JP3395809B2 - Sound image localization processor - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a sound image localization processing device capable of imparting a sense of direction or a sense of distance to sound over a wide frequency band.

[0002]

2. Description of the Related Art Human beings perceive the distance and direction of a sound source by listening to sound with both ears, using the interaural time difference, the interaural level difference, frequency characteristics, etc. as criteria. In other words, it has a sense of localization. Therefore, even when sound is reproduced in both ears by a sound reproducing device such as headphones, it is possible to perform sound signal processing for localizing the sound at a desired position.

Conventionally, it has been proposed to localize a sound image by convolving an acoustic signal with a sound source and the acoustic transfer characteristics to the left and right ears of the listener and listening. That is, the acoustic transfer characteristics between the sound source and the left and right of the listener in FIG. 6A are H _l and H _r (hereinafter, H ₁ and H _r are collectively referred to as transfer function), and as shown in FIG. 6B. transmitted to the sound source signal S function H _l, convolving H _r in parallel, transferred to the left and right ears with a headphone or the like function H _l, H acoustic signal _r is convolved H _l · S, by H _r · S By presenting the sound, the time variation of the sound wave in both ears of FIG. 6A is reproduced. Also in FIG. 6B, the listener can perceive a sound image at the position of the sound source in FIG. 6A.

[0004]

However, under the present circumstances, it is difficult to accurately measure the transfer function over a wide band due to the limitation of the measurement system. That is, the frequency band in which the transfer function can be accurately measured is limited to a band narrower than the audible frequency band. When a narrow-band transfer function is convoluted with a wide-band source signal for listening, the components of the source signal in the frequency band not limited by the transfer function are lost by the convolution calculation of the transfer function. That is, since the convolution operation in the time domain is equivalent to the multiplication in the frequency domain, the convolution operation results in the loss of the sound source signal component in the frequency band not limited by the transfer function, resulting in the deterioration of the sound quality and the loss of presence. .

An object of the present invention is to provide a sound image localization processing device which can localize a sound in a wider frequency band by using a transfer function in a narrow frequency band without deteriorating the feeling of localization.

[0006]

In the invention described in claim 1 of this application, the sound source signal is limited in the transfer function in order to provide a means for localizing a sound in a frequency band wider than the transfer function. A frequency band component, a band branching unit for branching into an unrestricted frequency band component, a convolution operation unit for convolving a transfer function with a sound source signal component in a frequency band restricted in the transfer function, and a frequency band not restricted in the transfer function Means for adjusting the delay and gain of the sound source signal component of, and adding means for adding the sound source signal component of which the transfer function is convoluted and the sound source signal component of which the delay and gain are adjusted ,

[0007] Transfer set by transfer function setting means
A control unit that calculates delay and gain based on a function, and this control
Transfer function using delay and gain calculated in the control section
The audio signal component of the band which is not limited in
A delay and gain adjusting means for adjusting the delay and gain is provided.
It proposes a configuration.

In the invention according to claim 2 of this application, the transfer for calculating the transfer function in the frequency band not restricted in the transfer function by using the delay and the gain obtained for the sound source signal component in the frequency band not restricted in the transfer function. A function calculating means is provided, the transfer function calculated by the transfer function calculating means is combined with the set transfer function, a wideband transfer function is obtained from the set transfer function, and the wideband transfer function is convoluted with the sound source signal. The proposed sound image localization processing device is configured.

[0009]

The present invention utilizes the human auditory characteristic of the sound generated by adding the sound source signal in which the transfer function is convoluted and the sound source signal component in the frequency band not limited by the transfer function. That is, in human perception of sound image localization, there are individual differences, but the up-and-down direction is used by using the frequency characteristics of the sound in the high frequency range of approximately 1 kHz or more, and the interaural time difference is mainly estimated in the low frequency range of approximately 1.5 kHz or less. In the high frequency range of 1.5 kHz or more, the left-right direction is mainly determined by using the interaural level difference, and the sound image position is perceived by combining these.

When the frequency band of the sound source signal is wider than the transfer function, the sound source signal is branched into components of a frequency band limited by the transfer function and a frequency band not limited by the transfer function, and the sound source signal of the frequency band limited by the transfer characteristic. The transfer function is convoluted with the component, and the time difference and the gain difference between both channels are given by adjusting the delay and the gain for the source signal component of the frequency band which is not limited by the transfer function. A sound source signal component in which a transfer function is convoluted and a sound source signal component in which delay and gain are adjusted are added, and a sound reproduced from the added sound signal is listened to, or a transfer function having a wide band is obtained. By listening to the sound reproduced from the audio signal obtained by convolving the transfer function with the sound source signal, it is possible to obtain a sound image localization feeling for the wideband sound source signal.

