JP2000059894A - Sound image localization device and processing method for fir filter coefficient and arithmetic method for the fir filter - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a sound image localizing device where a FIR filter with many taps regardless of a small capacity of a coefficient memory is configured and a rewrite time of the coefficient memory is reduced in the case of sound localization processing, a processing method for the FIR filter coefficients where localization with excellent tracking performance is realized and an arithmetic method of the FIR filter. SOLUTION: The localization device is provided with a separate means 2 that separates a localization response impulse train into two or over, a compression processing means 3 that applies scaling to the separated impulse trains for compression processing, a rearrangement means 4 that rearranges a scaling coefficient and a compressed localization coefficient into reduced word number according to a prescribed format, a coefficient memory 6 that reads and stores the rearranged scaling coefficient, a coefficient separate means 7 that separates data stored in the coefficient memory into two localization coefficients or over, two FIR filters 8 or over that apply product- sum operation between the localization coefficient and an input signal, and a re-scaling section 9 that applies re-scaling to the result of product-sum of the FIR filters respectively.

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 apparatus and a FIR filter coefficient when a localization filter is implemented by an FIR filter in sound image localization techniques such as Dolby surround two-speaker reproduction and out-of-head sound image localization using headphones. The present invention relates to a processing method and an FIR filter calculation method.

[0002]

2. Description of the Related Art Sound image localization using a speaker or headphones is a method of localizing a sound image as if the sound source is located in a place different from the actual sound source. Head-related transfer functions (HRTFs) measured by a real head or a dummy head are known. In the case of speaker playback, crosstalk cancellation is performed for speaker playback, and headphone characteristics cancellation and crosstalk addition processing are performed for headphones. (This is called a localization filter).

[0003] There are various types of configurations of the localization filter, but those using an FIR filter have the best localization and natural sound quality. In other words, it has excellent features such that the sound image does not rise much and the sound image is less reversed back and forth. This is because the FIR filter coefficients have a head related transfer function (HR
TF) and can reproduce the desired sound pressure on the eardrum most faithfully.

[0004]

The number of the FIR filter coefficients is called a tap length, and the longer this is, the better the localization becomes. However, there is a limit to the tap length that can be used due to the limitation of the arithmetic capacity and memory capacity of a DSP (Digital Signal Processor).

FIG. 12 is a characteristic diagram showing an example of a head related transfer function (HRTF) in an acoustic room having an appropriate reflection. From FIG. 12, it can be said that in order to realize the HRTF in the sounding room with the FIR filter, the number of taps of several hundred units is required. However, the currently commonly used DS
The number of taps that can be realized by P is about 256 to 512.
Considering that four localization filters are required for localization of normal 2-channel audio, the number of taps allocated to one localization filter is at most about 100 taps. With only about 100 taps, only the direct sound part of the HRTF can be realized, so that good localization cannot be expected.

FIG. 13 is a conceptual diagram in the case where sound image localization is applied to an interactive device such as a game. In such a device, the localization position is changed by returning the position information obtained from the joystick or the keyboard of the computer to the left and right localization filters.

FIG. 14 shows a system in which the rotation angle of the head is detected by a geomagnetic azimuth sensor and the localization coefficient is changed based on the angle information obtained therefrom. In such a system, since the position of the sound image with respect to the surroundings does not change, a more realistic localization can be obtained.

However, FIG. 13 and FIG.
In such a system as shown in (1), the localization coefficient is frequently changed. Therefore, the coefficient must be rewritten instantaneously. If the localization filter of such a system is realized by an FIR filter, there is a problem that rewriting cannot be performed in time because there are many coefficients to be rewritten.

SUMMARY OF THE INVENTION The present invention has been made in view of the above points, and in the sound image localization processing for an acoustic signal, a FIR filter having a large number of taps can be configured with a small coefficient memory, and a natural and undistorted sound image localization is possible. It is an object of the present invention to provide a sound image localization device which can be realized with less hardware and can reduce the rewriting time of the coefficient memory. It is another object of the present invention to provide a method of processing FIR filter coefficients and a method of calculating an FIR filter, which can realize very good localization when the localization system has an interactive configuration.

