KR100571922B1 - Method and apparatus for using room inverse filter - Google Patents

Abstract

A method and apparatus for using a spatial inverse filter is disclosed. In this method, the low-frequency and high-frequency components of the inverse filter are obtained by applying the least square technique and the octave gain control technique to the predetermined target signal and the measured spatial impulse response, and the inverse filter is synthesized by synthesizing the low and high frequency components. And a high frequency component and a low frequency component are divided based on a predetermined reference frequency. Therefore, the designed inverse filter can be used to provide an equalized response even when the listener deviates from the measurement point, and in the band below the reference frequency, the characteristics of the phase response can be corrected to obtain a result that is not influenced by space. In particular, while the conventional octave gain control method cannot compensate for abrupt power fluctuations present in each band, the present invention uses a continuous average value in frequency based on the measured spatial impulse response to compensate for abrupt power fluctuations. It is possible to correct the distortion of the spatial transfer function in a wide listening area.

Description

Method and apparatus for using room inverse filter}

1 is a flowchart illustrating an embodiment of a method for using a space inverse filter according to the present invention.

2 is a flowchart for explaining a preferred embodiment of the present invention in the tenth step of obtaining the low frequency component of the inverse filter.

3 is a flowchart for explaining a preferred embodiment of the present invention in the tenth step of obtaining the high frequency components of the inverse filter.

FIG. 4 is a flowchart for explaining an exemplary embodiment of the present invention with respect to step 80 of FIG. 3.

FIG. 5 is a flowchart for explaining another embodiment of the present invention with respect to step 80 of FIG. 3.

6 is a block diagram of an apparatus for using a space inverse filter according to the present invention.

FIG. 7 is a block diagram of a preferred embodiment of the present invention of the low frequency component generator shown in FIG. 6.

FIG. 8 is a block diagram of a preferred embodiment of the present invention of the least squares technique application shown in FIG.

FIG. 9 is a block diagram of an embodiment of the present invention of the least squares technique application shown in FIG. 8.

FIG. 10 is a block diagram of a preferred embodiment of the present invention of the high frequency component generating unit shown in FIG. 6.

FIG. 11 is a block diagram of a preferred embodiment of the present invention of the moving average filter shown in FIG.

12 is a block diagram of another embodiment according to the present invention of the moving average filter shown in FIG.

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a room inverse filter that can be used to improve sound quality in a home theater and the like, and more particularly, to a method and apparatus for using a space inverse filter for designing and using a space inverse filter. .

In general, the input signal x is distorted by the space transfer function G, which is a characteristic function of space. Conventional methods for compensating for such distortion include a method using a least square (LS) technique and a method using an octave gain control technique.

First, conventional methods for compensating distortion using least squares techniques have been titled "Further Investigations of Inverse Filtering". Soulodre and MCLavoie and published in 2003 by the American Audio Engineering Society, "115th Audio Engineering Society Convention, preprint 5923" or "Digital Filter Design for Inversion Problems in Sound Reproduction." The Journal of Audio Engineering Society, Volumn 47, No., published by Kirkeby and Ph.A.Nelson and published in July 1999 by the American Audio Engineering Society. 7/8 (pp. 583-595) '. The basics that accompany it are entitled "Discrete-Time Signal Processing" under the title 'Oppenheim. A. V 'and' Schafer.AW ', published in 1989 by Prentice-Hall Publishers in Englewood Cliffs, or by the author, Haykin.S under the title "Adaptive Filter Theory," The book is published in 1996 in 3rd edition by Prentice-Hall, based in Englewood Cliffs.

 The conventional method disclosed herein provides an error between the ideal target signal and the actual target signal when the sound source passes through the spatial inverse filter, assuming that there is little environmental change such as the sound source, the position of the listener, or the arrangement of objects over time. Minimize the square of. This conventional method has the advantage of compensating for the phase of the impulse response with mixed-phase characteristics. However, this conventional method not only requires a large amount of computation to generate an inverse filter, but also has a problem of distorting the signal more when the source or the listener's position changes. As a result, considering that a slight head movement is essential during listening, there is a problem in that sound quality may be lost if the phase characteristics of the listening space are corrected to a certain high frequency band by using the conventional method.

Next, a conventional method of compensating distortion using the octave gain control technique is entitled "Harma, A." entitled "Frequency-Warped Signal Processing for Audio Applications." And it is disclosed in Journal of the Audio Engineering Society's 108 ^th AES Convention preprint 5171 presented in 2000 by the 'Karjalainen, M.'. In the conventional method disclosed herein, after dividing a signal into bands, the gain is set differently for each band according to the power of each band to smooth the overall frequency response. This conventional method has a short filter length, does not require a large amount of computation, and minimizes signal processing to prevent sound quality damage. However, this conventional method has a problem in that the frequency resolution is so low that the characteristics of the acoustic listening space cannot be precisely corrected, and the consideration of phase characteristics cannot be taken into account at all.

