JP3367625B2 - Sound image localization control device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention
Any desired position different from the installation position of the speaker (speaker)
Sound localization control device that makes the sound image seem to be localized
Installation, especially for amusement game machines and computers.
Excellent sound image localization that can be installed on data terminals, etc.
And a sound image localization control device capable of reducing the circuit scale.
Things. 2. Description of the Related Art Conventionally, signal level differences between both ears have been known.
Sound source at a specific position (direction) depending on the phase difference (time difference)
There is a sound image localization device that makes you feel. For example, Japanese Unexamined Patent Publication
298200, "Sound image forming method and device therefor"
Left and right channels emitted from the two speakers
Gives a frequency-dependent level difference and phase difference to the channel signal
Instead, it controls the localization of the sound image. In addition, this application
Person is Japanese Patent Application No. 4-343449 and Japanese Patent Application No. 4-356
No. 358 was proposed. This is the HRTF (head-related transfer function)
To determine sound characteristics and localize sound image based on actual measurements
It is like that. [0003] In either case, the amplitude frequency characteristics and
Process phase frequency characteristics by signal processing and localize sound image
I have. [0004] However, the sound image
The localization position is dispersed at the intended position (the sound image is blurred),
There was a problem that it deviated from the intended position. So book
The applicant has incorporated the psychological element of HRTF and
In order to improve the localization, Japanese Patent Application No. 5-34946 was proposed.
This improves the localization of the sound image using a band-by-band correction process.
It is a thing. The present invention is not limited to the position of the localization of such a sound image.
Based on these improvements, the external environment or
Is always due to individual differences due to differences in the distance between the ears.
Sound image localization control that can optimally obtain sound image localization
It is intended to provide a device. [0005] In order to solve the above problems,
In addition, the present invention provides a sound image localization control device having the following configuration.
Offer. A signal output from a sound source (X) is subjected to a convolution operation.
Audio signal is transmitted to a pair of transducers (speakers sp
1, sp2), the pair of transducers
In a position different from the arrangement of the speakers (speakers sp1, sp2)
Sound image localization control device that performs sound image localization so that the sound image is localized
1 and a signal output from the same sound source (X) is supplied.
A pair of convolution operation processing circuits (convolvers) 2
Of a pair of transducers (speakers sp1, sp2)
Sound image localization to localize the sound image at a position different from the arrangement
Coefficient (coefficient data (cfLx, c
fRx)) to the convolution operation processing circuit (convolver) 2
Coefficient memory 3 to be supplied and listening position (position of listener M)
To periodically change the angular position of the localized sound image at
(For example, a sound image is shaken ± 10 degrees every 3 to 5 seconds.
To ensure the recognition of the localized sound image
A sound image localization control device, characterized in that: [0008] The sound image localization control device according to the present invention will be described below.
This will be described with reference to the drawings below. FIG. 1 shows a sound image localization system according to the present invention.
FIG. 2 is a configuration diagram of an embodiment of a control device, and FIG.
FIG. 3 is a diagram for explaining mapping to be corrected, and FIG.
FIG. 4 is a diagram for explaining how the image is shaken. FIG.
FIG. 5 is a block diagram showing the basic principle of sound image localization control, and FIG.
Shows the configuration of the HRTF (head-related transfer function) measurement system
FIG. 7 is a diagram for explaining the points of HRTF measurement, FIG.
FIG. 9 is a diagram for explaining a calculation example of a cancel filter, and FIG.
A specific example of IR (impulse response) of RTF is shown.
FIG. 10 and FIG. 10 show specific examples of coefficients of the cancel filter.
FIG. First, the basic principle of sound image localization control
explain. This consists of a pair of transformers
Uses a juicer (speaker as an example)
This is a technique for localizing a sound image at an arbitrary position in space. FIG. 2 is a block diagram showing the basic principle of sound image localization control.
In FIG. 5, sp1 and sp2 are the listeners (ie, the listeners).
Speakers placed on the front left and right of the person M), and sp1
Transfer characteristics from head to listener's left ear (impulse response)
H1L, head transfer characteristics to the right ear h1R, sp2
The transfer characteristics from the head to the left and right ears are defined as h2L and h2R.
In addition, an actual speaker is placed at the intended localization position x.
Head transfer characteristics to the listener's left and right ears when pLx, p
Rx. Here, each transfer characteristic is represented by a speaker in the acoustic space.
And the microphones on both sides of the dummy head (or head)
Appropriate waveform processing etc. are applied to the actual measurement
It was done. Next, the sound source X to be localized is signal-transformed.