[0010]

According to the present invention, there is provided a sound image localization apparatus comprising: separating means for separating a localization response impulse train into two parts, a first half and a second half; a first half and a second half separated by the separating means. First and second half compression processing means for scaling and compressing the impulse train signal so as to be within a predetermined number of bits, and a scaling coefficient and compression used in a compression process by the first and second half compression processing means. Relocation means for relocating the determined localization coefficient in accordance with a predetermined format; a coefficient memory for reading and storing the scaling coefficient and the compressed localization coefficient rearranged by the relocation means; and data stored in the coefficient memory. Separating means into a localization coefficient of the first half and a localization coefficient of the second half, a first half and a second half FIR for performing a product-sum operation of the localization coefficient and the input signal And a filter, said first half portion and the rear half portion FI
Rescaling means for rescaling the product-sum results of the R filter.

According to another aspect of the present invention, there is provided a sound image localization apparatus, comprising: separating means for dividing a localization response impulse train into N blocks; Compression processing means for scaling and compressing to fit within the number, rearrangement means for rearranging the scaling coefficients and the compressed localization coefficients used in the compression processing by the compression processing means according to a predetermined format, A coefficient memory for reading and storing the scaled coefficients and the compressed localization coefficients rearranged by the rearrangement means, a coefficient separation means for separating the data stored in the coefficient memory into localization coefficients for each block, N FIR filters for performing a product-sum operation on the input signal and the input signal, and a product-sum result of the N FIR filters It is obtained by a rescaling unit to re-ring.

Further, in the method of processing FIR filter coefficients according to the present invention, a separating step of separating a localization response impulse train into a first half and a second half, and a method of separating a signal of the separated impulse train into a predetermined number of bits. It comprises a compression processing step of performing scaling and compression processing, and a rearrangement step of rearranging the scaling coefficients and the compressed localization coefficients used in the compression processing in the compression processing step according to a predetermined format.

[0013] Further, in the calculation method of the FIR filter according to the present invention, a separation step of separating the localization response impulse train into a first half and a second half, and scaling the separated impulse train signal so as to be within a predetermined number of bits. A compression processing step of performing compression processing, a rearrangement step of rearranging a scaling coefficient and a compressed localization coefficient used in the compression processing in the compression processing step according to a predetermined format, and rearrangement by the rearrangement step. A storage step of storing a scaling coefficient and a compressed localization coefficient in an external ROM; reading the scaling coefficient and the compressed localization coefficient from the external ROM into a coefficient memory inside a DSP; The data in the coefficient memory is A coefficient separation step of separating position to the coefficients, FIR first half portion and second half portion for performing product-sum operation of the coefficient separating the localization coefficient and the input signal separated by the step
It comprises a filter calculation step and a re-scaling step of re-scaling the product-sum results of the respective FIR filter calculation steps.

According to another aspect of the present invention, there is provided a method for processing FIR filter coefficients, comprising the steps of: separating a localized response impulse train into N blocks; and scaling the separated impulse train signal so as to be within a predetermined number of bits. And a rearrangement step of rearranging the scaling coefficients and the compressed localization coefficients used in the compression processing in the compression processing step according to a predetermined format.

According to another aspect of the present invention, there is provided a method for calculating an FIR filter, comprising the steps of: separating a localized response impulse train into N blocks; and scaling the separated impulse train signal so as to be within a predetermined number of bits. A compression processing step of performing compression processing, a rearrangement step of rearranging a scaling coefficient and a compressed localization coefficient used in the compression processing process of the compression processing step in accordance with a predetermined format, and a rearrangement step performed by the rearrangement step. A storage step of storing a scaling coefficient and a compressed localization coefficient in an external ROM; reading the scaling coefficient and the compressed localization coefficient from the external ROM into a coefficient memory inside a DSP; The data stored in the coefficient memory is A coefficient separation step of separating the signals into local coefficients, a product-sum operation of the localization coefficients of each block separated by the coefficient separation step and an input signal to calculate an FIR filter of each block, and the FIR filter calculating step. And a re-scaling step of re-scaling the product-sum results in the filter operation step.