[0011]

[Embodiment] FIG. 1 shows an embodiment of a sound image localization processor proposed in 請 Motomeko 1 of the present invention. An acoustic signal (monaural signal) is input to the input terminal 1. The acoustic signal input to the input terminal 1 is split into two signals at the acoustic signal branch point 2 and input to stereo signal generation circuits 100L and 100R for generating left and right stereo signals.

Stereo signal generation circuits 100L and 100R
The band branching means 10-1 and 10- are provided on the front side of each of them.
Two are provided. These band branching means 10-1 and 1
0-2, convolutional operators 23-1 and 23-2 described later.
The frequency band component limited in the transfer function used in the convolution operation performed in (1) and the frequency band component not limited in the transfer function are branched.

In the illustrated embodiment, the high-pass filters 11-1 and 11- are provided at the acoustic signal branch points 3-1 and 3-2, respectively.
2 and low pass filters 21-1 and 21-2 are connected,
The high-pass side frequency component W ₁ (FIG. 2) which is not limited in the transfer function is extracted from the high-pass filters 11-1 and 11-2, and the low-pass filters 21-1 and 21-2
The low frequency component W ₂ + W ₃ including the frequency component W ₂ limited in the transfer function is taken out from.

Acoustic signal branch points 4-1 and 4-2 are further provided on the output sides of the low-pass filters 21-1 and 21-2. Analog / digital converters 22-1 and 22-2 and low-pass filters 31-1 and 31-2 are connected to the acoustic signal branch points 4-1 and 4-2. The low pass filters 31-1 and 31-2 are the low pass filters 21.
Low-frequency components W ₂ + W ₃ extracted by branching at −2 and 21-2
The operation of extracting the signal component W ₃ of a lower frequency which is not limited in the transfer function included in

High-pass filters 11-1, 11-2 and
The transfer function extracted by the low-pass filters 31-1 and 31-2
Frequency component W not limited in number₁, W₃Acoustic communication
No. is amplifiers 12-1, 12-2 and 32-1, 32-3.
And then A_Hr, A_LrAnd A_Ll, A_LrDoubled and further delayed
Delayed by 13-1, 13-2 and 33-1, 33-2
τ _Hl, Τ_HrAnd τ_Ll, Τ_Lr(Msec) is added and added
It is input to the calculators 5-1 and 5-2.

Analog / digital converters 22-1, 2-2
In 2-2, the low-pass frequency components W ₂ + W ₃ branched by the low-pass filters 21-1 and 21-2 are directly analog / digital converted into the sampling frequency f _m of the transfer function, and the digital signal is convoluted. To the vessels 23-1 and 23-2. The convolution calculators 23-1 and 23-2 convolve the transfer functions H _l (t) and H _r (t) with the digitized acoustic signal, respectively. The digital acoustic signal in which the transfer functions H _l (t) and H _r (t) are convoluted is converted into an analog signal by the digital / analog converters 24-1 and 24-2, and the analog signal is added to the adder 5-1. , 5-2, and 13-
1, 13-2 and 33-1 and 33-2 are added to the signals of the frequency components not limited in the transfer functions H _l (t) and H _r (t) output from the acoustic signal output terminals 6-1 and 6-2 outputs as a stereo signal.

In addition to the above configuration, the present invention further includes a setting section 51, a storage section 52, and a control section 53. Setting unit 5
1, the virtual sound source position, the position of the listener with respect to the sound source,
Set the orientation of the listener. The storage unit 52 stores the transfer functions H _l (t) and H _r (t). The control unit 53 controls the transfer functions H _l (t) and H _r (t) and other variables.
The flow of processing in the control unit 53 will be described with reference to FIG.