Conversion devices cfLx, cfRx (transmission by convolver etc.)
The signals obtained through the characteristics are respectively referred to as sp1 and sp
Consider playing on 2. At this time, the listener
Assuming that the signals to be obtained are eL and eR, eL = h1L · cfLx · X + h2L · cfRx · X eR = h1R · cfLx · X + h2R · cfRx · X Here, both equations are combined to form (Equation 1). On the other hand, the source X is reproduced from the target localization position.
The signals obtained at the listener's left and right ears are dL and dR.
Then, dL = pLx · X dR = pRx · X Here, both equations are combined to obtain (Equation 2). Here, the listener listens by playing back the sp1 and sp2.
Signals from the left and right ears play the source from the target position
If the signal matches the signal at the time of the
The sound image is recognized as if a speaker is
You. The conditions eL = dL, eR = dR and (Equation 1), (Equation 1)
From 2), X is deleted and h1L · cfLx + h2L · cfRx = pLx h1R · cfRx + h2R · cfRx = pRx Here, both equations are combined to form (Equation 3). (Equation 3)
When cfLx and cfRx are obtained, cfLx = (h2R · pLx−h2L · pRx) / H cfRx = (− h1R · pLx + h1L · pRx) / H Here, both equations are combined to obtain (Equation 4a). Where H = h1L · h2R−h2L · h1R (Equation 4b) Therefore, (Equation 4a) and (Equation 4b)
Using the transfer characteristics cfLx and cfRx calculated
The signal to be localized by Luba (convolution processing circuit)
By processing, the sound image can be localized at the target position x.
Wear. There are various ways to realize a specific signal conversion device.
But asymmetric FIR digital filter (convolver)
What is necessary is just to realize using. In addition, FIR digital fill
The final transfer characteristic when used in a
You. That is, the transmission characteristics at the required localization position x
As the properties cfLx and cfRx, (Equation 4a) and (Equation 4b)
Realized by one FIR filter process
The coefficients for cfLx and cfRx are
It is created in advance and prepared as ROM data.
The required coefficient of the sound image fixed position is read from the ROM by the FIR digital file.
Filter, and convolve the signal from the sound source
Playback from a pair of speakers
The sound image is localized. Real sound image localization control based on the above principle
Will be described in the order of ~. Head Related Transfer Function
Function; hereinafter referred to as HRTF) As shown in FIG. 6, a dummy head (or
Human head) A pair of microphones ML and MR are installed in both ears of DM
And receives the measurement sound from the speaker SP and sends it to the recorder DAT.
Source sound (reference data) refL, refR
Sounds to be measured (measurement data) L and R are synchronously recorded. As the source sound XH, an impulse sound, an e
White noise and other noise can be used
You. In particular, from the viewpoint of statistical processing, white noise is
Energy distribution with continuous sound and over the audio band
Is constant, it is possible to obtain SN by using white noise.
The ratio improves. The position of the speaker SP is set to 0 degrees in front.
Angles θ in the space determined as
As shown, it is set at 12 points every 30 degrees)
Recording is continuously performed for a predetermined time. HRTF impulse response (Impulse Re
calculation of sponse (hereinafter referred to as IR).
Sound (reference data) refL, refR and measured
A constant sound (measurement data) L, R and a workstation
(Not shown). Source sound (reference data) frequency
The response is X (S), the frequency response of the measured sound (measurement data)
Is Y (S), and the frequency response of the HRTF at the measurement position is
Assuming IR (S), the input / output shown in the following (Equation 5)
There is a relationship. Y (S) = IR (S) · X (S) (Equation 5) Therefore, the frequency response of the HRTF is represented by IR (S) as follows: IR (S) = Y (S) / X (S) ... (Equation 6) Thus, the frequency response of the reference X (S),
The frequency response Y (S) of the measurement data is calculated in step 10
Cut out the data obtained in step 1 using a window synchronized with the time, and
Finite Fourier series expansion by FFT
HRTF is calculated as
The answer IR (S) is obtained by a well-known calculation method. In this case, the accuracy of IR (S) is increased (S
Hundreds of windows that differ in time to improve the N ratio)
It is better to calculate IR (S) and average them.
No. Then, the calculated HRTF frequency response IR (S)
Is subjected to inverse FFT transform, and the time axis response of HRTF (impulse
Response) IR (first IR). IR (Impulse Response) Shaping Process Here, the HRTF impulse response was obtained.