Furthermore, a sound image localization apparatus according to another invention is characterized in that:
In a sound image localization device having front, left, right, rear left, and rear right audio signal inputs of five channels, and each of the five channel inputs being sound image localized at five positions by ten localization filters, a headphone-mounted receiver is provided. A rotation angle detection means for detecting a rotation angle of the listener's head, and a speaker for rewriting the coefficients of the respective localization filters in accordance with the rotation angle detected by the rotation angle detection means and moving the localization position of the sound image. Means for realizing a sound field equivalent to the sound field of (a), storing the localization coefficients of the localization filter in a compressed form in an internal memory of the DSP, and performing a product-sum operation while decompressing the compressed coefficients; It is characterized by having.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS At present, the quality of digital audio generally used is 16 bits. On the other hand, 24-32 bits are mainly used for the coefficient memory of a normal DSP. Further, the IIR filter has a feedback loop and is therefore sensitive to the accuracy of the coefficients, whereas the FIR filter is a simple sum of products and does not require strict accuracy.

In consideration of such points, by assigning a plurality of localization coefficients to one of the coefficient memories of the DSP,
The coefficient memory can be saved and rewriting can be speeded up.
For example, if two localization coefficients are assigned to one coefficient memory to realize a 256-tap FIR filter, only 128 coefficient memories are needed, and the number of rewrites at the time of changing the localization coefficients is only 128.

Further, if bits are simply assigned with a reduced number of bits, the accuracy is degraded, or the sound quality is degraded. Therefore, by using data compression at the time of assignment, it is possible to prevent deterioration in accuracy. For data compression, the number of bits can be reduced by utilizing the psychological properties of human auditory psychology, and furthermore, deterioration in accuracy due to scaling can be prevented.

FIG. 1 is an explanatory diagram for explaining the principle of coefficient assignment according to the present invention. To make the description concrete,
The coefficient memory of the DSP is 24 bits, and the localization coefficient is 2
A description will be given on the assumption that 256 localization coefficients are allocated to 128 coefficient memories C0 to C127.

First, referring to FIG. 12, which is a time response of the head related transfer function (HRTF), it can be seen that a portion having a large amplitude is limited to a time corresponding to a direct sound. Also, from the auditory psychology of humans, the impression of a sound image is almost determined by the direct sound which is the preceding sound.

Taking these points into consideration, it can be seen that accuracy is required in the first half of the localization filter and not required in the second half. Therefore, the 24-bit memory is stored in the upper 16
Bits and lower 8 bits are divided, and the upper half of the localization filter is assigned to the upper bits and the lower half of the localization filter is assigned to the lower bits.

However, it is not wise to assign the first half and the second half on the same scale. Therefore, the first half of the localization filter has a full scale (3276) whose maximum value is 16 bits.
8) The second half of the localization filter is scaled to be 8-bit full scale (128), and then DS
Converted to the coefficient of P. Then, the scaling coefficient at this time is separately stored.

FIG. 2 shows this scaling. This scaling coefficient is used for rescaling to match the results of the preceding and following product-sum operations.

FIG. 1 shows a case where the localization coefficient is allocated to two blocks before and after the allocation. However, FIG. 3 shows that the number of blocks is further increased to N blocks (block 0 to block N-1). is there. Also, a case is shown in which three or more localization coefficients are assigned to DSP coefficients using a compression algorithm such as ADPCM instead of compression by simple scaling.

Hereinafter, a sound image localization processing apparatus and a processing method according to the present invention will be described with reference to the accompanying drawings. <First Embodiment> FIG. 4 is a configuration diagram showing a first embodiment of the present invention. In the figure, reference numeral 1 denotes a desired impulse response, which is an HRTF corresponding to a direction to be localized.
Corresponds to this. Reference numeral 2 denotes a response separation unit, and 3.0 and 3.1 denote a first half compression processing unit and a second half which scale and compress the signals of the first half and the second half separated by the response separation unit 2 so as to be within a predetermined bit number. The partial compression processing section 4 is a coefficient rearrangement for rearranging the scaling coefficients and compressed localization coefficients from the first half compression processing section and the second half compression processing sections 3.0 and 3.1 to the number of words reduced according to a predetermined format. Arrangement part, 5
Is an external ROM.