The technical problem to be achieved by the present invention is to generate an inverse filter for each frequency component differently, and to provide an equalized response even when the listener of the acoustic signal moves by using the generated inverse filter. The present invention provides a method of using a spatial inverse filter that can correct the characteristics of a phase response of an acoustic signal.

Another technical problem to be solved by the present invention is to generate an inverse filter differently for each frequency component, and to provide an equalized response even when a listener of an acoustic signal moves by using the generated inverse filter. An object of the present invention is to provide an apparatus for using a spatial inverse filter capable of correcting characteristics of phase response of an acoustic signal.

In accordance with an aspect of the present invention, there is provided a method of using a spatial inverse filter, by applying a least square technique and an octave gain control technique to a predetermined target signal and measured spatial impulse response to obtain low and high frequency components of an inverse filter, respectively. And obtaining the inverse filter by synthesizing the low frequency component and the high frequency component, wherein the high frequency component and the low frequency component are divided based on a predetermined reference frequency.

According to another aspect of the present invention, there is provided an apparatus for using a spatial inverse filter, including: a low frequency component generator for generating a low frequency component of an inverse filter by applying a least square technique to a predetermined target signal and a measured spatial impulse response; A high frequency component generator for generating a high frequency component of the inverse filter by applying an octave gain control technique to a signal and the measured spatial impulse response, and synthesizing the low frequency component and the high frequency component and outputting the synthesized result as the inverse filter The high frequency component and the low frequency component are preferably divided around a predetermined reference frequency.

Hereinafter, a method of using a space inverse filter according to the present invention will be described with reference to the accompanying drawings.

1 is a flowchart illustrating an embodiment of a method for using a spatial inverse filter according to the present invention, comprising: obtaining an inverse filter (10th and 12th steps) and filtering an input acoustic signal (14th and Sixteenth steps).

In the method for using a space inverse filter according to the present invention, first, a high frequency component and a low frequency component of a space inverse filter (hereinafter referred to as an inverse filter) are obtained (step 10). Here, the low-frequency component of the inverse filter is obtained by applying the least square technique described above to the predetermined target signal and the measured spatial impulse response. Here, the predetermined target signal may be given by the user as a signal to be finally obtained by applying the inverse filter to be generated by the spatial inverse filter using method according to the present invention to the measured spatial impulse response. In addition, the high frequency component of the inverse filter is obtained by applying the above-described octave gain control technique to the predetermined target signal and the measured spatial impulse response. At this time, the high frequency component and the low frequency component of the inverse filter obtained in the tenth step are divided around a predetermined reference frequency.

According to the present invention, the reference frequency f _ref may be obtained as in Equation 1 below.

Here, b is, for example, '1' as a variable, fs means sampling frequency, and N indicates wavelet analysis order. For example, when the sampling frequency fs is 44.1 Hz and the wavelet analysis is performed four times (N = 4), the reference frequency f _ref is 1.378 Hz.

FIG. 2 is a flowchart for explaining a preferred embodiment of the present invention in a tenth step of obtaining a low frequency component of an inverse filter, by decomposing a predetermined target signal and a measured spatial impulse response. Obtaining a response (steps 40 to 46), obtaining an impulse response of the inverse filter (step 48), and performing low pass filtering after smoothing the impulse response of the inverse filter (50th and 46th steps). 52 steps).

Referring to FIG. 2, a wavelet analysis technique is applied to a predetermined target signal to represent a predetermined target signal in a wavelet domain, and the signal of the lowest band in the wavelet region is a low-pass subband target. Separate as a response (step 40). Here, the wavelet analysis technique is authored by 'Vetterli.M' and 'Kovacevic.J' entitled "Wavelets and SubbandCoding.Signal Processing Series" and published in 1995 by Prentice-Hall, located in Englewood Cliffs, NJ. Disclosed in the book.

After step 40, the measured spatial impulse response is applied to the measured spatial impulse response to represent the measured spatial impulse response as a wavelet region, and the signal of the lowest band in the wavelet region is separated as a low-pass subband spatial impulse response. (Step 42).

After step 42, the unit impulse signal is applied to the unit impulse signal to represent the unit impulse signal in the wavelet region, and the high band signal except the lowest band in the wavelet region is separated as a high-frequency subband unit impulse response. Do it (step 44).

The above forty-fourth, twenty-fourth, and forty-fourth steps may be performed in any order, or may be performed simultaneously. In this case, each time the wavelet transform is applied once in each of the 40th, 42nd and 44th steps, the bandwidth of each subband is reduced by 1/2.