Shape the IR. First, for example, by FFT transform,
The first IR into a discrete circle over the audio spectrum
Unfolding by wave number, unnecessary band (large dip in high frequency
However, this does not have a significant effect on sound image localization.
Is removed by a BPF (Band Pass Filter).
You. By limiting the band in this way, unnecessary
Peaks and dips have been removed and applied to the cancel filter
Since unnecessary coefficients do not occur, convergence is improved, and
The number can be shortened. Then, the band-limited IR (S) is converted into the inverse F
FT conversion and cut IR (impulse response) on time axis
Window (for example, a cosine function window)
(The second IR). Wind processing
As a result, the effective length of IR will no longer be long and will be canceled
The convergence of the filter is improved so that the sound quality does not deteriorate
Swell. FIG. 9 shows the IR (impulse response) of the HRTF.
A specific example will be described. The horizontal axis is time (sample clock is 4
8kHz clock time), vertical axis is amplitude level
It is. The two-dot chain line indicates the window. Cancel filters cfLx, cfRx
Calculation convolver (convolution operation processing circuit)
Ruta cfLx and cfRx are as described above (Equation 4a) and
It takes the value shown in (Equation 4b). Here, the speakers sp1 and sp
2 transfer characteristics h1L, h1R, h2L, h2R
And an actual speaker is placed at the intended localization position x.
Head transfer characteristics pLx and pRx when
The shaping process for each angle θ obtained by
The obtained second IR (impulse response) is substituted. The head transfer characteristics h1L and h1R are represented by L in FIG.
It corresponds to the position of the channel speaker, from the front
For example, if it is set at 30 degrees to the left (θ = 330 degrees)
For example, an IR of θ = 330 degrees is used. Head transfer characteristics h2
R and h2L correspond to the position of the R channel speaker in FIG.
For example, 30 degrees from the front to the right (θ = 30
Degrees), use an IR of θ = 30 degrees
(That is, the system at the time of actual sound image reproduction (for example, FIG. 4
). Then, the head transfer characteristics pLx and pRx are defined as
90 degrees from the front, which is the target sound source localization position.
The range of 180 degrees is of course
IR for every 30 degrees in a wide space (all spaces)
By entering, cfLx of the whole space corresponding to it,
cfRx, that is, 12 sets of cancel buffers every 30 degrees
Filters cfLx and cfRx are obtained (in FIG. 8,
An example is a position at 240 degrees). Cancel filter
The cfLx and cfRx groups finally show the response on the time axis.
Is obtained as IR (impulse response). It should be noted that the cancel fill according to (Equation 4a)
The calculation of the parameters cfLx and cfRx is as follows. First
H which is a kind of inverse filter for H in (Equation 4b) ^-1 To
It is obtained by the least-squares method, and this is inverse-FFT transformed to obtain
Let h (t) be the number. Further, each term h1L, h1 of (Equation 4a)
R, h2L, pRx, pLx, h2R
The following formula is established by expressing by a number. cfLx (t) = (h2R · pLx−h2L · pRx) ·
h (t) cfRx (t) = (− h1R · pLx + h1L · pRx)
H (t) Here, both equations are combined to obtain (Equation 7). Therefore, from these (Equation 7),
The coefficients of the cell filters cfLx and cfRx are determined.
And As is apparent from (Equation 7),
To shorten the filter cfLx and cfRx coefficients,
Transfer characteristics h1L, h1R, h2L, pRx, pLx,
It is very important to shorten h2R. This
Therefore, as described above, window processing and shaping
Various processing such as processing are performed, and each head transfer characteristic h1L,
h1R, h2L, pRx, pLx, h2R are shortened.
You. FIG. 10 shows the coefficient cfL of the cancel filter.
A specific coefficient sequence of x and cfRx is shown. The horizontal axis is time (sa
Clock time when the sample clock is 48kHz),
The vertical axis is the amplitude level. Two-dot chain line indicates window
You. Note that the coefficients cfLx and cfR of the cancel filter are used.
x is subjected to FFT to obtain a frequency response,
Moving average, inverse FFT transform and final
The time response of the cell filter may be obtained. Move like this
Averaging eliminates unnecessary peaks and dips.
Time response to be achieved earlier
Thus, the size of the cancel filter can be reduced. Cancellation filter for each localization point x
Scaling Also, the actual sound image with a convolver (cancellation filter)
The spectrum distribution of the processed sound source (source sound)
There is something distributed like pink noise
Or something that goes down gently in the high range.