Reference numeral 6 denotes a coefficient memory inside the DSP for storing filter coefficients given via the external memory 5;
Is a coefficient separating means for separating the data of the coefficient memory 6 into the first half and the second half of the localization coefficient. Part FIR filter H0 and latter half FIR filter H
1, 9.0 and 9.1 are rescaling units, 10 is an input signal, and 11 is an output signal. In the drawing, a portion surrounded by a dotted line shows an internal configuration of the DSP.

The input signal 10 is supplied to an FIR filter H0, which processes the first half of the desired impulse response 1,
8.0 and the FIR filter H responsible for the latter half of the processing
1 and 8.1, the amplitude is adjusted by the respective scaling coefficients by the re-scaling units 9.0 and 9.1, and the sum is added to the output signal 1
Output as 1.

Here, how the filtering coefficients inside the DSP are set and calculated in the first and second half, respectively, will be described. The desired impulse response 1 is
Divides the data into two blocks, a first half and a second half. The first half is scaled by the compression processing unit 3.0 so that the maximum value becomes 31767, and is passed to the coefficient rearrangement unit 4 together with the scaling coefficient. Similarly, the latter half is scaled by the compression processing unit 3.1 so as to have a maximum value of 127, and is passed to the coefficient rearrangement unit 4 together with the scaling coefficient. The coefficient rearrangement unit 4 uses the coefficients compressed by the first half compression processing unit 3.0 and the second half compression processing unit 3.1.
Rearranged according to the format of the coefficient memory 6 of the SP.

The above operation is performed for all the HRTFs in the desired localization direction,
Save to 5. The compression coefficient stored in the external ROM 5 is
It is called appropriately according to the direction control information of the game machine or the geomagnetic direction sensor, and rewrites the coefficient memory 6 of the DSP. D
Although one coefficient memory stored in the coefficient memory 6 of the SP is 24 bits, in the present invention, two 16-bit and 8-bit localization coefficients are stored in one coefficient memory. This is separated by the coefficient separating means 7 to obtain localization coefficients of the front and rear FIR filters 8.0 and 8.1.

FIG. 5 is a block diagram showing the FIR filters 8.0 and 8.1 in FIG. 4 in detail. As shown in FIG. 5, the FIR filter 8.0 is for performing a product-sum operation of a plurality of delay units D _{00 to} D ₀₁₂₇ for delaying the input signal 10 and the localization coefficients separated by the coefficient separation unit 7. A plurality of multipliers 8.0.0 to 8.0127 and an adder AD8.
0 is provided. Similarly, FIR filter 8.1 has
A plurality of multipliers 8 for performing a product-sum operation of a plurality of delay units D _{10 to} D ₁₁₂₇ for delaying a signal input via the IR filter 8.0 and a localization coefficient separated by the coefficient separation unit 7. .1.0 to 8.1127 and adder AD8.1
And the coefficients separated by the coefficient separating means 7 are used for the actual product-sum operation. On the other hand, the scaling coefficients in the re-scaling units 9.0 and 9.1 are used for re-scaling for matching the former and latter blocks.

FIG. 6A is a flowchart for explaining a method of processing FIR filter coefficients for processing coefficients for efficient FIR filter operation in advance. FIG. 6B shows an FIR in which a DSP performs real-time processing inside the DSP using the coefficient to calculate an FIR filter.
9 is a flowchart illustrating a filter calculation method.
These flowcharts are suitable for providing a computer program for implementing the above method.

As shown in FIG. 6A, in the FIR filter coefficient processing method according to the present embodiment, first, in step S11, the localization response impulse train is separated into a first half and a second half. Thus, the separated impulse train signal is scaled and compressed so as to be within a predetermined number of bits.