The position of the listener set in the setting section 51 and
The transmission function stored in the storage unit 52 based on the direction and the sound source position.
Number H_l(t), H_rCall (t). Called transfer function
Number H _l(t), H_r(t) allows amplifiers 12-1, 12-
2 and the gain A set for each of 32-1 and 32-2
_Hl, A_Hr, A_Ll, A_LrAnd delay devices 13-1, 13-2 and
And delay τ set for each of 3-1, 33-2_Hl,
τ_Hr, Τ_Ll, Τ_LrTo calculate. The gain A_Hl, A_Hr…etc
And delay τ_Hl, Τ_HrEach subscript H, L, l, r such as ... is that
H is high side, L is low side, l is left channel, r is
Means right channel.

The transfer function H_l(t), H_r(t) it
Transfer to the convolutional computing units 23-1 and 23-2, respectively, and gain
A_Hl, A_HrTo amplifiers 12-1 and 12-2, respectively.
Profit A _Ll, A_LrTo amplifiers 32-1 and 32-2,
Delay τ_Hl, Τ_HrTo delay units 13-1 and 13-2, respectively.
And the delay τ_Ll, Τ_LrTo the delay units 33-1 and 33-
Transfer to 2.

The transferred transfer function, gain and delay are
A convolution operation on the acoustic signal at each transfer destination,
Used as a variable for amplification and delay. In addition,
A limited frequency band in the reaching function, that is, the transfer function H
_l(t), H_rUpper cutoff frequency f of (t)_HAnd lower cutoff circumference
Wave number f_L(See FIG. 2) and the sampling frequency f of the transfer function._mAlso
Treated Here, the upper cutoff frequency f of the transfer function_HIs high
Low-pass filters 11-1 and 11-2
21-1 and 21-2 are given as cutoff frequencies,
Lower cutoff frequency f of transfer function_LIs a low-pass filter 31
-1, 31-2 are used as cutoff frequencies. Folding performance
Calculations in the calculations 23-1 and 23-2 are performed digitally
Therefore, the sampling frequency f of the transfer function is_mIs analog / de
Analog sound in the digital converters 22-1 and 22-2
Sound signals are converted to digital audio signals
Digital sound in the analog converters 24-1, 24-2
Required to convert the sound signal into an analog sound signal
It Therefore, the upper cutoff frequency f_HHigh pass filter
To 11-1 and 11-2, lower cutoff frequency f_LIs low pass
Sampling frequency f to filters 31-1 and 31-2_mA
Analog / digital converters 22-1, 22-2 and di
Transfer to digital / analog converter 24-1, 24-2
It Here, the upper cutoff frequency f_HAnd the lower cutoff frequency f_L
And the sampling frequency f _mIn the storage unit 52 with the transfer function
Stored in the beginning and refer to it when calling the transfer function.
The control for transferring to each of the above parts is executed by the control part 53.
It

Gain A in the control unit 53_Hl, A_Hr,
A_Ll, A_LrAnd delay τ_Hl, Τ_Hr, Τ_Ll, Τ _LrHow to calculate
The methods include the following methods. 1) Gain A_Hl, A_Hr, A_Ll, A_LrTo A_Hl= | H_l(F_H) | (1) A_Hr= | H_r(F_H) | (2) A_Ll= | H_l(F_L) | (3) A_Lr= | H_r(F_L) | (4) Set as follows. Here, the transmission to the called left ear
The reaching function is H_l(t), the transfer function for the right ear is H_r(t),
Transfer function H for the left ear_lThe frequency characteristic of (t) is H _l(f)
 , The transfer function H for the right ear_rThe frequency characteristic of (t) is H
_r(f). Where t is time, f is frequency, and | ... |
Shows the logarithmic value. And the delay τ_Hl, Τ_Hr, Τ_Ll, Τ_LrIn
The rise time of the reaching function is used. That is, τ_Hl= T {H_l(t)} (5) τ_Hr= T {H_r(t)} (6) τ_Ll= T {H_l(t)} (7) τ_Lr= T {H_r(t)} (8) And However, H_l(t), H_r(t) is from the sound source to both ears
To the transfer function, T {...} is the rise time
Show.

2) Gain A _Hl , especially for high frequency sound signals,
A _Hr is defined as A _Hl = <| H _l (f) |> (9) A _Hr = <| H _r (f) |> (10) However, <...> represents an average value from a certain frequency f _C (<f _H ) to the upper cutoff frequency f _H. The reason for adopting this method is that the absolute values of the frequency characteristics of the transfer function | H _l (f) | and | H _r (f) |
It can be said that in the high region near _H , the frequency changes remarkably. However, in the low frequency range, the change of the transfer function with frequency is generally gradual, and thus the gains A _Ll and A _Lr for the low frequency range acoustic signal are determined using the equations (3) and (4).