After the 44th step, the lowpass subband inverse filter response of the inverse filter for the lowpass subband spatial impulse response in the wavelet region is obtained using a least square technique using the lowpass subband target response as the target signal (step 46).

After operation 46, the impulse response of the inverse filter is obtained in the time domain by applying a wavelet reconstruction technique to the low pass subband inverse filter response and the high pass subband impulse response.

After step 48, the impulse response of the inverse filter is smoothed (step 50). After step 50, the low frequency component of the impulse response of the smoothed inverse filter is filtered, and the filtered result is determined as the low frequency component of the inverse filter (step 52).

According to the present invention, the tenth step shown in FIG. 2 may not be provided with the fifty step. In this case, after step 48, the low frequency component of the impulse response of the inverse filter is filtered, and the filtered result is determined as the low frequency component of the inverse filter (step 52).

3 is a flowchart illustrating a preferred embodiment of the present invention in a tenth step of obtaining a high frequency component of an inverse filter, comprising: moving average filtering the measured spatial impulse response (step 80) and moving average filtered High pass filtering the impulse response of the inverse filter obtained by using the result (82 th and 84 th steps).

Referring to FIG. 3, moving average filtering of the measured spatial impulse response is performed (operation 80). Here, moving average filtering means replacing the measured power value of the spatial impulse response with the average power value of a specific band located around the measured space impulse response. Herein, the length of the specific band is set narrow in the low frequency band and long in the high frequency band in consideration of human hearing characteristics.

FIG. 4 is a flowchart for describing an embodiment 80A according to the present invention with respect to step 80 of FIG. 3, wherein the power of the spatial impulse response (steps 100 and 102) and power are obtained. Cyclically convolution with the window signal (step 104).

According to one embodiment of the present invention, the length of the moving average filtering is set short in the low frequency band and is set frequency dependent in the high frequency band. In this case, first, in order to perform moving average filtering, the measured spatial impulse response is expressed in the frequency domain (step 100). For this purpose, fast Fourier transform (FFT) of the measured spatial impulse response is performed. After step 100, the power of the spatial impulse response expressed in the frequency domain is obtained (step 102).

Here, step 102 may be performed before step 100. In this case, the power of the measured spatial impulse response is obtained, and the power of the spatial impulse response is expressed in the frequency domain. That is, fast Fourier transform the power of the spatial impulse response.

After step 102, the power of the spatial impulse response is cyclically convolved with the window signal as in Equation 2 below, and the cyclically convolved result is determined as the moving average filtered result (step 104). Here, the window signal, which is a low pass filter, is referred to as "Generalized Fractional-Octave Smoothing of Audio and Acoustic Responses". Hatziantoniou 'and' J.N. Mourjopoulos' is published in a paper published in 2000 in Vol. 48, No. 4 of the Journal of Audio Engineering Society.

는 이동 평균 필터링된 결과를 나타내고,

는 주파수로 표현되는 공간 임펄스 응답의 전력을 나타내고, W_sm(i)는 윈도우 신호를 나타내고,

는 순환 콘볼루션을 나타내고, M은 이동 평균 필터링의 길이를 나타낸다.here,

Represents the moving average filtered result,

Denotes the power of the spatial impulse response expressed in frequency, W _sm (i) denotes the window signal,

Denotes cyclic convolution and M denotes the length of the moving average filtering.

FIG. 5 is a flowchart for explaining another embodiment 80B according to the present invention with respect to step 80 of FIG. 3, wherein power of the measured spatial impulse response is obtained (steps 110 and 112). Converting the power to a log scale and performing interpolation (steps 114 and 116) and reducing the frequency interpolated result with a window signal and circulating the interpolated result (steps 118 and 120). Steps).

According to another embodiment of the present invention, the length of the moving average filtering is fixed. In other words, the bandwidth of the moving average filtering is set to a constant. In this case, first, in order to perform moving average filtering, the power of the measured spatial impulse response is expressed by expressing the frequency domain (steps 110 and 112). Steps 110 and 112 shown in FIG. 5 correspond to steps 100 and 102 shown in FIG. 4 and perform the same function, and thus a detailed description thereof will be omitted.

는 로그 척도로 표현되고 보간된 전력을 의미한다.After step 112, the power of the spatial impulse response expressed on the frequency scale is represented on the log scale (step 114). After step 114, the power of the spatial impulse response expressed on the log scale is interpolated (step 116). After step 116, the interpolated result is cyclically convolved with the window signal as shown in Equation 2 (step 118). In this case, in Equation 2,

Denotes the power interpolated and expressed on a logarithmic scale.

After step 118, the cyclically convoluted result represented by the log scale is represented on the frequency scale, and the cyclically convolved result represented by the frequency scale is determined as the moving average filtered result (step 120).