After the compression processing in step S12, in step S13, the scaling coefficients and the compressed localization coefficients used in the compression processing are rearranged according to a predetermined format. Through these processing steps, the coefficients of the FIR filter are processed. Further, in step S14, the scaling coefficients and the compressed localization coefficients rearranged by the rearrangement processing in step S13 are stored in the coefficient memory.

The processing inside the DSP will be described below. In step S15a of FIG. 6B, the scaling coefficient and the compressed localization coefficient are read from the coefficient memory into the internal memory of the DSP. Then, in step S15, the data in the coefficient memory are stored in the first half and the second half. In the FIR filter calculation step in step S16, the first half and the second half FIR filter calculations are performed by performing the product sum calculation of the separated localization coefficients and the input signal. The product-sum results from the FIR filter calculation step are rescaled. Through such processing steps, the calculation of the FIR filter is performed.

If the processing in each of the above steps is made to correspond to the processing in the configuration in FIG. 4, the separation processing in step S11 corresponds to the processing in the response separation unit 2, and the compression processing in step S12 is the first half compression. 2. Processing unit and second half compression processing unit
0 and 3.1, the rearrangement processing in step S13 corresponds to the processing in the coefficient rearrangement unit 4, and the processing in step S13 is performed.
The processing in 14 corresponds to the processing in the external ROM 5, the processing in step S15a corresponds to the processing in the coefficient memory 6 inside the DSP, and the processing in step S15 corresponds to the coefficient separating means 7.
In step S16, the processing in the first half F
IR filter H0 and second half FIR filter H1, 8.0
And 8.1, and the process in step S17 corresponds to the rescaling units 9.0 and 9.1.

Here, an unsigning method at the time of packing compressed localization coefficients in the rearrangement processing in step S13 is described in, for example, Japanese Patent Application Laid-Open No. Hei 6-314 proposed by the present applicant.
No. 955 can be applied. That is, the addition output of the adders AD8.0 and AD8.1 is cross-fade by the cross-fade means composed of the multipliers 9.0 and 9.1 shown in FIG. 5, and the switching signal of FIG. Are switched. Also in this case, switching noise is reduced, rewriting is remarkably speeded up, and followability is improved.

As described above, according to the first embodiment, with regard to sound image localization, when realizing a localization filter with an FIR filter, two or more localization coefficients are assigned to one DSP coefficient memory. So the same DSP
Can be assigned twice as many FIR filter coefficients as possible in the coefficient memory capacity, and the accuracy of each coefficient is not required as in the FIR filter, but especially when many coefficients are used. It is valid.

When the localization position is frequently changed, such as in the application of a game sound field, the DSP coefficient must be updated as needed. In the first embodiment, one DSP coefficient is used.
Rewriting the SP coefficient means that the two FIR coefficients have been updated, so the coefficient update time can be reduced by half, which is effective when the localization coefficient is frequently updated in the limited DSP performance. It is.

According to the first embodiment, an FIR filter having a large number of taps with a small number of coefficient memories can be configured in a sound image localization process for an audio signal. Since it can be realized with less hardware and the rewriting time of the coefficient memory can be shortened, when the localization system has an interactive configuration, a very good localization can be realized.

<Second Embodiment> FIG. 9 is a block diagram showing a second embodiment of the present invention. Portions common to those in the first embodiment shown in FIG. 4 are denoted by the same reference numerals, and description thereof will be omitted. Although the basic flow is the same as that of the first embodiment, the number of block divisions is increased from 2 to N in the second embodiment.

In the configuration shown in FIG. 9, a desired impulse response 1 is separated by a response separation unit 2 into N blocks from block 0 to block N-1. The separated signals are subjected to compression processing sections 3.0 to 3.0. After entering into N−1, each block is compressed by a compression algorithm such as ADPCM and sent to the coefficient rearrangement unit 4. At the same time, the scaling coefficient used in the scaling process before compression is also sent to the coefficient rearrangement unit 4.

