transfer function when the head is present to the transfer function when the head is not present, as defined in the above document. Since this is the product of the function and the LSTF, and the spatial propagation is only the amount of delay, HRTF ≒ SSTF / LSTF.

On the other hand, in FIG. 2B, in order to create a transfer function equivalent to the binaural earphone (or stereo headphone) 104 using the binaural earphone 104, a microphone attached to the external auditory canal from the input of the binaural earphone 104 The transfer function up to the output of 105, that is, the ear canal transfer function (ECTF: Ear Cana)
l Transfer Function), and if the transfer function of the product of the ECTF and the transfer function of the digital filter 103 matches the transfer function SSTF / LSTF, the same listening signal as the speaker listening is placed at the location of the microphone 105 installed in the ear canal. Can be played.

As described above, the transfer function of the digital filter 103 can be obtained by a computer. That is, this transfer function (out-of-head sound image localization transfer function) is represented by S
LTF (Soundimage Localization Transfer Functio
n)

[0011] SLTF = ｛SSTF / (ECTF · LS
TF)｝ ≒ HRTF / ECTF, and all the terms on the right side of this equation can be obtained by measurement. After that, the mathematical operation is performed to obtain the SLTF.

[0012]

However, even if the above-mentioned SLTF can be obtained, the transfer function EC
The TF and the HRTF differ depending on the size of the ear canal, the size of the ear, and the size of the face of the listener. That is,
If the transfer function does not match the shape of the person's face, an accurate sound image cannot be localized outside the head, and in the worst case, it may not be possible to determine the front and back of the sound image, or it may not be localized outside the head.

[0013] For this reason, if the head is not located outside the head due to individual differences in the transfer function, it cannot be used universally for an unspecified number of listeners, so that it is difficult to select several kinds of transfer functions measured in advance. There is a need for a solution or a system that can measure the transfer function for each individual. But,
The above-mentioned hardware-based solution has a problem that the amount of hardware is increased, and the method using a system for measuring a transfer function for each individual has a problem that such a measurement system is very expensive.

[0014]

The present invention employs the following structure to solve the above-mentioned problems. <Configuration 1> A speaker that is provided in contact with both ears and generates a sound wave based on a sound source signal, a microphone that collects sound waves generated from the speaker in the ear canal, and a (sound transfer function / sound source located in space) Transfer function), and a function of inputting a sound source signal through the digital filter and outputting the signal to a speaker, and an inverse impulse of an ear canal transfer function which is a transfer function from the speaker to the output of the microphone. An adaptive filter for obtaining a response value, a memory for storing initial values for obtaining an inverse impulse response value in the adaptive filter, an impulse response value of an ear canal transfer function, and an inverse impulse response of an ear canal transfer function obtained by the adaptive filter Has a filter characteristic such that the product of the value and the value becomes a band-pass filter characteristic. A bandpass filter to be applied, a subtractor for subtracting the output of the bandpass filter and the output of the microphone, and an approximate digital filter for inputting a sound source signal through a digital filter and storing a value of a predetermined approximate transfer function. A convergence calculation circuit that updates the inverse impulse response value of the adaptive filter based on the result of the subtraction using the value of the approximate transfer function of the approximate digital filter until the inverse impulse response value of the adaptive filter converges. Out-of-head sound image localization device.

<Structure 2> In the out-of-head sound image localization apparatus according to Structure 1, the memory stores an inverse impulse response value corresponding to the center of gravity of the Euclidean distance of the inverse impulse response of the ear canal transfer function of a plurality of listeners. An out-of-head sound image localization apparatus, wherein an approximate digital filter stores impulse response values corresponding to the center of gravity of the Euclidean distance of impulse responses of a plurality of listeners' ear canal transfer functions.