)의 크기(G_ps)에 역(1/G_ps)과 소정의 목표 신호(p)를 승산하고, 승산된 결과를 역 필터의 임펄스 응답(h")의 크기(

)로서 결정한다(제82 단계).On the other hand, after the 80th step shown in Figure 3, the moving average filtered result as shown in the following equation (3)

) Multiplies the magnitude (G _ps ) by the inverse (1 / G _ps ) and the predetermined target signal p, and multiplies the result by the magnitude of the impulse response (h ") of the inverse filter (

(Step 82).

)와 선형 위상을 갖는 역 필터의 임펄스 응답의 고역 성분을 필터링하고, 필터링된 결과를 역 필터의 고주파 성분으로서 결정한다(제84 단계).The size determined in step 82 (

And the high frequency component of the impulse response of the inverse filter having a linear phase and determine the filtered result as the high frequency component of the inverse filter (step 84).

2 and 3 described above may be performed simultaneously or sequentially. Therefore, after the 52nd step shown in FIG. 2 is performed, the process may proceed to the twelfth step as shown in FIG. 2, and unlike the FIG. 2 step, the 52nd step may be performed after the 52nd step shown in FIG. It may be. Similarly, after the 84th step shown in FIG. 3 is performed, the process may proceed to the twelfth step as shown in FIG. 3, and the 40th step shown in FIG. You can also proceed.

On the other hand, after the tenth step, a low frequency component and a high frequency component are synthesized, and the synthesized result is output as an inverse filter (step 12).

After the twelfth step, the inverse filter resulting from the synthesis of the low frequency component and the high frequency component is smoothed (step 14). Therefore, the distortion of the overlapping portion when the low frequency component and the high frequency component of the inverse filter are synthesized can be minimized by the smoothing of the fourteenth step.

After the fourteenth step, the inverse filter hi smoothed to the input acoustic signal Xi is convolved as in Equation 4 below, and the convoluted result is determined as the output acoustic signal Xo (16). step).

Where * is a symbol representing a convolution.

According to an embodiment of the present invention, the method for using a space inverse filter according to the present invention provides all the tenth to sixteenth steps shown in FIG. 1 as described above.

According to another embodiment of the present invention, the method for using the space inverse filter according to the present invention, unlike FIG. 1, provides the tenth, twelfth and sixteenth steps, and does not provide the fourteenth step. In this case, an inverse filter hi not smoothed to the input acoustic signal Xi is convolved, and the convoluted result is determined as the output acoustic signal Xo (step 16).

According to another embodiment of the present invention, the method of using the space inverse filter according to the present invention, unlike FIG. 1, provides the tenth, twelfth and fourteenth steps, and does not provide the sixteenth step.

According to another embodiment of the present invention, the method of using the space inverse filter according to the present invention, unlike FIG. 1, provides only the tenth and twelfth steps and does not provide the fourteenth and sixteenth steps.

Hereinafter, with reference to the accompanying drawings, the configuration and operation of the apparatus for using a space inverse filter according to the present invention will be described as follows.

6 is a block diagram of the apparatus for using a space inverse filter according to the present invention, wherein a low frequency component generator 200, a high frequency component generator 202, a synthesizer 204, a first smoothing unit 206, and an acoustic signal cone ball are shown. It is composed of a cushion portion 208.

The apparatus for using a space inverse filter illustrated in FIG. 6 serves to perform the method for using a space inverse filter illustrated in FIG. 1.

The low frequency component generator 200 and the high frequency component generator 202 illustrated in FIG. 6 perform a tenth step illustrated in FIG. 1. That is, the low frequency component generator 200 applies the least squares technique to the predetermined target signal p input through the input terminal IN1 and the measured spatial impulse response g input through the input terminal IN2 to perform the inverse filter. low-frequency component (h "generates a _L) and _the low-frequency component (h of the generated inverse filter, and outputs the _L) to the synthesis unit 204. the The high frequency component generator 202 applies an octave gain control technique to a predetermined target signal p input through the input terminal IN1 and a measured spatial impulse response g input through the input terminal IN2, thereby applying a high frequency of the inverse filter. The component h " _H is generated and the generated high frequency component h" _H is output to the synthesis part 204.

In order to perform the twelfth step, the synthesis unit 204 may combine the low frequency component h ′ _l input from the low frequency component generator 200 and the high frequency component h ″ _H input from the high frequency component generator 202. The synthesized result is output to the first smoothing unit 206 as an inverse filter.

In order to perform the fourteenth step, the first smoothing unit 206 smoothes the inverse filter, which is the output of the combining unit 204, and outputs the smoothed inverse filter to the acoustic signal convolution unit 208.

In order to perform the sixteenth step, the acoustic signal convolution unit 208 convolves the smoothed inverse filter input from the first smoothing unit 206 to the input acoustic signal input through the input terminal IN3, The result is output through the output terminal OUT1 as an output sound signal.