The coefficient rearrangement section 4 rearranges the compression coefficients according to the format of the coefficient memory of the DSP to be used. Perform the above operations for HRTFs in all directions to be localized, and collect them into an external RO
Save to M5. The compression coefficient stored in the external ROM 5 is appropriately called according to the direction control information of the game machine or the geomagnetic direction sensor, and rewrites the coefficient memory 6 of the DSP.

The compression coefficient read into the coefficient memory 6 is:
Blocks 0 to N-1 separated by the coefficient separating means 7
Are used in the product-sum operation as the localization coefficients of the FIR filters H0 to HN-1. Also, the rescaling coefficients 9.0 to 9. N-1 is used to rescale the output of each block.

Also in the second embodiment, as in the first embodiment, the coefficients for efficient FIR filter calculation according to the flowchart shown in FIG. It is possible to provide a computer program by a processing method and an FIR filter calculation method of calculating an FIR filter using the coefficients.

That is, in the method of calculating the FIR filter coefficients according to the second embodiment, in step S11, a process of separating the localization response impulse train into N blocks is performed, and in step S12, the separated impulse train is processed. After the column signal is scaled and compressed so as to be within the predetermined number of bits, in step S13, the scaling coefficient and the compressed localization coefficient used in the compression processing are rearranged to a reduced number of words according to a predetermined format. Thus, the coefficients of the FIR filter are processed. In step S14, the scaling coefficients and the compressed localization coefficients rearranged by the rearrangement processing in step S13 are stored in the coefficient memory.

The processing inside the DSP will be described below. In step S15a of FIG. 6B, the scaling coefficient and the compressed localization coefficient are read from the coefficient memory into the memory inside the DSP. Then, in step S15, the data in the coefficient memory is separated into the localization coefficient for each block. In the FIR filter calculation step in step S16, the product sum calculation of each of the separated localization coefficients and the input signal is performed to perform the FIR filter calculation for each block. Thereafter, in step S17, the product sum in each FIR filter calculation step is performed. The calculation of the FIR filter is performed by re-scaling the results.

<Third Embodiment> FIG. 10 shows a third embodiment of the present invention in which a 5-channel surround sound field is realized by headphones and the listener's head rotates. This example shows an application example in a system for updating a localization position so that a sound image does not relatively move, that is, in order to reproduce a localization situation similar to speaker listening.

As shown in FIG. 10, the head-related transfer function (HRTF) in a 5-channel surround sound field is represented by H _Ll ,
H _Lr , H _Cl , H _Cr , H _Rl , H _Rr , H _SLl , H _SLr ,
H _SRl and H _SRr . For the 5-channel speaker, LR is 30 degrees left and right, SL · S
It is good to arrange R at 110 degrees left and right.

These head related transfer functions (HRTFs) are determined by measuring the impulse response of the dummy head in a room with some reverberation. However, in the present embodiment, since it is necessary to localize at all positions on the horizontal plane, for example, it is preferable to measure every 5 degrees and 360 degrees.

In this case, (360/5) × 2 (c
h) = 144 head transfer functions (HRTFs) are required, and if each FIR filter has 256 taps, a ROM that holds 144 x 256 = 36864 coefficients is required. Therefore, these coefficients are compressed and stored in the coefficient ROM 48.

FIG. 11 shows the sound image localization of five channels.
Prepare a system like In the drawing, the head-related transfer functions from each direction in FIG. 10 are stored in the coefficient memory of the DSP at 33 to 42 as the coefficients of the FIR filter. The signal of the laser disk or DVD player 31 is AC-3.
The signal is decoded into a 5-channel signal by the decoder 32, input to the localization filters 33 to 42, mixed by the adders 43 and 44, and input to the headphones 45.

A headphone 45 is provided with a terrestrial magnetism sensor 46, which sends rotation angle information of the listener's head to a rotation angle calculator 47. The rotation angle is calculated by the rotation angle calculation unit 47, and the FIR filter coefficient of the corresponding rotation angle is read from the coefficient ROM 48 every time the head rotates by an angle (for example, 5 degrees) of the head-related transfer function (HRTF) measured in advance. .
Since these coefficients are stored in ROM in compressed form,
After being decompressed by the compression coefficient decompression means 49, each F
It becomes an IR filter coefficient.