Even though the sound source is different from a single sound, convolution operation
Overflow occurs when (integration) is performed, causing distortion
Danger. So, to prevent overflow
Between the coefficients of the cancellation filters cfLx and cfRx
At the maximum gain (for example, cancel filter cfL
x, cfRx)
When the coefficient and 0db white noise are convolved
Scale all coefficients to avoid overflow.
Ring. In practice, the maximum absolute value is
(Amplitude) 1 to 0.1 to 0.4 (for example, 0.1 to 0.4).
It is good to attenuate so that it becomes 2). Then, the window shown in FIG.
In window), according to the actual number of convolver coefficients
Window processing so that both ends become 0, and
Shorten the effect length. Is scaled in this way.
Data that is finally supplied to the convolver as coefficients
Group (in this example, 12 sets capable of sound image localization every 30 degrees
Of the convolver) cfLx and cfRx are obtained. FIG. 4 shows an example of a sound reproducing device of a game machine, which is obtained by convolving and reproducing a signal from a sound source.
In this way, the game operator (listener) M
A pair of speakers sp1 and sp2 are arranged at intervals
The pair of speakers sp1 and sp2 have a pair of speakers sp1 and sp2.
Acoustic signal processed by convolver (convolution processing circuit)
Is configured to be reproduced. In a pair of convolvers,
The same sound source X (for example, a flight from a game synthesizer)
Signal, etc.), and
Coefficients cfLx and cfRx created in step 105
(For example, the flight sound is 120 degrees left rear (θ = 240 degrees)
(If you want to localize the sound image at the position, θ = 240 degree coefficient)
Is selected and set in the convolver. For example, game
Sound image measurement from main CPU (central processing unit)
The sub CPU for control uses the coefficient R
From the OM, the coefficient of the desired localization position is applied to a pair of convolvers.
Forward. In this way, the pair of convolvers
The signal from sound source X is not convoluted on the time axis
And a pair of speakers sp1 and sp
Reproduced from 2. From a pair of speakers sp1 and sp2
The reproduced sound is transmitted from the speakers sp1 and sp2 to both ears.
HRTF canceled, binaural from any desired position
HRTF is reproduced to the desired position
The sound image is localized as if there is a source, heard by the listener M,
It is reproduced as a sound full of reality. Convolver
The number is based on the movement of the airplane in response to the operation of the listener M.
Then, the optimal sound image position is sequentially selected and switched. Ma
When changing from a flight sound to a missile sound, for example,
Source sound from source X is changed from flight sound to missile sound
You. In this way, the sound image can be freely localized at any position.
Can be made. As a transducer for reproduction,
Headphone instead of the pair of speakers sp1 and sp2
Can also be used. In this case, the HRTF measurement
Since the constant conditions are different, prepare the coefficients separately and
It is better to switch between them. In addition, IR (impulse response) shaping processing
Is not always necessary, and even if omitted, control of sound image localization
Is possible. In the real sound image localization control,
The same sound source is supplied from a pair of transducers
A structure for reproducing a signal processed by a pair of supplied convolvers
Is the minimum configuration to obtain the effect of the present application.
It is. Therefore, if necessary, a pair, ie, 2
Additional configuration with one or more transducers and convolvers
Of course, you can also
If the coefficient is long, split the coefficient to
It may be constituted by a bus. Further, the coefficient of the convolver is
The opening angle of the mosquito (that is, the angle sp1-M-sp in FIG. 4)
2) It depends on the opening angle of the speaker.
Can be used selectively according to the actual playback system.
You may do it. That is, in the above, the game
Separate left and right by 30 degrees around the operator (listener)
A pair of speakers sp1 and sp2 with an opening angle of 60 degrees
The convolver's coefficient for installation was determined.
When calculating the cancellation filter, the speaker sp
1, sp2, head transfer characteristics h1L, h1R, h2
As L and h2R, other opening angles, for example, 45 degrees and 30 degrees
What is necessary is just to substitute the one corresponding to and find it. The convolver coefficient is measured by HRTF.
It may vary depending on the fixed conditions.
No. That is, since there is an individual difference in the size of the head,
The size of the dummy head (or human head) when measuring RTF
Change, and find several types, depending on the audience (for example,
(For adults with big heads and for children with small heads)
You may make it possible to use it. Further, positions not actually measured, for example, θ =
If you want to localize the sound image more precisely every 15 degrees
Is based on the measured coefficients cfLx and cfRx.