<Structure 3> In the out-of-head sound image localization apparatus according to Structure 1, the memory stores the inverse impulse response of the listener closest to the center of gravity of the Euclidean distance of the inverse impulse response of the ear canal transfer function of the plurality of listeners. An external head sound localization apparatus characterized in that a value is stored, and an approximate digital filter stores impulse response values of listeners closest to a center of gravity of a Euclidean distance of impulse responses of external auditory canal transfer functions of a plurality of listeners. .

<Structure 4> In the out-of-head sound image localization apparatus according to any one of Structures 1 to 3, the initial value of the inverse impulse response value of the ear canal transfer function used in the previous adaptive operation is calculated by the inverse of the adaptive filter previously converged. An out-of-head sound image localization apparatus, comprising: an initial value calculation / setting update circuit for storing a value updated using an impulse response value as a new initial value in a memory.

[0018]

Embodiments of the present invention will be described below in detail.

<< Principle of the Present Invention >> First, the principle of the present invention will be described. Transfer functions ECTF and SS in sound localization outside the head
There are the following documents regarding the study on the personality of TF. S.Yano, H.Hokari, andS.Simada: "A Study on
Personal Difference in the Transfer Functions of S
sound Localization ", AES the 106th Convention, NO.492
2May8-11 (1999)

According to the above document, it has been concluded that the transfer function ECTF has greater individuality than the transfer function SSTF.

Therefore, by introducing an adaptive signal processing technique, an ECT that matches one's own while listening to musical sounds and voices without measuring the transfer function of the ECTF for each individual.
Consider a method for determining the transfer function of F.

FIG. 3 is a functional block diagram for explaining the principle of the present invention. The illustrated system shows a block for only one ear for realizing the function of FIG. 2B, and includes a digital filter 1, an adaptive filter 2, an earphone (speaker) 3, a microphone 4, a bandpass filter 5, and a subtractor. 6
Consists of

FIG. 4 is a functional block diagram showing a configuration for realizing the system of FIG. That is, in FIG. 4, the inverse ear canal transfer function of the adaptive filter 2 (1 / ECT
To obtain F), an approximate digital filter 7 and a convergence operation circuit 8 are provided, and the adaptive filter 2, the approximate digital filter 7 and the convergence operation circuit 8 constitute an external auditory canal transfer function adaptive operation circuit 9.

In these figures, since the transfer functions LSTF and SSTF are less affected by individual differences, numerical values for which SSTF / LSTF has been calculated in advance are input to the digital filter 1. Let this output signal be x (n).

Next, the operation after the signal x (n) is a basic configuration known as a Filtered-X LMS algorithm. However, what differs from the active noise control (ANC) is that the bandpass filter 5 having a fixed transfer function is an unknown transfer function and is known as a primary path. Since the adaptive filter 2 and the unknown transfer function ECTF are connected in series, the time domain of the transfer function when the input signal is x (n) and the output signal is z (n) is If the waveform is an impulse waveform with a delay, the transfer function of adaptive filter 2 is EC
It becomes the inverse transfer function of TF.

However, since the ECTF includes the earphone 3 and the microphone 4, attenuation occurs outside the band. Therefore, if the convolution operation result of the impulse responses of the adaptive filter 2 and the ECTF is the impulse response of the bandpass filter 5, the tap coefficient value or the impulse response value of the adaptive filter 2 can be obtained stably. For this reason, S.Yano, H.Hokari, S.Shimada, and
H. Irisawa: "A Study on the Derivation of Transfer Fu
nctions for Sound Image Localization UsingStereo E
arphones ", JAES, Vol. 47, No. 6, p. 469-p. 478, (1999)
It is revealed in.

That is, if the band of the bandpass filter 5 is made to pass through a band narrower than the band of the adaptive filter 2, the out-of-band portion of the transfer function from the adaptive filter 2 is canceled by the subtractor 6, and A stable solution can be obtained.

The function of the external auditory canal transfer function adaptive operation circuit 9 composed of the adaptive filter 2, the approximation digital filter 7 and the convergence operation circuit 8 is based on the output e (n) of the subtractor 6 and the function of the external auditory canal transfer function. Inverse transfer function ECTF
^This is a function that ^{calculates -1} .