As described above, when the space inverse filter using method shown in FIG. 1 does not provide the fourteenth step, the space inverse filter using device shown in FIG. 6 does not provide the first smoothing unit 206. In this case, in order to perform the sixteenth step, the acoustic signal convolution unit 208 convolves the inverse filter input from the combining unit 204 to the input acoustic signal input through the input terminal IN3 and convolves the result. Is output through the output terminal OUT1 as an output sound signal.

In addition, when the method for using the space inverse filter illustrated in FIG. 1 does not provide the sixteenth step, the apparatus for using the space inverse filter illustrated in FIG. 6 does not provide the acoustic signal convolution unit 208.

Hereinafter, with reference to the accompanying drawings, the configuration and operation of each embodiment of the apparatus for using a space inverse filter according to the present invention will be described as follows.

FIG. 7 is a block diagram of a preferred embodiment 200A of the low frequency component generator 200 shown in FIG. 6 according to the present invention, and includes a least square method applying unit 220, a second smoothing unit 222, and a low range. It consists of a pass filter 224.

The low frequency component generator 200A illustrated in FIG. 7 performs the steps 40 through 52 illustrated in FIG. 2.

In order to perform the 40 th to 48 th steps shown in FIG. 2, the least-square method applying unit 220 shown in FIG. 7 is provided through a predetermined target signal p and an input terminal IN5 input through the input terminal IN4. The least square method is applied to the input measured spatial impulse response g, and the result h 'of the least square method is applied to the second smoothing unit 222.

FIG. 8 is a block diagram of a preferred embodiment 220A according to the present invention of the least squares technique applying unit 220 shown in FIG. 7, including a target response decomposition unit 240, a spatial impulse response decomposition unit 242, The unit impulse response decomposition unit 244, the least squares technique inverse filter 246, and the impulse response synthesis unit 248.

The least squares technique applying unit 220A shown in FIG. 8 performs the steps 40 through 48 shown in FIG. 2.

In order to perform step 40, the target response decomposition unit 240 applies a wavelet analysis technique to a predetermined target signal p input through the input terminal IN6 to convert the predetermined target signal p into a wavelet region. And separate the signal of the lowest band in the wavelet region as the low pass subband target response p _k , and output the separated low pass sub band target response p _k to the least squares technique inverse filter 246.

In order to perform the forty-second step, the spatial impulse response decomposition unit 242 applies a wavelet analysis technique to the measured spatial impulse response g input through the input terminal IN7 to apply the spatial impulse response g to the wavelet region. the signal representation, and the lowest in the wavelet domain into a band with a low sub-band space impulse response (g _k) separation, and the separated low-pass sub-band space impulse response (g _k) a least squares technique the inverse filter 246 as Output

In order to perform the 44th step, the unit impulse response decomposition unit 244 applies a wavelet analysis technique to the unit impulse signal inputted through the input terminal IN8, that is, the delta signal δ, and processes the unit impulse signal δ. The high band signal, except for the lowest band, in the wavelet region and the lowest band in the generated wavelet region, the impulse response (δ ₁ , δ ₂ , ... and δ _km ), where 1 ≦ m ≦ k -1), and outputs the separated high-band subband unit impulse responses δ ₁ , δ ₂ ,..., And δ _km to the impulse response synthesizer 248.

In order to perform the forty-seventh step, the least square technique inverse filter 246 is a low-pass subband space by a least square technique using a low-pass subband target response p _k input from the target response decomposition unit 240 as a target signal. The low pass subband inverse filter response h _k of the inverse filter with respect to the impulse response g _k is generated as in Equation 5 below, and the generated low pass subband inverse filter response h _k is generated by the impulse response synthesizer 248. )

Here, g _k ^T means a transpose of g _k .

In order to perform the 48 _th step, the impulse response synthesizer 248 inputs the low pass subband inverse filter response h _k input from the least square technique inverse filter 246 and the high pass input from the unit impulse response decomposition unit 244. The wavelet reconstruction technique is applied to the subband impulse responses (δ ₁ , δ ₂ , ... and δ _km ) to generate the impulse response (h ') of the inverse filter in the time domain, and the impulse response of the generated inverse filter. Output (h ') through output terminal OUT3.

FIG. 9 is a block diagram of an embodiment of the present invention of the least square method applying unit 220A illustrated in FIG. 8, and includes a target response decomposition unit 240A, a spatial impulse response decomposition unit 242A, and a unit impulse response decomposition unit ( 244A), least square technique inverse filter 246, and impulse response synthesizer 248A.

The target response resolver 240A illustrated in FIG. 9 performs the same role as the target response resolver 240 illustrated in FIG. 8. To this end, the target response decomposition unit 240A includes the decomposition units 260, 262, 264, 266, ..., 268 and 270 and the down samplers 310, 312, ..., 314 and 316). Here, two decomposition parts 260 and 262 and one down sampler 310 form one cell 250. Thus, the target response resolver 240A has a plurality of cells.