As described above, the intermediate value may be calculated. This
, Two measured values (for example, an intermediate value at θ = 15 degrees)
Is obtained by simply measuring θ = 0 degrees and θ = 30 degrees)
Convolver based on actual measurement instead of arithmetic averaging
Frequency response by transforming the coefficients cfLx and cfRx of
, And then refer to the frequency-amplitude characteristic as the amplitude of the transfer characteristic.
Calculated as the geometric mean of the characteristics.
Phase component of the vector average of the frequency complex vector
It is good to ask. Find the intermediate value in this way
And an intermediate transfer characteristic that closely approximates the measured value.
There is almost no deterioration in feeling or sound quality. The measurement of HRTF in the above-mentioned case
At the time of measurement, only the semicircular portion from θ = 0 to 180 degrees was actually measured
For the remaining semicircle, use the symmetry of the coefficient.
Then, the measured value may be used. In this way, the measurement
And the calculation of IR and coefficient
Unnecessary fineness and a coefficient excellent in sound image localization
May be. As described in detail above, the sound image localization system described above
According to the technology, the signal from the sound source is
The signal is processed on the time axis to localize the sound image
As a circuit for actually performing the sound image processing,
You need one set of convolvers on the time axis as shown
Therefore, the circuit scale is very small and inexpensive. Also, the sound image localization processing of the convolver
Numerical data is finally converted to IR (impulse response) on the time axis.
A) Because it is supplied as data,
To reduce the scale, reduce the number of convolver coefficients.
It would be good to lose it. Further, the convolver is provided with an IR (impulse response).
A) data is supplied as a coefficient and processed.
Therefore, the optimal solution in the time domain can be easily calculated using the coefficient as IR.
Moreover, it is not adaptive and can be determined uniquely.
Since the delay time of the time axis response waveform can be clearly specified, multiple
The time relationship between the response waveforms of the points can be accurately controlled. Ma
In addition, the convolver coefficient is set to the phase for each frequency based on the actual measurement.
And the amplitude can be accurately measured,
It is possible to localize the sound image in a wide space
Was. The IR shaping process described above,
That is, the first IR corresponding to the measured HRTF is obtained.
From the first IR over the audio spectrum
Perform predetermined processing (band limitation) in the scattered frequency band,
Window on time and time axis (for example, cosine function window)
Multiplied by window processing to converge the length to a predetermined value
A second IR is obtained, and a pair of cancels are obtained from the second IR.
If the coefficients of the filter are determined,
In addition, the accuracy of the calculation in the cancel filter calculation process
Is improved. In the above description, white noise
HRTF at each sound image localization using
, The S / N ratio can be increased by using white noise.
HRTF (IR,
Number). Also the smell mentioned above
Average the multiple IRs determined for each HRTF
That is, as mentioned above, several hundred
Calculate IR (S) for each window and average them
In this case, the SN ratio is improved and the accuracy is improved. As described above, the magnitude of the coefficient is
Scale the maximum value to be 0.1 to 0.4 of the maximum level
If it is attenuated by processing, overflow and playback
There is no distortion in the sound. The above-described sound image localization control technique has a
Image at a position different from the position of the speakers sp1 and sp2
To localize the coefficient data of a pair of convolvers (c
fLx, cfRx) to a pair of convolvers 2
The horizontal angle of 360 degrees about the listener M
Control and distance control from the listener M
And the elevation angle from just beside the listener M to just above
No controls are included. For this reason, the present inventor has proposed this sound image localization control technique.
We studied to incorporate elevation control into the surgery. As a result,
Convolver coefficient data for horizontal angle and distance control
(CfLx, cfRx)
The sound source to be reproduced by listener M
(Source) determines the probability of realization of sound image localization in the elevation direction,
The accuracy of the position is large and
However, if the sound image is stationary on the rear side of the listener M,
It was noted that the sound image is likely to be flipped back and forth when the sound is heard. That is, in the elevation direction in the sound image localization control,
To perform sound image localization control optimally, for each listener M,
Fine sound image localization correction is required for each playback source.
It is important. Therefore, correction based on mapping described later is performed.
By using the data, it is possible to change
You can do good, every listener M, you play source
At the same time, optimal sound image localization can be performed. (1) A method for correcting a shift in auditory localization recognition
Mapping of the position filter The present inventor conducted an experiment of localization of a spatial sound image by a real sound. this
As a result, the position where the listener M recognizes the sound image is not necessarily provided.