In the above processing, the decompression of the coefficients is performed by the DSP
, The amount of information sent from the coefficient ROM 48 to the DSP can be reduced. That is, the time for updating the coefficient of the FIR filter can be reduced. This is particularly effective when the filter coefficient needs to be updated instantaneously following the rotation angle of the head as in the present embodiment.

[0055]

As described above in detail, according to the sound image localization apparatus, the FIR filter coefficient processing method, and the FIR filter calculation method according to the present invention, in the sound image localization processing for the acoustic signal, the localization filter is constituted by the FIR filter. In this case, by allocating two or more compressed localization coefficients to the coefficient memory, a FIR filter having a large response number and a long response length can be configured with a small number of coefficient memories. Can be realized with less hardware.

Further, since the rewriting time of the coefficient memory can be shortened, when the localization system has an interactive configuration, it is possible to realize a very good localization.

Further, according to the sound image localization apparatus of the present invention, when the sound image is localized at five positions outside the head, the coefficients of the FIR filter are rewritten according to the rotation angle of the head, and the sound image localization position is moved. When a sound field equivalent to the sound field of a speaker is realized by
Since the product-sum operation is executed while the compressed coefficients are stored in the P internal memory and the compressed coefficients are decompressed, the filter coefficients can be updated efficiently and the coefficient update time can be reduced.

[Brief description of the drawings]

FIG. 1 is a conceptual diagram relating to coefficient memory allocation according to the present invention.

FIG. 2 is an explanatory diagram related to optimization by scaling according to the present invention.

FIG. 3 is an explanatory diagram relating to a compression process according to the present invention.

FIG. 4 is a configuration diagram showing a sound image localization device according to the first embodiment of the present invention.

FIG. 5 is an explanatory diagram relating to an internal configuration of an FIR filter part of FIG. 4;

FIG. 6 is a flowchart (a) for explaining a method of processing an FIR filter coefficient according to the present invention and a flowchart (b) for explaining a method of processing an FIR filter inside a DSP.

FIG. 7 is an explanatory diagram of cross-fade processing in the coefficient switching method of the FIR filter.

FIG. 8 shows a DSP in a coefficient switching method of an FIR filter.
FIG. 4 is an explanatory diagram showing an example of an impulse response for performing convolution operation processing according to FIG.

FIG. 9 is a configuration diagram illustrating a sound image localization device according to a second embodiment of the present invention.

FIG. 10 is a diagram for explaining a sound image localization device according to a third embodiment of the present invention, and is an explanatory diagram of a head-related transfer function (HRTF) in a 5-channel surround sound field.

FIG. 11 is a configuration diagram illustrating a sound image localization device according to a third embodiment of the present invention.

FIG. 12 is a characteristic diagram illustrating an example of a head related transfer function (HRTF).

FIG. 13 is an explanatory diagram of a system in which localization is changed in a game sound field.

FIG. 14 is an explanatory diagram of a system in which localization is changed according to a head rotation angle.

[Explanation of symbols]

1. Desired impulse response 2. Response separation unit (separation means) 3.0 to 3. N-1 compression processing section (compression processing means) 4 coefficient rearrangement section (relocation means) 5 external ROM 6 coefficient memory 7 coefficient separation means 8.0 to 8. N-1 FIR filter 9.0-9. N-1 Rescaling unit (rescaling means) 10 Input signal 11 Output signal

Claims (7)