The spatial impulse response decomposition unit 242A performs the same role as the spatial impulse response decomposition unit 240 illustrated in FIG. 8. To this end, the spatial impulse response resolver 242A consists of resolvers 272, 273, 274, 276, 278, and 280 and down samplers 318, 320, ..., 322 and 324. do.

The unit impulse response decomposition unit 244A performs the same role as the unit impulse response decomposition unit 244 illustrated in FIG. 8. To this end, the unit impulse response decomposition section 244A includes resolution sections 282, 284, 274, 286, 288, ..., 290, and 292 and down samplers 326, 328, 330, 332, ..., 334 and 336).

The impulse response synthesizer 248A performs the same role as the impulse response synthesizer 248 shown in FIG. 8. To this end, the impulse response synthesizer 248A includes reconstruction sections 294, 296, ..., 298, 300, 302 and 304 and upsamplers 338, 340, ..., 342, 344, 346 and 348).

Each of the down samplers illustrated in FIG. 9 may lower the sampling frequency by q times, and each of the up samplers may increase the sampling frequency by q times. Here, q may be, for example. The detailed operation of these down samplers, disassembly and reconstructions is Prentice, written by 'Vetterli.M' and 'Kovacevic.J' under the title "Wavelets and SubbandCoding.Signal Processing Series" and located in Englewood Cliffs, NJ. It is published in a book published in 1995 by Hall.

In order to perform the 50th step shown in FIG. 2, the second smoothing unit 222 shown in FIG. 7 smoothes the output of the least squares technique applying unit 220 and outputs the smoothed result to the low pass filter 224. Will output

In order to perform the fifty-second step, the low pass filter 224 filters the low pass component of the smoothed result in the second smoothing unit 222, and outputs the filtered result as the low frequency component h ' _L of the inverse filter. Output to the combining unit 204 through OUT2.

When the method illustrated in FIG. 2 does not provide the fifty step, the low frequency component generator 200A shown in FIG. 7 does not provide the second smoothing unit 222. In this case, the low pass filter 224 filters the low pass component of the result of applying the least squares technique, that is, the impulse response h 'of the inverse filter output from the impulse response synthesis unit 248, and filters the filtered result. in "determining as _(L, a low-frequency component (h station of the determined low-frequency filter component h), and outputs a _L) through an output terminal OUT2.

FIG. 10 is a block diagram of a preferred embodiment 202A according to the present invention of the high frequency component generator 202 shown in FIG. 6, which includes a moving average filter 400, a magnitude generator 402 and a high pass filter ( 404).

The high frequency component generator 202A illustrated in FIG. 10 performs the 80th through 84th steps illustrated in FIG. 3.

In order to perform step 80, the moving average filter 400 performs moving average filtering on the measured spatial impulse response input through the input terminal IN9, and outputs the moving average filtered result to the size generator 402.

FIG. 11 is a block diagram of a preferred embodiment 400A of the moving average filter 400 of FIG. 10 according to the present invention, and includes a power calculation unit 420 and a first circular convolution unit 422. .

The moving average filter 400A illustrated in FIG. 11 serves to perform step 80A illustrated in FIG. 4.

In order to perform the steps 100 and 102, the power calculation unit 420 expresses and calculates the power of the measured spatial impulse response input through the input terminal IN12 in the frequency domain, and calculates the calculated power represented in the frequency domain. Output to the first circular convolution unit 422.

In order to perform step 104, the first cyclic convolution unit 422 cyclically convolves the power of the spatial impulse response input from the power calculator 420 with the window signal, and performs the moving average filtering on the cyclically convoluted result. As a result, the result is output to the size generator 402 through the output terminal OUT5.

12 is a block diagram of another embodiment 400B of the moving average filter 400 of FIG. 10 according to the present invention, which includes a power calculation unit 440, a scale conversion unit 442, an interpolation unit 444, The second circulation convolution unit 446 and the scale reduction unit 448.

Moving average filter 400B shown in FIG. 12 serves to perform step 80B shown in FIG.

In order to perform steps 110 and 112, the power calculator 440 calculates the power of the measured spatial impulse response input through the input terminal IN13 in the frequency domain, and calculates the power of the spatial impulse response expressed in the frequency domain. The power is output to the scale converter 442.

In order to perform step 114, the scale converter 442 expresses the power of the spatial impulse response expressed on the frequency scale input from the power calculator 440 on a log scale, and the spatial impulse response expressed on the log scale. The power is output to the interpolator 444.