Here, the basic principle of the adaptive filter 2 and the block orthogonal projection algorithm (hereinafter referred to as BOP algorithm) used here will be described with reference to FIG.

It is generally known that the BOP algorithm converges at a higher speed than the LMS algorithm. The details of this algorithm are described in Shigeo Tsujii: "Adaptive Signal Processing", Shokodo, p. 70 to p. 97.

FIG. 5 is an explanatory diagram for explaining the operation of the principle of the present invention. Output x of digital filter 1
(N) is first input to an adaptive filter 2 for adaptively identifying the inverse characteristic of ECTF. This 1 / ECTF
In order to identify the characteristics of the FIR filter, an FIR filter configuration is often used. The FIR filter characteristic, by giving tap coefficient h _M from the outside, it is possible to obtain easily the desired characteristics in a stable operation. Assuming that this output is y (n), y (n) is represented by equation (1) in FIG. here,
T denotes the transpose of the matrix,

X _M (n) = [x (n), x (n−1),
.., X (n−M + 1)] ^T , h (n) = [h ₀ (n),
h ₁ (n),..., h _M−1 (n)] ^T.

Here, x (n) is a vector x
Elements of _M (n), _{h i} is the element of the vector h (n), i and the tap position of the FIR filter structure, M is the number of taps of the adaptive filter 2, n represents the time.

The output y (n) of the adaptive filter 2 is input to one earphone 3. The signal z (n) output from the microphone 4 is obtained by setting the impulse response of the transfer function ECTF to an unknown transfer function and setting the impulse response to c _N.
z (n) is represented by equation (2) in FIG. Where y _N =
[Y (n), y (n-1), ..., y (n-N +
1)] ^T , c _N = [c ₀ , c ₁ ,..., C _N−1 ] ^T Note that y _N (n) is the y represented by the equation (1).
(N) is an output vector of the adaptive filter 2 of N elements from the present to the past.

[0035] Here, (in the figure, represented by Hat c _N (∧)) value of the approximate digital filter 7 in Fig. 4 Filtered
Those used to obtain the (hat the u (n) (∧)) -X input signal vector has a correlation of certain degree of impulse response vector c _N of unknown transfer function ECTF.
_Let the tap length of the hat (∧) of c _N and c _{N be N} , the tap length of the adaptive filter h (n) be M (≧ 2N), and output the error signal e (n) and the hat (∧) of c _N. Signal u
The hat (∧) in (n) can be expressed as in equations (3) and (4) in FIG. 5, respectively.

In FIG. 4, the BOP algorithm for adaptive inverse filter estimation is given as in equation (5) in FIG. Where μ is the relaxation coefficient, j is the block number,
L is a block length and is given by L = N + r-1. Note that r
Represents the number of Filtered-X input signal vectors used for coefficient correction. At this time, U (j) (wh
(J)) uses the r-th order error signal vector e (j),
This is given by equation (6) in FIG. Where e (j) = (e
(JL-1), e (jL-2), ..., e (jL-r))
^T.

Also, a Filtered-X input signal matrix (U (j)
⁺ (Represented by a hat (∧)) represents a general inverse matrix of the Moore-Penrose type of the hat (∧) of the matrix U (j), and the matrix U
The hat (∧) in (j) is a matrix in which each element has a hat (∧) in u (n), and is represented as in equation (7) in FIG.

[0038] Further, w of the bandpass filter 5 is an inverted filter of unknown impulse response c _N, adaptive inverse filter as a result of the linear convolution of c _N and w is the target impulse response b bandpass filter 5 is fixed You. That is, as shown in Expression (8) in FIG. 5, c _N · w = b or w =
c _N ⁻¹ · b.