Position does not match, the presentation position and the localization recognition position
The deviation from the position depends on the presentation position regardless of the listener M,
(A) and (b) show almost the same tendency
Was confirmed. (A) The deviation of the horizontal angle is entirely before the listener M.
This is particularly likely to occur in the direction
Is a strong tendency. That is, as shown in FIG.
At a certain distance from listener M with listener M at the center
Presentation position a (mapped to the listener M)
(Directly in front of), b, c, d (directly behind listener M)
You. After this, the localization recognition in which the listener M actually recognizes the sound image
Map the location. As a result, the display positions a and d
Therefore, no deviation from the localization recognition position occurs. On the other hand, presentation
For the position c, a deviation from the localization recognition position occurs. Present
Whether the display position c is the same as the display position b
The following positional deviation occurs. In particular, listener M
Between the presentation position and the localization recognition position from right behind to the rear
Occurs. The deviation between the presenting position and the localization recognition position
Actively use, in this example, if you show an example,
When the sound image is localized at the presentation position b,
The corresponding pair of convolver coefficient data (cfLx, c
fRx) only needs to be corrected and set.
Position can be obtained optimally. (B) The deviation of the elevation angle is entirely above the listener M
This tendency tends to occur especially in the area immediately behind the listener M.
Is strong. That is, as shown in FIG.
From the back of the listener M to the crown, with the listener M at the center
The presentation position mapped to the elevation position that is separated by a certain distance
e (directly behind listener M), f, g, h (head of listener M)
Top) is set. After this, the listener M actually recognizes the sound image
The localization recognition position to be mapped is mapped. As a result, the presentation position
For the positions e and h, no deviation from the localization recognition position occurs.
No. On the other hand, the deviation from the localization recognition position for the presentation position f
Occurs. For the presentation position f, it is as if the presentation position g
, A localization recognition position shift occurs as if it were the same.
In particular, the presentation position right behind and above the back of listener M excluding the top of the head
The position and the localization recognition position are shifted. The deviation between the presented position and the localization recognition position
Actively use, in this example, if you show an example,
When the sound image is localized at the presentation position g,
The corresponding pair of convolver coefficient data (cfLx, c
fRx) only needs to be corrected and set.
Position can be obtained optimally. By the way, the degree differs depending on the listener M.
However, sound that has undergone sound localization can be easily localized in motion.
However, when the sound image is stationary on the rear surface,
Inversion may occur easily. Make the sound image stand still
If you place it, you can localize it by shaking the localization point little by little.
To determine if it is possible to clarify
Was. As a result, regarding the swing width, the swing cycle is
When it was short, the playback source tremble was felt. That is, FIG.
(A) As shown in FIG.
And oscillated the localization point in small increments at 1 second intervals
(Angle -θ → 0 → + θ → −θ 1 second pause -θ → 0 → + θ
→ -θ ...). By the way, the angle + θ is the position diagonally right before the localization point.
Required angle, angle -θ is the required angle diagonally behind and left of the localization point.
You. In addition, when the shaking cycle is long,
That was felt. That is, as shown in FIG.
In addition, within the range of angle + θ to angle -θ, + θ (1 second)
Interval) → 0 (for 2 seconds) → -θ (for 1 second) → 0 (for 2 seconds) →
... caused the localization point to fluctuate in a long cycle. Now, the correction data based on the above-described mapping
Data to improve rear localization and vertical plane localization
The sound image localization control device for performing
An audio signal obtained by convolving an audio signal output from a sound source (X)
Is reproduced from the speakers sp1 and sp2 and the speakers s
The sound image is localized at a position different from the arrangement of p1 and sp2
Device 1 that performs sound image localization, and outputs from a sound source (X)
A pair of convolvers 2 to which signals to be supplied are supplied, and a speaker
A sound image is localized at a position different from the arrangement of sp1 and sp2.
Data cfLx and cfRx for convolver 2
The supplied coefficient memory 3 and a listening position (a listener (for example,
The coefficient data localized in the sound image behind the game operator) M position)
Data localized by the data cfLx and cfRx
And a conversion table 4. Same structure as above
The same components are denoted by the same reference numerals and description thereof is omitted. Thus, the listening position (the position of the listener M)
If the optimal sound image localization cannot be obtained in
Cannot obtain the optimal sound image localization using an input device (not shown).
Specify and input the signal corresponding to the position. This input state
When the main CPU recognizes, the sub CPU for control
Is supplied from the coefficient memory 3 to the convolver 2
Coefficient data cfLx, cfRx corresponding to the position
Instead of the coefficient data cfLx, c corresponding to the position
fRx (correction data) is converted from the conversion table 4 into a coefficient memory.