3) Further, the time difference between the left and right ears is a low frequency component.
Contributes to the left-right localization with respect to
Considering that the ear-to-ear time difference becomes larger than the high range,
Total τ _Hl, Τ_Hr, Τ_Ll, Τ_LrTo τ_Hl= Θ_Hl/ 2πf_H    (11) τ_Hr= Θ_Hr/ 2πf_H    (12) τ_Ll= Θ_Ll/ 2πf_L    (13) τ_Lr= Θ_Lr/ 2πf_L    (14) There is also a method to determine. Where θ_Hl, Θ_HrIs the upper cutoff frequency
Number f_HFrequency characteristics H of the transfer function to the left and right ears at_l
(F_H), H_r(F_H) Continuous phase (unwrapp
ed phase). Similarly, θ_Ll, Θ_LrIs below
Limiting cutoff frequency f_LOf the transfer function to the left and right ears in
Number characteristic H_l(F_L), H_r(F_L) Represents the continuous phase of
You

4) A gain is obtained by using a mathematical model shown in FIG.
A_Hl, A_Hr, A_Ll, A_LrAnd delay τ_Hl, Τ _Hr, Τ_Ll, Τ_Lr
A method for calculating is shown. If the sound waves scattered by the human head are
Rigid sphere of sound waves generated from a point source in a space without reflection
Hypothetical sound source, assuming that it can be approximated to scattering by
To observation point J on the surface of the sphere corresponding to both ears_l, J_rWhat
Transfer function H of_l, H_rIs expressed by the following equation.

[0025]

[Equation 1] Where h _n ⁽²⁾ (r) is the spherical Bessel function, P
_n ^m (cos θ) is the Legendre function,
h _n ⁽²⁾ ′ (kr ₀ ) = ∂h _n ⁽²⁾ (x) / ∂x _{x = kr0} ,
k is the wave number of the sound wave, r ′ is the distance between the sound source and the center of the sphere, r ₀
Is the radius of the sphere, ψ is the azimuth angle between the sound source direction and the front direction of the sphere (a positive value when the sound image direction is left with respect to the observation points J _l and J _r ), and θ is the vertex direction of the sphere and the observation point J. _The perspective angle formed by _l and J _r , and θ ₀ is the perspective angle formed by the vertex of the sphere and the sound source direction. An example is considered in which a human head width is applied as the radius r ₀ of the sphere. Α is the front direction of the sphere and the observation points J _l , J _r
Is an azimuth angle formed by, and in particular, assuming that the left and right observation points J _l and J _r are in opposite azimuth angles on the horizontal plane, θ
= Π / 2 and α = π / 2. And (15), (1
The gains A _Hl , A _Hr , A _Ll , and A _Lr are obtained by applying the transfer functions H _l and H _r obtained by the formula (6) to the formulas (1), (2), (3), and (4), for example. , (11), (1
The delays τ _Hl , τ _Hr , τ _Ll , and τ _Lr are calculated by applying the expressions (2), (13), and (14).

Like the cutoff frequency, etc., the gain A _Hl obtained by any one of the above 1) to 4),
A _Hr , A _Ll , A _Lr and delays τ _Hl , τ _Hr , τ _Ll , τ _Lr are calculated in advance, stored in the storage unit 52 together with the transfer function, and transferred to the above units. Procedures are also possible.
Although the above-described embodiment shows the sound image localization processing device that inputs an analog signal and outputs an analog signal, a digital / analog converter is provided between the sound source signal input terminal 1 and the acoustic signal branch point 2 of the embodiment. To configure a digital input and digital output type sound image localization processing device by connecting an analog / digital converter between the adders 5-1 and 5-2 and the acoustic signal output terminals 6-1 and 6-2, respectively. You can

Further, as another embodiment of the sound image localization processor of the digital input and digital output type, a digital / analog converter, an analog / digital converter and an analog / digital converter which are inserted in the input side and the output side are provided.
The digital converters 22-1 and 22-2 and the digital / analog converters 24-1 and 24-2 become unnecessary, and the sampling frequency of the sound source signal is transferred to the analog / digital converters 22-1 and 22-2. Of the sampling frequency f _m in the sampling frequency converter and the digital / analog converters 24-1 and 24-2 for converting the sampling frequency f _m into
A configuration including a sampling frequency converter that converts the sampling frequency of the sound source signal is conceivable.