The length of b is N + M-1. Furthermore, the use of linear convolution matrix C is defined by the equation in FIG. 5 (9) instead of _{c N} in equation (2), Equation (8), as shown in equation (10), Cw = b And the matrix U (j) in equation (6) is expressed as U (j) = X (j) C, as shown in equation (11) in FIG. However, each element c
_i (i = 0, 1,..., N−1) corresponds to the components of the vector c _N.

FIG. 6 is an explanatory diagram showing the convergence characteristics by the BOP algorithm obtained by the equation (5). In the illustrated example, the convergence characteristics are 118 for both ears of 59 listeners, the horizontal axis represents time, and the vertical axis represents the estimation error ε (j) defined by the equation (12) in FIG. Represent. As shown in Expression (12), the estimation error is a normalized square error in the same frequency band as that of the bandpass filter 5. Where B (f), C (f), H
(F, j) are b and c _N , respectively, and the adaptive filter h
(J) is DFT at 2M points, and the range of the sum corresponds to the band of the bandpass filter 5 used for b.

By the way, the shorter the convergence time for the adaptive filter 2 to obtain the characteristic matching the inverse characteristic of the ECTF, the better. As described in the above-mentioned document (document on individuality of transfer functions ECTF and SSTF in out-of-head sound localization), the time until convergence varies depending on the error value and individual difference.

Therefore, the contents of the examination of the required error are described in Takashi Inada, Shohei Yano, Haruhide Hokari, and Masaharu Shimada: "Study on Required Estimation Accuracy of Inverse Filter in Out-of-head Sound Image Localization", 1999 IEICE Shin-etsu Branch Meeting (19
99.10.2). According to this, the error is −
It is reported that if it is less than 10 dB, more than 80% of people will get a feeling of localization of an out-of-head sound image.

As shown in FIG. 6, each listener has different convergence characteristics. Therefore, the convergence time during which the error becomes −10 dB or less is required from about 0.3 seconds to about 0.5 seconds. Moreover, this convergence is the case where white noise is used for input. It takes more time to converge with normal music and voice. Therefore, when applying the adaptive filter 2, it is necessary to consider an algorithm that can converge in a short time, and the present invention has realized a system that can converge in such a short time. Hereinafter, a specific example of such an out-of-head sound image localization device will be described.

<< Specific Example 1 >> Before describing a specific example of the present invention, the convergence characteristics of the adaptive filter 2 when the BOP algorithm is applied to Filtered-X will be theoretically described.

FIG. 7 is an explanatory diagram for explaining the operation of the specific example. Generally, the convergence characteristic is represented by a difference between the tap coefficient vector h (j) of the adaptive filter 2 and a target tap coefficient vector w. That is, assuming that the coefficient error vector is E (j), the equation (13) in FIG. 7 is obtained. At this time, the equation (5) in FIG. 5 is replaced by the equation (14) in FIG. It can be expressed as.

When the input signal x (n) is white noise,
From (11), U (k) in equation (14) ⁺Hat (∧)
U (k) is almost unit vector l because of the nature of white noise
Furthermore, the coefficient error norm should be evaluated as a scalar quantity.
Can be. This is shown by equation (15) in FIG.

As a result, the value in [] of the equation (14) is a constant scalar amount, and the ECTF ^-1 for each listener is now represented by w _i ,
The initial value is h (0), and the filter coefficient after j times of updating is h
Assuming (j), equation (16) in FIG. 7 is obtained. Where α
<1, w _i is the impulse response of the transfer function ECTF for each listener i. Since equation (16) is a discrete value system, if it is converted to a continuous time function t for simplicity, the coefficient error norm at time t ₀ when the initial value h (0) is given is expressed by equation (16) 17). Similarly, if the time required for the coefficient error norm to reach a certain value p is defined as t _j , Expression (18)
Becomes Given an initial value h (0), when determining the time T _i to reach a certain predetermined estimation error in equation (1
9).