The control signal for switching and outputting to the coefficient memory 3 and
Output to the conversion table 4. These correction data are described above.
Of the localization filter that corrects the displacement of
This correction data is converted
It is stored in advance in Table 4. Thus, the optimum sound image definition at the listening position
If a specific position for which rank cannot be obtained occurs, the specific position
By switching and outputting correction data for
The listener can improve rear and vertical orientation.
Performs optimal sound image localization for each M and for each source to be reproduced
be able to. For example, the listening is performed based on the above-described deviation of the horizontal angle.
When the person M cannot optimally obtain the backward localization, FIG.
As shown in (A), at the angular position corresponding to the presentation position b
There is no sound image localization recognition position so that the sound image at the presentation position c is localized.
This occurs. A pair of combos output from coefficient ROM3
Luba coefficient data (cfLx, cfRx) is the localization target point
Is not the coefficient data obtained from the HRTF to both ears,
Considering the misalignment of localization recognition, if you feel that you are localized there
(Ie, localized at the presentation position b)
The coefficient is a coefficient created from the HRTF from the presentation position c.
And). Therefore, the output from the coefficient ROM 3 is
The coefficient data (cfLx, cfRx) of the pair of convolvers is
When indicating the presentation position b, the presentation position b is used instead.
Change the correction data for outputting the coefficient data at the indicated position c.
What is necessary is just to output from the conversion table 4 to the coefficient memory 3.
No. Reading of the correction data from the conversion table 4
This is performed under the control of the troll sub CPU. Similarly, the listener is determined by the above-mentioned elevation angle deviation.
FIG. 2 shows a case where M cannot optimally obtain the backward localization.
As shown in (B), at the angular position corresponding to the presentation position g
There is no sound image localization recognition position so that the sound image at the presentation position f is localized.
This occurs. A pair of combos output from coefficient ROM3
Luba coefficient data (cfLx, cfRx) is the localization target point
Is not the coefficient data obtained from the HRTF to both ears,
Considering the misalignment of localization recognition, if you feel that you are localized there
(Ie, localized at the presentation position g)
The coefficient is a coefficient created from the HRTF from the presentation position f.
And). Therefore, the output from the coefficient ROM 3 is
The coefficient data (cfLx, cfRx) of the pair of convolvers is
When it indicates the presentation position g, the presentation position g
Change the correction data for outputting the coefficient data at the indicated position f.
What is necessary is just to output from the conversion table 4 to the coefficient memory 3.
No. Further, the sound image is reproduced by reversing the sound image.
FIG. 3 shows a case where the taker M cannot optimally obtain the rearward orientation.
As shown in (A), the period of shaking the sound image is fluctuated little by little.
(Angle -θ → 0 → + θ → −θ 1 second pause -θ → 0 → + θ
→ −θ ...), the stationary sound image is
It can be recognized stably and reliably. That is, the coefficient R
OM3 has a pair of capacitors corresponding to -θ, 0, + θ in advance.
Storing the VOLBA coefficient data (cfLx, cfRx)
deep. As an example, the coefficient data of the convolver at the presentation position e
Data (cfLx, cfRx) is -θ (for example,
Presentation position at an angle of about -10 degrees from the center 0)
coefficient data of the presentation position e at the center 0
Data, + θ (for example, an angle of about +10 degrees from the center 0)
The coefficient data of the presentation position e at the position is stored respectively.
Keep it. Then, at the listening position (the position of the listener M),
When stable localization of a stationary sound image cannot be obtained
Means that the listener M has a stable sound image using an input device (not shown).
Specify and input the signal corresponding to the position where localization cannot be obtained.
When the main CPU recognizes this input state,
For the sub CPU for the
Pair of combos corresponding to the obtained coefficient data -θ, 0, + θ
Luba coefficient data pauses for -θ → 0 → + θ → −θ1 second
-Θ → 0 → + θ → −θ, etc., sequentially output from coefficient ROM3
A control signal to the effect is output to the coefficient ROM 3. In this way, the sound corresponding to the specific position
Since the image can be shaken periodically in small increments,
At the position (position of the listener M),
It can be recognized stably and reliably without inversion. Similarly, the sound image is reproduced by reversing the sound image.
FIG. 3 shows a case where the taker M cannot optimally obtain the rearward orientation.
(B) As shown in FIG.