[Claims] 1. A separating means for separating a localization response impulse train into two parts, a first half and a second half, and a signal of the first half and the second half of the impulse train separated by the separating means is contained in a predetermined number of bits. A first-half and a second-half compression processing means for performing compression processing by scaling to the first half, and a rearrangement for rearranging the scaling coefficients and the compressed localization coefficients used in the compression processing by the first-half and the second half compression processing means according to a predetermined format. Allocating means, a coefficient memory for reading and storing the scaling coefficients relocated by the relocating means, and a coefficient separating means for separating the data stored in the coefficient memory into a first half localization coefficient and a second half localization coefficient. A first half and a second half FIR filter for performing a product sum operation of the localization coefficient and the input signal; and a product sum result of the first half and the second half FIR filter. A sound image localization device comprising: rescaling means for re-scaling. 2. Separation means for dividing a localization response impulse train into N blocks, and compression processing by scaling an impulse train signal for each block separated by the separation means so as to be within a predetermined number of bits. Compression processing means, relocation means for rearranging the scaling coefficients used in the compression processing by the compression processing means and the compressed localization coefficients according to a predetermined format, and scaling coefficients rearranged by the relocation means. A coefficient memory for reading and storing data, coefficient separating means for separating the data stored in the coefficient memory into localization coefficients for each block, and N FIRs for performing a product-sum operation of the localization coefficients and an input signal
A sound image localization apparatus comprising: a filter; and a rescaling unit that rescales a product-sum result of the N FIR filters. 3. A separation step of separating a localization response impulse train into a first half and a second half, a compression processing step of scaling and compressing a signal of the separated impulse train so as to be within a predetermined number of bits, A rearrangement step of rearranging the scaling coefficients used in the compression processing in the compression processing step and the compressed localization coefficients in accordance with a predetermined format; and a FIR filter coefficient processing method. 4. A separation step of separating a localization response impulse train into a first half and a second half, a compression processing step of scaling and compressing a signal of the separated impulse train so as to be within a predetermined number of bits, A rearrangement step of rearranging the scaling coefficient and the compressed localization coefficient used in the compression processing in the compression processing step according to a predetermined format; and externally transferring the scaled coefficient and the compressed localization coefficient rearranged by the rearrangement step. A storing step of storing in the ROM; a step of reading the scaling coefficient and the compressed localization coefficient from the external ROM into a coefficient memory inside the DSP; and a data of the coefficient memory read by the reading step in a first half localization coefficient. And a coefficient separation step for separating into a localization coefficient in the second half, A first-half and a second-half FIR filter calculation step of performing a product-sum operation on each of the localization coefficients separated by the coefficient separation step and the input signal; and a re-scaling step of re-scaling the product-sum results obtained by the FIR filter calculation steps, respectively. And an FIR filter operation method comprising: 5. A separation step of separating a localization response impulse train into N blocks, a compression processing step of scaling and compressing the separated impulse train signal so as to be within a predetermined number of bits, and said compression processing. A rearrangement step of rearranging the scaling coefficients used in the step-by-step compression processing and the compressed localization coefficients according to a predetermined format; and a method of processing FIR filter coefficients. 6. A separation step of separating a localization response impulse train into N blocks, a compression processing step of scaling the separated impulse train signal so as to be within a predetermined number of bits, and performing a compression process; A rearrangement step of rearranging the scaling coefficient and the compressed localization coefficient used in the compression processing step according to a predetermined format; and storing the scaled coefficient and the compressed localization coefficient rearranged by the rearrangement step in an external ROM. A storage step of saving; a step of reading the scaling coefficient and the compressed localization coefficient from the external ROM into a coefficient memory inside a DSP; and a step of reading the data stored in the coefficient memory read by the reading step for each block. A coefficient separating step for separating into localization coefficients; The product sum operation of the localization coefficient and the input signal of each block separated by the number separation step is performed, and the F
An FIR filter operation method, comprising: an FIR filter operation step of operating an IR filter; and a re-scaling step of rescaling the product-sum result of the FIR filter operation step. 7. A sound image localization apparatus which has front, left, right, rear left and rear right audio signal inputs of 5 channels and localizes the input of each 5 channels to five positions by means of ten localization filters, respectively. Rotation angle detection means for detecting the rotation angle of the head of the listener wearing headphones, and rewriting the coefficient of each of the localization filters according to the rotation angle detected by the rotation angle detection means to move the localization position of the sound image By doing so, when realizing a sound field equivalent to the sound field of the speaker, the localization coefficients of the localization filter are stored in a compressed form in the internal memory of the DSP, and the product-sum operation is performed while decompressing the compressed coefficients. Means for executing, a sound image localization device.