From Equation (19), it is difficult for each listener to find an initial value that minimizes the time to reach the desired coefficient error by an analytical method. Then, it confirms by simulation. That is, the average convergence time T _ave of the N listeners can be expressed by Expression (20) in FIG. Σ Using the values after as parameters, find the minimum value of the total time.

In equation (20), the EC of a certain listener is
With the impulse response of TF ^{-1 as} an initial value h (0), the natural log of the norm (Euclidean distance) of the difference component of the impulse response w _i of ECTF ^-1 of N listeners is obtained, The sum total average of the listeners (sum total value of the natural logarithm of the coefficient error) is obtained, and the value is shown on the horizontal axis, while the average convergence time T _ave for each of the N listeners from the above initial value is shown on the vertical axis . The value of the ECTF ^-1 impulse response of the listener having the smallest average convergence time T _ave is obtained.

FIG. 8 is an explanatory diagram showing the result. FIG. 8 shows the center of gravity of the Euclidean distance as an initial value h.
The value when (0) is set is also shown.

As shown in FIG. 8, it can be seen that the average convergence time T _ave when the center of gravity is set to the initial value is short. To clarify this, the relationship between the convergence time until the evaluation error when changing the initial value of the tap coefficient value reaches -10 dB and the percentage of the cumulative number for each listener was measured.

FIG. 9 is a diagram showing the relationship between the convergence time and the percentage of the cumulative number for each listener. FIG. 9 shows the convergence time until the evaluation error when the initial value of the tap coefficient value is changed reaches −10 dB on the horizontal axis, and the vertical axis shows the percentage of the cumulative number for each listener. .

In FIG. 9, (a) shows a case where the impulse response of the listener closest to the center of gravity is set as an initial value, (b) shows a case where the center of gravity is set as an initial value, and (c) shows an initial value h (0). ) =
0 or no initial value. Thus, when the initial value is set near the center of gravity of the Euclidean distance, 50% or more of the listeners instantly converge, and after 0.2 seconds, 8% of the listeners
It can be seen that 0% or more converges.

Therefore, in the out-of-head sound image localization, even if there is a different transfer function ECTF for each listener, the impulse response corresponding to the center of gravity of the Euclidean distance or the ECTF ^-1 of a certain listener close to the center of gravity is obtained. It can be seen that by giving the impulse response as an initial value, the convergence time is shortened to 4 to 5 times. From such a basic idea, a specific example as described below will be described.

<Structure of Embodiment 1> FIG. 1 is a functional block diagram showing Embodiment 1 of the out-of-head sound image localization apparatus of the present invention.
The device shown in the figure has a digital filter 1, an adaptive filter 2,
Earphone (speaker) 3, microphone 4, bandpass filter 5,
The adaptive filter 2, the approximate digital filter 7, the convergence calculation circuit 8 and the memory 10 constitute an external ear canal transfer function adaptive operation circuit 9.

As shown in FIGS. 3 and 4, the digital filter 1 previously stores the value of (spatial acoustic transfer function SSTF / transfer function LSTF of a sound source located in space) (≒ head acoustic transfer function HRTF). The filter is configured to input a sound source signal and transmit an output x (n).

The adaptive filter 2 has an external auditory canal transfer function (EC
This is a filter for calculating an inverse transfer function (1 / ECTF) of the TF), receives a sound source signal via the digital filter 1, and sends out an output y (n) to the earphone 3.

The earphone 3 is a speaker arranged to be in contact with the listener's ear. The microphone 4 is a microphone for collecting the acoustic signal transmitted from the earphone 3 in the external auditory canal of the listener, and outputs the output z (n) to the subtractor 6.

The bandpass filter 5 has an ear canal transfer function ECT
The product of F and the inverse transfer function (1 / ECTF) of the ear canal transfer function obtained by the adaptive filter 2 has a filter characteristic such that it becomes a bandpass filter characteristic. Filter.

The subtracter 6 outputs the output d of the bandpass filter 5.
The subtracter subtracts (n) from the output z (n) from the microphone 4 and sends out the output e (n) as an error signal.