Further, in addition to the above-mentioned embodiment, the source signal (that is, the branch point 2
Even if the transfer function whose bandwidth is widened in the convolutional computing units 23-1 and 23-2 is simply convoluted without branching and splitting the left and right outputs), the components and delay and gain of the transfer function are convolved. An acoustic signal equivalent to the sum of the adjusted components (that is, the output of the adders 5-1 and 5-2) is
Obtainable.

That is, when processing an analog audio signal, the delays τ _Hl , τ _Hr , τ _Ll , and τ _Lr described above are used as shown in FIG.
And transfer function calculation means 41-1 and 41-2 for obtaining a transfer function in the frequency domain which is not limited by the set transfer function based on the gains A _Hl , A _Hr , A _Ll , and A _Lr. 41-1 and 41-2 are provided, and transfer function synthesizing units 42-1 and 42-2 for synthesizing the transfer function calculated in 41-1 and 41-2 are provided.
Wideband sound localization is enabled by convolving the widebanded transfer functions obtained by 2-1 and 42-2 into digital acoustic signals by the convolution calculators 23-1 and 23-2. With this configuration, as a configuration for processing an analog sound source signal, as shown in FIG. 5, a sound source signal input terminal 1, an acoustic signal branch point 2, an analog / digital converter 22-1,
Only 22-2, convolution operation units 23-1, 23-2 and digital / analog converters 24-1, 24-2 are required.

When processing a digital source signal,
In addition to the configuration described above, the analog / digital converter 22
-1, 22-2 and digital / analog converter 24-
1, 24-2 becomes unnecessary, and a simple acoustic signal processing system can be realized. However, it is necessary to equalize the band-wide transfer function and the sampling frequency of the sound source signal. The following methods 5) and 6) may be mentioned in order to combine the transfer functions in the transfer function combining units 42-1 and 42-2.

5) Gains A _Hl , A _Hr , A _Ll , A _Lr and delays τ _Hl , obtained by any one of the above 1) to 4)
From τ _Hr , τ _Ll , and τ _Lr , it is assumed that the gain does not change with the frequency f at the frequency f that is equal to or higher than the upper limit cutoff frequency f _H and equal to or lower than the lower limit cutoff frequency f _L. That is, the left and right transfer functions H _l (f) and H _r (f) displayed in the frequency domain are determined by the following equations.

[0032]     H_l(f) = A_Hlexp (2πifτ_Hl) (F> f_H) (17)     H_r(f) = A_Hrexp (2πifτ_Hr) (F> f_H) (18) Similarly, the lower cutoff frequency f_LAt frequency f below
Left and right transfer function H displayed in wavenumber domain_l(f), H
_r(f) is defined by the following equation.     H_l(f) = A_Llexp (2πifτ_Ll) (F <f_L) (19)     H_r(f) = A_Lrexp (2πifτ_Lr) (F <f_L) (20) Calculated by equations (17), (18), (19), and (20)
Combined transfer function and set transfer function, that is, the whole band
Inverse Fourier transform of the transfer function displayed in the frequency domain
Display in the time domain required for convolution by
Transfer function H _l(t), H_ris synthesized into (t).

6) Unlike the method 5), the gain is frequency
The frequency of the left and right transfer functions, assuming that
Number characteristic H_l(f), H_r(f) is the mathematical model (1
It is obtained by the equations 5) and (16). However, the left and right transmission functions
Number frequency characteristic H_l(f), H _r(f) is the upper cutoff frequency f
_HIn order to satisfy equations (17) and (18),
Radius of the sphere that is the variable of the scientific model₀And center of sound image and sphere
It is necessary to correct the distance r ′ between and. Corrected radius r
_Hl, R_HrCorrected distance, where is the radius of the sphere
r '_Hl, R ′_HrIs the distance between the sound image and the center of the sphere.
Upper limit cutoff frequency f_HLeft at frequency f above
Right transfer function H_l(f), H_r(f) (frequency domain display)
calculate. Lower cutoff frequency f_LSimilarly, in (1
9), variables of the mathematical model so as to satisfy the equations (20)
Radius r of₀And the distance r ′ between the sound image and the center of the sphere
to correct. Corrected radius r of the sphere_Ll, R_LrAnd amended
Distance r ′ between the sound image and the center of the sphere_Ll, R ′_LrLower bound using
Cutoff frequency f_LLeft and right transfer functions at frequency f below
H_l(f), H_r(f) Calculate (frequency domain display). Since
Below is the transfer function H displayed in the time domain as in 5).
_l(t), H_r(t) is the transmission displayed in the frequency domain of the entire band.
The reaching function is synthesized by performing an inverse Fourier transform.