In order to perform step 116, the interpolator 444 interpolates the power of the spatial impulse response represented by the log scale input from the scale converter 442, and outputs the interpolated result to the second cyclic convolution unit 446. Will output

In order to perform step 118, the second circular convolution unit 446 circularly convolves the interpolated result input from the interpolator 444 with the window signal, and measures the circularly convolved result as the scale reduction unit 448. Will output

In order to perform step 120, the scale reduction unit 448 expresses the result of the cyclic convolution in the second cyclic convolution unit 446 represented by the log scale on a frequency scale, and the cyclic convolution represented by the frequency scale. The result is output to the size generator 402 through the output terminal OUT6 as a moving average filtered result.

In order to perform step 82, the magnitude generator 402 multiplies the inverse of the magnitude of the moving average filtered result by the moving average filter 400 and a predetermined target signal input through the input terminal IN10, and multiplies the multiplied result. The result is output to the high pass filter 404 as the magnitude of the impulse response of the inverse filter.

In order to perform step 84, the high pass filter 404 filters the high frequency component of the impulse response of the inverse filter having the magnitude and the linear phase determined by the magnitude generator 402, and filters the high frequency component of the inverse filter. (h " _H ) is output to the synthesis unit 204 through the output terminal OUT4. To this end, the high pass filter 404 may input the linear phase of the impulse response of the inverse filter from the outside through the input terminal IN11. You can also create your own internally.

As described above, the method and apparatus for using a spatial inverse filter according to the present invention use a least square method considering phases in a band below a reference frequency and an octave gain control scheme not dependent on the position of a space in a band larger than the reference frequency. The inverted filter is designed using, and the designed inverse filter can be used to provide an equalized response even when the listener deviates from the measurement point. In particular, while the conventional octave gain control method cannot compensate for the sudden power fluctuations present in each band, the present invention uses a continuous average value in frequency based on the measured spatial impulse response. So it can compensate for sudden power fluctuations The distortion of the spatial transfer function in the listening area can be corrected.

Claims (20)

Obtaining a low frequency component and a high frequency component of an inverse filter by applying a least square technique and an octave gain control technique to a predetermined target signal and measured spatial impulse response, respectively; Obtaining the inverse filter by synthesizing the low frequency component and the high frequency component; And Convolving said inverse filter to an input acoustic signal, and determining the convolved result as an output acoustic signal, The high frequency component and the low frequency component are divided based on a predetermined reference frequency. Obtaining the high frequency component, Moving average filtering the measured spatial impulse response; Multiplying a magnitude of a moving average filtered result by an inverse and the predetermined target signal and outputting the multiplied result as a magnitude of an impulse response of the inverse filter; And Filtering the high frequency component of the impulse response of the inverse filter having the determined magnitude and linear phase, and outputting the filtered result as the high frequency component of the inverse filter. The method of claim 1, wherein the spatial inverse filter is used. Convolving said inverse filter to an input acoustic signal and determining the convolved result as an output acoustic signal. The method of claim 1, wherein the reference frequency (fref) is obtained as follows.

