JP2005249989A - Sound field control method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a sound field control method in which characteristics having impulse response of precision with a previously set permissible error can be reproduced at each control point without increasing the number of speakers as the number of control points increases. <P>SOLUTION: In the sound field control method which has N pieces of filters 20 where the same input signal is inputted and N (N≥2) output elements 40 applying outputs to the N filters respectively and determine characteristics of the N filters so that predetermined desird signal transmission characteristics are reproduced at M pieces (M≥N+1) control points (where microphones 50 are arranged), the filter characteristics are determined by using precision corresponding to permissible ranges of errors set by the control points. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a sound field control method for changing sound field characteristics of a space such as a room.

When it becomes possible to freely control the sound pressure at any point in the space, for example, a speaker placed only in the front can be used to create a sound as if it is sounding from the back, or a concert The sound of the hall can be enjoyed while at home, and the sound can be reproduced only within a limited area. As one method for controlling sound pressures at a plurality of arbitrary points in this space, a multi-channel control device has been proposed (Patent Document 1). By using this multi-channel control device, it is possible to strictly control a point sound pressure that is at least one less than the number of speakers. However, it is necessary to increase the number of speakers as the number of points to be controlled increases, and the system becomes large, so that the place where it can be used is limited.
On the other hand, the concept of a method for configuring an apparatus that controls the sound pressure at a control point on the basis of the minimum square error using a smaller number of speakers than the control point has been reported (Non-Patent Document 1).

The sound field control method will be described with reference to FIG.
In FIG. 1, N (N ≧ 2) speakers (output elements) 200 _i (i = 1, 2,..., N) are arranged in a space. In addition, M control points (positions where microphones are arranged) are set in the space. In FIG. 1, a microphone 300 _j (j = 1, 2,..., M) is installed at each control point. The same input signal from the input is applied to the speaker 200 _i through the FIR filter 100 _i (i = 1, 2,..., N). By controlling the filter coefficients of these FIR filters 100 _i, the signals applied to the speakers can be individually controlled.
Here, each impulse response between the N speakers and the M control points can be represented by a discrete time series having the same finite length m. M desired impulse responses assumed for each control point are a _j (j = 1, 2,..., M), and the impulse response between the microphone and the speaker is h _{i, j} (i = 1, 2). ,..., N, j = 1, 2,..., M), and when the coefficients of each filter are w _i (i = 1, 2,..., N),
If w _i satisfying is obtained, a desired impulse response can be obtained at each control point. However, when M> N, the equation becomes impossible, so in practice, w _i that minimizes the square error is obtained.

Hereinafter, in order to simplify the description, the left side of equation (1) is denoted by A, and the right side matrix and vector are denoted by H and W, respectively. In this case, equation (1) can be expressed as equation (2).
Can be obtained. Here, “ ⁺ ” represents a pseudo-inverse matrix.

However, in this method, the filter coefficient is obtained so as to minimize the error, and the error is included in principle. It is unknown how much the error is included in each control point, and it is impossible to control the degree of the error at the present time.
Japanese Patent No. 2558445 (Claim 1) Yanagida et al. J. Acoust. Soc. Jpn (E), Vol.4, No.2, pp.107-109 (1983)

In the sound field control method based on the measure of the least square error, it is not known in what ratio the error is distributed at each control point.
An object of the present invention is to propose a sound field control method capable of arbitrarily setting an allowable error ratio at each control point.

  The present invention includes N filters to which the same input signal is supplied, and N (N ≧ 2) output elements that apply the outputs of the N filters, and M (M ≧ N + 1). In the sound field control method for determining the characteristics of the N filters so as to reproduce a predetermined desired signal transfer characteristic with respect to the control points of the control points, the control points correspond to the allowable error range set for each control point. The filter characteristic is determined using accuracy. That is, in the sound field control method for controlling the sound pressure at a plurality of locations (control points) to be a desired one, by performing an operation using a weighting coefficient matrix in which an allowable amount of error at each control point is set, It is possible to set an allowable error ratio for each control point.

  According to the present invention, it is possible to reproduce a characteristic having an impulse response having a preset tolerance error accuracy at each control point without increasing the number of speakers as the number of control points increases. It becomes. By using this method, it is possible to realize a device that emits sound from behind using a speaker arranged in front. Useful for various sound field control, such as the ability to reproduce the characteristics of an arbitrary sound field such as a concert hall in another space such as at home, or to achieve it with high accuracy by localizing the sound field It is.

Consider setting an error tolerance for each control point,
In the configuration shown in FIG. 1, the following equation is obtained using a weighting coefficient matrix G in which the diagonal element is g _j and the other elements are 0 with the allowable amount of error for the jth control point as the weighting coefficient g _j . Constitute. At this time, the weight coefficient g _j is set to be the reciprocal of the allowable ratio of error to the j-th control point.