[0034]

As described above, according to the present invention, means for band-splitting a sound source signal, means for convoluting a transfer function with a sound source signal component in a band limited by the transfer function, and limitation for the transfer function. A means for adjusting the delay and the level of the sound source signal component in the non-converted band, a means for adding the convoluted component and the component for which the delay and the level are adjusted, or a transfer function synthesizing section for widening the transfer function. By providing the sound image localization processing apparatus, it is possible to control the sound image localization by using the transfer function even for a sound source signal having a sampling frequency or a band wider than the transfer function.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a graph for explaining the operation of the embodiment of the present invention.

FIG. 3 is a flowchart for explaining a part of the operation procedure of the embodiment of the present invention.

FIG. 4 is a diagram for explaining an example of a processing method used in the sound image localization processing device according to the present invention.

FIG. 5 is a block diagram showing a modified embodiment of the present invention.

FIG. 6 is a diagram for explaining a conventional technique.

[Explanation of symbols]

1 Acoustic signal input terminal 2 Sound signal branch point 3-1, 3-2 Acoustic signal branch point 4-1 and 4-2 Acoustic signal branch point 5-1 and 5-2 adder 6-1, 6-2 Acoustic signal output terminal 10-1, 10-2 Band splitting means 11-1, 11-2 Bandpass filter 12-1, 12-2 amplifier 13-1, 13-2 Delay device 21-1, 21-2 Low pass filter 22-1, 22-2 analog / digital converter 23-1, 23-2 Convolutional computing unit 24-1, 24-2 Digital / Analog converter 31-1, 31-2 Low-pass filter 32-1 and 32-2 amplifier 33-1 and 33-2 delay device 41-1 and 41-2 transfer function calculating means 42-1 and 42-2 transfer function synthesis unit 51 setting section 52 storage 53 Control unit

─────────────────────────────────────────────────── ─── Continuation of the front page (56) Reference JP-A-4-79600 (JP, A) JP-A-5-292600 (JP, A) JP-A-8-102999 (JP, A) (58) Field (Int.Cl. ⁷ , DB name) H04R 3 / 00,5 / 00 H04S 1 / 00,5 / 00,7 / 00

Claims (2)

(57) [Claims] 1. In a sound image localization processing device capable of locating a sound image at a virtual position by performing signal processing that simulates the spatial transfer characteristic of a sound wave on a sound source signal, a listener from a virtual sound source as the spatial transfer characteristic Setting means for setting a transfer function up to both ears, band branching means for branching a sound source signal into a frequency band component limited in the transfer function and a frequency band component not limited in the transfer function, and a frequency band limited in the transfer function Means for convolving the transfer function with the source signal component of, the means for adjusting delay and gain for the source signal component of the frequency band not limited in the transfer function, the source signal component with the transfer function convolved and the delay and means for gain adds sound source signal component which has been adjusted, before Symbol group transfer functions set in the setting means of the transmission characteristics Based on the delay and gain, and means for adjusting the delay and gain of the source signal component of the frequency band not limited by the transfer function using the delay and gain calculated by the delay and gain calculating means . Ruoto image localization processor to <br/> characterized in that configuration was. 2. In a sound image localization processing device capable of locating a sound image at a virtual position by performing signal processing that simulates the spatial transfer characteristic of a sound wave on a sound source signal, a listener from a virtual sound source as the spatial transfer characteristic Setting means for setting a transfer function up to both ears, and means for calculating a delay and a gain for adjusting a sound source signal component of a frequency band not limited by the transfer function based on the transfer function set by the setting means, A transfer function calculating means for newly calculating a transfer function of a frequency band not limited by the transfer function using the delay and gain calculated by the delay and gain calculating means, and a transfer calculated by the transfer function calculating means A transfer function combining unit for combining the function and the set transfer function to obtain a transfer function in a wider band than the set transfer function. A sound image localization processing device characterized by being configured.