The approximate digital filter 7 determines the value of the center of gravity of the ear canal transfer function ECTF of a plurality of listeners obtained up to that time or the value of the ear canal transfer function ECTF of the listener closest to the center of gravity. This is a filter that stores a value of a hat (∧) of the approximate transfer function c _N , receives a sound source signal via the digital filter 1, and outputs
It is configured to send the hat (∧) of (n).

The convergence calculation circuit 8 uses the hat (∧) of the output u (n) from the approximate digital filter 7 and the adaptive filter 2 based on the error signal e (n) from the subtractor 6.
Has a function of updating the value of the tap coefficient h.

The memory 10 is a memory for storing the hat (∧) of the value w of the inverse impulse response of the ear canal transfer function as the initial value of the adaptive filter 2.

The digital filter 1, the adaptive filter 2, the bandpass filters 5 and the convergence operation circuit 8 are provided, for example, on a personal computer, and are constituted by software corresponding to each function and a processor and a memory for executing the software. Each function is realized by dedicated hardware or dedicated hardware.

<Operation of Specific Example 1> First, an impulse response corresponding to the center of gravity of the Euclidean distance of the impulse response of ECTF measured from a plurality of listeners in advance is obtained, and the coefficient (c
_N (ッ ト)) is stored. Further, an inverse impulse response corresponding to the center of gravity of the Euclidean distance of the inverse impulse response of the ECTF is obtained, and the value is stored in the memory 10. Here, as described above, the impulse response may be the impulse response of a specific listener close to the center of gravity.

Next, in order to operate this circuit, the value of the memory 10 is input to the coefficient value of the adaptive filter 2 as soon as the power is turned on. As described above, when the listener wears the earphone 3 on the auricle and a sound source signal is input, the signal x passes through the digital filter 1 which specifies the sound source direction in advance.
(N) is generated. The signal x (n) is supplied to the bandpass filter 5
Is input to the subtractor 6 with the target band of the signal and the time delay.

On the other hand, at the same time as the signal x (n) is input to the adaptive filter 2, the sound radiated from the earphone 3 is received by the microphone 4 via the external auditory canal and input to the subtracter 6. The output signal of the subtractor 6 becomes an error signal e (n).
The convergence calculation circuit 8 updates the coefficient of the adaptive filter 2 by the BOP algorithm, and generates an impulse response substantially reverse to the impulse response of the ECTF of the listener.

As described above, the initial value of the adaptive filter 2 is
By inputting the impulse response of the ear canal transfer function ECTF of a large number of listeners, in this example, the inverse impulse response, a fast convergence time can be obtained, and even if the external noise is mixed in, the adaptive operation is unstable. If the unstable state can be detected, the control system can perform a stable operation by inputting the information of the initial value from the memory 10 again.

<Effects of Specific Example 1> Specific Example 1 as described above
According to the above, a transfer function that matches the listener is obtained by the adaptive operation of the ear canal transfer function ECTF, and the value of the center of gravity of the ECTF- ¹ of a plurality of listeners is used as an initial value during the adaptive operation. Therefore, general-purpose use can be performed for an unspecified number of listeners, and an adaptive operation can be performed in a short time.

<< Example 2 >> In Example 2, the initial value is updated each time the adaptive operation is performed. For example, the adaptive operation of the adaptive filter 2 is performed on a certain listener, and the converged coefficient of the adaptive filter 2 corresponds to ECTF ^-1 of the listener at that time. Therefore, it is conceivable to perform the next adaptive operation of the adaptive filter 2 using the converged coefficient of the adaptive filter 2.

For example, if the initial value of the coefficient of the adaptive filter 2 used in the current adaptive operation is h ₀ (m) and the coefficient of the adaptive filter 2 converged at this time is h _∞ (m), then the adaptive operation Initial value h ₀ of the coefficient of the adaptive filter 2 when performing
(M + 1)

H ₀ (m + 1) = αh ₀ (m) + (1−
α) h _∞ (m).