(Where b is a variable, fs is the sampling frequency, and N is the number of wavelet transforms.) The method of claim 1, wherein the spatial inverse filter is used. And smoothing the inverse filter that is a result of synthesizing the low frequency component and the high frequency component. The method of claim 1, wherein the obtaining low frequency components of the inverse filter Separating a signal of the lowest band in the wavelet region generated by applying a wavelet analysis technique to the predetermined target signal as a low pass subband target response; Separating the lowest band signal as a low pass subband spatial impulse response in the wavelet region generated by applying the wavelet analysis technique to the measured spatial impulse response; Separating a high band signal except for the lowest band as a high pass subband impulse response from a wavelet region generated by applying the wavelet analysis technique to a unit impulse signal; Obtaining a lowpass subband inverse filter response of the inverse filter with respect to the lowpass subband spatial impulse response using the least square technique of using the lowpass subband target response; Obtaining an impulse response of the inverse filter in a time domain by applying a wavelet reconstruction technique to the low pass subband inverse filter response and the high pass subband impulse response; And Filtering the low pass component of the impulse response of the inverse filter, and determining the filtered result as the low frequency component of the inverse filter. 6. The method of claim 5, wherein obtaining the low frequency component of the inverse filter Smoothing the impulse response of the inverse filter, The low frequency component of the inverse filter corresponds to a result of filtering the low frequency component of the smoothed result. The method of claim 1, wherein the step of obtaining the high frequency component of the inverse filter Moving average filtering the measured spatial impulse response; Multiplying a magnitude of a moving average filtered result by an inverse and the predetermined target signal, and determining the multiplied result as a magnitude portion of an impulse response of the inverse filter; And Filtering the high frequency component of the impulse response of the inverse filter having the determined magnitude and linear phase, and determining the filtered result as the high frequency component of the inverse filter. 8. The method of claim 7, wherein the length of the moving average filtering is set short in the low frequency band and long in the high frequency band. 8. The method of claim 7, wherein the moving average filtering Expressing the power of the measured spatial impulse response in a frequency domain; And And cyclically convolving the power of the spatial impulse response and determining the cyclically convolved result as the moving average filtered result. 8. The method of claim 7, wherein the length of the moving average filtering is fixed. The method of claim 10, wherein the moving average filtering step Expressing and calculating the power of the measured spatial impulse response in a frequency domain; Expressing a power of the spatial impulse response expressed on a frequency scale on a logarithmic scale; Interpolating the power of a spatial impulse response represented by the logarithmic scale; Cyclically convolving the interpolated result with a window signal; And Representing the cyclically convoluted result represented by a logarithmic scale on a frequency scale, and determining the cyclically convoluted result represented by a frequency scale as the moving average filtered result. How to use. A low frequency component generator for generating a low frequency component of an inverse filter by applying a least square method to a predetermined target signal and a measured spatial impulse response; A high frequency component generator for generating a high frequency component of the inverse filter by applying an octave gain control technique to the target signal and the measured spatial impulse response; A synthesizer which synthesizes the low frequency component and the high frequency component and outputs the synthesized result as the inverse filter; And An acoustic signal convolution unit for convolving the inverse filter to an input acoustic signal and outputting the convoluted result as an output acoustic signal, The high frequency component and the low frequency component are divided based on a predetermined reference frequency. The high frequency component generation unit, A moving average filtering unit for moving average filtering the measured spatial impulse response; A magnitude generator for multiplying a magnitude of a moving average filtered result by an inverse and the predetermined target signal and outputting the multiplied result as a magnitude of an impulse response of the inverse filter; And And a high pass filter for filtering a high frequency component of an impulse response of the inverse filter having the determined magnitude and linear phase and outputting the filtered result as a high frequency component of the inverse filter. The apparatus of claim 12, wherein the apparatus for using a space inverse filter is And a sound signal convolution unit for convolving the inverse filter to an input sound signal and outputting the convoluted result as an output sound signal. The apparatus of claim 12, wherein the apparatus for using a space inverse filter is And a first smoothing unit for smoothing an inverse filter which is an output of the synthesis unit. The method of claim 12, wherein the low frequency component generating unit A least squares technique applying unit for applying the least squares technique to the target signal and the measured spatial impulse response; And And a low pass filter for filtering the low pass component of the result to which the least squares technique is applied and determining the filtered result as the low frequency component of the inverse filter. The method of claim 15, wherein the least squares technique application unit A target response decomposition unit for separating and outputting a signal of the lowest band in the wavelet region generated by applying a wavelet analysis technique to the predetermined target signal as a low-pass subband target response; A spatial impulse response decomposition unit for separating and outputting the lowest band signal as a low-pass subband spatial impulse response in the wavelet region generated by applying the wavelet analysis technique to the measured spatial impulse response; A unit impulse response decomposition unit for separating and outputting a high band signal except for the lowest band as a high frequency subband unit impulse response in a wavelet region generated by applying the wavelet analysis technique to a unit impulse signal; A least squares inverse filter for generating a low-pass subband inverse filter response of the inverse filter to the low-pass subband spatial impulse response by the least squares technique having the low-pass subband target response as a target signal; And An impulse response synthesizer for generating an impulse response of the inverse filter in a time domain by applying a wavelet reconstruction technique to the low pass subband inverse filter response and the high pass subband impulse response; And the low pass filter filters a low pass component of an impulse response of the inverse filter, and outputs the filtered result as a low frequency component of the inverse filter. The method of claim 15, wherein the low frequency component generating unit A second smoothing unit is further provided to smooth the output of the least square technique application unit. And the low pass filter filters a low pass component of the result smoothed by the second smoothing unit, and outputs the filtered result as the low frequency component of the inverse filter. delete The method of claim 12, wherein the moving average filtering unit A power calculation unit expressing and calculating the power of the measured spatial impulse response in a frequency domain; And And a first cyclic convolution unit configured to cyclically convolve power of the spatial impulse response with a window signal, and output a cyclically convolved result as the moving average filtered result. The method of claim 12, wherein the moving average filtering unit A power calculation unit expressing and calculating the power of the measured spatial impulse response in a frequency domain; A scale converter for representing the power of the spatial impulse response expressed on a frequency scale on a log scale; An interpolator for interpolating power of a spatial impulse response expressed by the log scale; A second circular convolution unit for circularly convolving the interpolated result with a window signal; And A scale reduction unit configured to express a cyclically convoluted result in the second circular convolution unit represented by a log scale on a frequency scale and output the cyclically convoluted result represented by a frequency scale as the moving average filtered result Apparatus using a space inverse filter, characterized in that.

Images