The configuration shown in FIG. 2 will be specifically described as an example.
First, as an initialization operation, the FIR filter 20 _i inputs an impulse as an input signal x (t) as an all-band pass characteristic. This input signal is discretized into a signal X (k) by the A / D converter 10, digital / analog converted again by the D / A converter 30, and then output from the speaker 40 _i . A microphone is installed at the control point, and each microphone output is an impulse response from each speaker to the microphone. The N × M impulse response vectors h _{i, j} (i = 1, 2,..., N, j = 1, 2,..., M) are recorded in the response recorder 60. On the other hand, the desired response recorder 70 stores a desired impulse response vector a _j (j = 1, 2,..., M) determined for each control point, and its output is a response record. This is input to the filter coefficient determiner 90 together with the output of the filter 60. Here, a zero response can be set as a _j . The weight recorder 80 holds weight coefficients g _j (j = 1, 2,..., M) of each control point, which are also input to the filter coefficient determiner 90. The filter coefficient determiner 90 determines the coefficient of each filter by performing the calculation shown in the equation (7) using these inputs h _{i, j} , a _j , g _j , and each FIR filter 21 ₁ , 21 ₂ , ..., 21 _N. At this time, N × M impulse response vectors h _{i, j} can be obtained by a method of measuring an impulse response (Japanese Patent No. 2725838) using a time-extended pulse as the input signal x (t).
After the initialization operation is completed, filter coefficients are set for each filter. In this state, the input signal x (t) is supplied from the input terminal to the A / D converter 10, and the signal X (k) obtained by discretizing the input signal x (t) by the A / D converter 10 is N. Each of the FIR filters 20 _i (i = 1, 2,..., N) is input. The filtered signal is subjected to digital / analog conversion by the D / A converter 30 and then output from the speaker 40 _i . As a result, characteristics having an impulse response with accuracy corresponding to the allowable error are reproduced at the M control points.

  An example of the simulation result of the sound field control by this invention is shown. FIG. 3 shows the arrangement of speakers and the arrangement of control points (positions where microphones are arranged) set as simulation conditions. In the figure, o indicates the position of the speaker, and x and * indicate control points. The control point was set so that the sound pressure was always 0 at the point of x, and * was set so as to be an impulse response when the sound was emitted from the center speaker alone. This arrangement is a speaker and control point arrangement for realizing local reproduction in which sound is reproduced only within a specific area. The weight is calculated by calculating the sound pressure distribution (FIG. 4) when * is 0.1 and x is 1 and the frequency is 1000 Hz (FIG. 4) and the conventional method, that is, the expression (5) without weighting. The pressure distribution (FIG. 5) is shown. The curve in each figure shows an iso-sound pressure curve, the shading corresponds to the sound pressure intensity, and “•” shows the position of the control point. In the sound pressure distribution obtained by the conventional method, the sound pressure does not necessarily become 0 at the point “•” where the sound pressure is 0 (see FIG. 5: center “•”), and as a result, local reproduction can be realized. Absent. On the other hand, when the weighting control according to the present invention is performed, the sound pressure at the control point “•” is small, and it can be seen that local reproduction can be realized (see FIG. 4: “•”).

The figure for demonstrating the principle of a sound field control method. The block diagram which shows the structural example of the sound field control apparatus for demonstrating the sound field control method of this invention. The figure which shows the arrangement | positioning of the speaker and control point used as the simulation conditions for showing the effect of this invention. The figure which shows the example of the sound pressure distribution at the time of using the sound field control method of this invention. The figure which shows the example of the sound pressure distribution at the time of using the conventional sound field control method.

Explanation of symbols

20 ... FIR filter, 30 ... D / A converter, 40 ... speaker, 50 ... microphone, 60 ... response recorder, 70 ... desired response recorder, 80 ... Weight recorder, 90 ... filter coefficient determiner

Claims (3)

N filters to which the same input signal is supplied and N (N ≧ 2) output elements for applying the outputs of the N filters, and M (M ≧ N + 1) control points. On the other hand, in the sound field control method for determining the characteristics of the N filters so as to reproduce a predetermined desired signal transfer characteristic,
A sound field control method, wherein a filter characteristic is determined using accuracy according to an allowable range of error set for each control point. The sound field control method according to claim 1,
The characteristic of N (N ≧ 2) filters is a transfer characteristic between N output elements and M (M ≧ N + 1) control points, where W is a matrix having the characteristic as an element. A matrix H, a matrix A having M desired transfer characteristics assumed for each control point as elements, and a weighting coefficient having a value corresponding to an allowable error set for each control point as elements Based on a measure of minimum squared error using a generalized inverse matrix
The sound field control method characterized by calculating | requiring and calculating | requiring. The sound field control method according to claim 2,
When the characteristics of N filters (N ≧ 2) are obtained using a generalized inverse matrix, a positive constant δ smaller than the maximum eigenvalue of the matrix is used as a diagonal element so that the inverse matrix does not become unstable. A filter field sound field control method characterized by adding a matrix δI in which other elements are 0 (where I is a unit matrix) and obtaining by the following equation.