By converging the initial value of adaptive filter 2 in this manner, the number of listeners who take the average of ECTF- ¹ is increased. Also, assuming that only one listener uses this device and uses the adaptive filter 2 again every time the device is started up, the initial value that is more consistent with the ECTF ^-1 of that listener is Given. Hereinafter, an out-of-head sound image localization apparatus that realizes such a function will be described as a specific example 2.

<Structure of Specific Example 2> FIG. 10 is a structural diagram of an out-of-head sound image localization apparatus of Specific Example 2. The illustrated out-of-head sound image localization apparatus includes a digital filter 1 to a memory 10 and an initial value calculation / setting update circuit 1 as in the first embodiment.
1 is provided.

The initial value calculating / setting updating circuit 11 uses the adaptive filter 2 used at the time of the previous adaptive operation of the adaptive filter 2.
Is stored in the memory 10.

<Operation of Specific Example 2> First, an impulse response corresponding to the center of gravity of the Euclidean distance of the ECTF impulse response measured from a plurality of listeners in advance is obtained, and the approximate impulse response coefficient (c
_N (ッ ト)) is stored. This operation is the same as in the first embodiment.

The initial value calculation / setting updating circuit 11
As the initial value, for example, ECT which is the same value as in the first example
The value of the inverse impulse response corresponding to the center of gravity of the Euclidean distance of the inverse impulse response of F is stored in the memory 10.
Alternatively, the initial value may be stored in the memory 10 as an initial value of 0 from the viewpoint that there is no listener up to the previous time.

Thereafter, as in the first embodiment, the adaptive filter 2
And the adaptive filter 2 has the listener's E
CTF ^-1 is set. Thereby, the initial value calculating / setting updating circuit 11 updates the initial value based on the converged value of ECTF- ¹ , and stores the updated value in the memory 10.

The above operation is performed when the coefficient of the adaptive filter 2 is
Perform each time convergence and update the initial value stored in memory 10
I will do it. Therefore, the adaptation of the adaptive filter 2 is performed for different listeners.
ECT of many listeners
F^-1Center of gravity is required and can be applied to more listeners.
Device can be realized. For this reason, note first
The value stored in the file 10 is not necessarily the value of many listeners.
ECTF calculated from the value ^-1Has the effect of not requiring the value of
You.

The value set in the approximate digital filter 7 is fixed, and is not updated like the initial value of the memory 10. This is because the value of the approximate digital filter 7 does not significantly affect the convergence time of the coefficient in the adaptive filter 2, and the ECTF value corresponding to the value updated by the initial value calculation / setting update circuit 11 is complicated. This is to avoid complicated arithmetic processing.

<Effect of Specific Example 2> Specific Example 2 as described above
According to the first embodiment, in addition to the configuration of the first embodiment, an initial value calculation / setting update circuit 11 that updates the initial value of the memory 10 based on the convergence value of the adaptive filter 2 up to the previous time is provided, and is stored in the memory 10. First, there is an effect that a value based on a large number of listeners is not required as an initial value. Further, assuming a case where the present apparatus is used by one listener and the adaptive filter 2 is adapted again each time the apparatus is started up, an initial value that is more consistent with the ECTF- ¹ of the listener becomes It has the effect of being given.

<< Usage Mode >> The out-of-head sound image localization apparatus of the present invention is a digital audio tape recorder (DAT).
In addition, the present invention can be used not only for a listening device such as a compact disk (CD) or a virtual reality system but also for ideal voice communication using a communication line.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a specific example 1 of an out-of-head sound image localization apparatus of the present invention.

FIG. 2 is a diagram illustrating the principle of realizing out-of-head sound image localization.

FIG. 3 is a functional block diagram for explaining the principle of the present invention.

FIG. 4 is a functional block diagram showing a configuration for realizing the system of FIG. 3;

FIG. 5 is an explanatory diagram for explaining the operation of the principle of the present invention.