(+ Θ (1 second) → 0 (2 seconds) → −θ (1 second) → 0
(2 seconds) →…)
It can be recognized stably and reliably without inversion. That is,
The coefficient ROM 3 previously stores a pair corresponding to -θ, 0, + θ.
Of the convolver's coefficient data (cfLx, cfRx)
Keep it. As an example, the convolver at the presentation position e
As numerical data (cfLx, cfRx), -θ (example
(For example, an angle position of about −10 degrees from the center 0)
Coefficient data of the indicated position e, of the indicated position e at the center 0
Coefficient data, + θ (for example, about +10 degrees from the center 0)
The coefficient data of the presentation position e at the angle position
And store it. Then, at the listening position (the position of the listener M),
When stable localization of a stationary sound image cannot be obtained
Means that the listener M has a stable sound image using an input device (not shown).
Specify and input the signal corresponding to the position where localization cannot be obtained.
When the main CPU recognizes this input state,
For the sub CPU for the
Pair of combos corresponding to the obtained coefficient data -θ, 0, + θ
Luba coefficient data is + θ (1 second) → 0 (2 seconds) →
-Θ (1 second) → 0 (2 seconds) → ...
From the coefficient ROM3.
Power. Thus, the sound corresponding to the specific position
Since the image can be shaken periodically and long, the listening position
(The position of the listener M)
It can be recognized stably and reliably without inversion. The above sound image localization recognition position shift or sound
In the environment of the listening position where external noise is large,
It depends greatly on the individual differences of the listeners or the music source, etc.
Because it is different, the sound image localization is recognized in advance by the test pattern.
Detecting displacement or reversal of the sound image
Conversion table for correction data as described above based on detection
4 so that the above correction can be performed efficiently.
It can be carried out. As described in detail above, the sound image according to the present invention
According to the localization control device, the sound source is controlled by a pair of convolvers.
Processing these signals on the time axis to localize the sound image
Therefore, as a circuit that actually performs sound image processing, the time axis
Only the convolutional processing circuit above is required, and the circuit
The scale is very small and inexpensive. In addition, the
The coefficient data of the sound image localization processing of Volva
Supply as IR (impulse response) data on axis
In order to make the circuit size smaller,
It suffices to reduce the number of convolver coefficients.
As a result, compared to the conventional approximation of data on the frequency axis,
HRTF can be accurately and effectively approximated.
From this point, the circuit scale without impairing the sound image localization feeling
Can be reduced. In addition, a ring with much external noise
Listening position below the border, listening with a difference greater than the standard distance between both ears
Good for uniform sound image localization control, such as when the listener listens
Especially when the listening position is not
By correcting the sound image localization position,
An appropriate sound image localization feeling can be obtained.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a configuration diagram of an embodiment of a sound image localization control device according to the present invention. FIG. 2 is a diagram illustrating mapping for correcting a shift in sound image localization recognition. FIG. 3 is a diagram illustrating how a sound image localization point fluctuates. FIG. 4 is a configuration diagram of a general sound image localization device. FIG. 5 is a configuration diagram showing a basic principle of sound image localization control. FIG. 6 is a configuration diagram showing an HRTF (head-related transfer function) measurement system. FIG. 7 is a diagram illustrating points of HRTF measurement. FIG. 8 is a diagram illustrating a calculation example of a cancel filter. FIG. 9 is a diagram showing a specific example of IR (impulse response) of HRTF. FIG. 10 is a diagram illustrating a specific example of coefficients of a cancel filter. [Description of Signs] 1 sound image localization control device 2 convolution operation processing circuit, convolver 3 coefficient ROM (coefficient memory) 4 conversion table (correction means) cfLx, cfRx cancel filter (convolver) and its coefficient M listeners sp1, sp2 speakers (Transducer) X sound source

Claims (1)

(57) [Claim 1] A sound signal obtained by convolving a signal output from a sound source with a pair of transducers is reproduced from a pair of transducers, and a sound image is localized at a position different from the arrangement of the pair of transducers. A sound image localization control device that performs sound image localization as described above, comprising: a pair of convolution operation processing circuits to which signals output from the same sound source are supplied; and a sound image localization control device for localizing the sound image at a position different from the arrangement of the pair of transducers. A coefficient memory for supplying a sound image localization position coefficient to the convolution operation processing circuit, wherein the sound image is localized at the listening position by using a sound image localization position coefficient for periodically changing an angular position. A sound image localization control device for ensuring recognition of a sound image.