FIG. 6 is an explanatory diagram showing convergence characteristics by a BOP algorithm.

FIG. 7 is an operation explanatory diagram for explaining an operation of a specific example.

FIG. 8 is an explanatory diagram of an average convergence time.

FIG. 9 is a diagram showing the relationship between the convergence time and the percentage of the cumulative number for each listener.

FIG. 10 is a configuration diagram of an out-of-head sound image localization device of a specific example 2.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Digital filter 2 Adaptive filter 3 Earphone (speaker) 4 Microphone 5 Band filter 6 Subtractor 7 Approximate digital filter 8 Convergence operation circuit 9 External ear canal transfer function adaptive operation circuit 10 Memory 11 Initial value calculation / setting update circuit

 ──────────────────────────────────────────────────続 き Continuation of the front page (71) Applicant 500139822 Shunji Horiuchi 1603-1 Kamitomiokacho, Nagaoka City, Niigata Prefecture Nagaoka University of Technology (72) Inventor Shinichi Kawada 1-7-12 Toranomon, Minato-ku, Tokyo Oki Electric Within Industrial Co., Ltd. (72) Inventor: Masaharu Shimada 1603-1, Kamitomioka-cho, Nagaoka, Niigata Prefecture Within Nagaoka University of Technology (72) 72) Inventor Shunji Horiuchi 1603-1 Kamitomiokacho, Nagaoka City, Niigata Prefecture F-term in Nagaoka University of Technology 5D011 AB11 5D062 AA71 AA74 5J023 DA05 DB03 DB05 DC00 DC01 DD05

Claims (4)

[Claims] 1. A speaker provided in contact with both ears and generating a sound wave based on a sound source signal; a microphone collecting sound waves generated from the speaker in an external auditory canal; Transfer function of sound source)
And a function of inputting a sound source signal through the digital filter and outputting the signal to the speaker, the inverse of the ear canal transfer function being a transfer function from the speaker to the output of the microphone. An adaptive filter for obtaining an impulse response value; a memory for storing an initial value for obtaining an inverse impulse response value in the adaptive filter; an impulse response value of the ear canal transfer function; and an ear canal transfer function obtained by the adaptive filter A product of the inverse impulse response value of the filter has a filter characteristic to be a band filter characteristic, a band filter that inputs a sound source signal through the digital filter, an output of the band filter, and an output of the microphone A sound source signal via the digital filter, and a predetermined value. An approximate digital filter for storing the value of the approximate transfer function, and using the value of the approximate transfer function of the approximate digital filter, based on the subtraction result of the subtractor, until the inverse impulse response value of the adaptive filter converges. An out-of-head sound image localization device, comprising: a convergence calculation circuit for updating an inverse impulse response value. 2. The out-of-head sound image localization apparatus according to claim 1, wherein the memory stores an inverse impulse response value corresponding to a center of gravity of a Euclidean distance of an inverse impulse response of an ear canal transfer function of a plurality of listeners. An external sound image localization apparatus characterized in that an approximate digital filter stores impulse response values corresponding to the center of gravity of the Euclidean distance of impulse responses of a plurality of listeners' ear canal transfer functions. 3. The out-of-head sound image localization apparatus according to claim 1, wherein the memory stores the inverse impulse response value of the listener closest to the center of gravity of the Euclidean distance of the inverse impulse response of the ear canal transfer function of the plurality of listeners. And an approximate digital filter storing an impulse response value of the listener closest to the center of gravity of the Euclidean distance of the impulse response of the external auditory canal transfer function of a plurality of listeners. 4. The out-of-head sound image localization apparatus according to claim 1, wherein the initial value of the inverse impulse response value of the ear canal transfer function used in the previous adaptive operation is the inverse impulse of the adaptive filter previously converged. An out-of-head sound image localization apparatus comprising: an initial value calculation / setting update circuit for storing a value updated using a response value as a new initial value in a memory.