JPH06149277A - Filter coefficient generation method - Google Patents

Abstract

PURPOSE:To obtain a coefficient to be set in an FIR filter mechanically and easily and to obtain the coefficient for utilizing the FIR filter at a maximum by finding many extremal values of an impulse response curve and extracting a specific number of extremal values in the decreasing order of the absolute values of data values. CONSTITUTION:A signal processor 23 consists of an initial reflected sound generation part 231 which simulates an initial reflected sound and a rear reverberation sound generation part 232 which simulates a reflected sound (reverberation sound) following the initial reflected sound. Here, the object characteristic impulse response curve is found by an initial reflected sound generating means 232-1-1. This impulse response curve is found by measuring an actual sound field. Many extremal values of the impulse response curve are found and a desired number of data and their generation time data (delay quantity) are extracted in the decreasing order of the absolute values among the found extremal values, thereby finding the filter coefficient mechanically.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a filter for obtaining a filter coefficient set in a filter when an acoustic signal is passed through a filter for the purpose of simulating the acoustic characteristic of a hall or the like and the characteristic is converted. The present invention relates to a coefficient generation method.

[0002]

2. Description of the Related Art In recent years, acoustic facilities such as halls are required to have various acoustic characteristics due to diversification of applications. For example, when a lecture is held, acoustic characteristics such as reverberation are required so that the speaker can easily hear the story, and when a concert is held, an acoustic sound having a reflected sound or reverberation suitable for the concert. Characteristics are required.

When various acoustic characteristics are to be obtained by architectural means in one hall, for example, the reflection characteristics are changed by making a wall or the like a movable structure, resulting in a very large system. It will end up. Therefore,
In recent years, it has been proposed to install an acoustic device using electric means in the hall or the like and electrically generate various acoustic characteristics by the acoustic device.

When controlling the characteristics of such an acoustic space, a sound is emitted from a sound source, and an initial reflected sound is reflected back from a wall or the like within about 150 msec after reaching the listening position. Is known to be important, and in order to generate a desired initial reflected sound, a plurality of delay signals are generated by delaying the input signal by different predetermined delay amounts from each other. It is known that it is effective to use an FIR filter having a configuration of multiplying each of the delayed signals by each predetermined coefficient and adding them together (for example, see Japanese Patent Laid-Open No. 3-254298).

[0005]

However, there is a limit in increasing the number of stages of the delay means constituting the FIR filter, and in order to obtain the desired filter characteristics with the smallest possible number of stages, the delay amount and multiplication coefficient (these are generically referred to as Then, it may simply be referred to as a "coefficient").

In view of the above circumstances, the present invention is a filter coefficient capable of technically easily obtaining a coefficient to be set in an FIR filter and obtaining a coefficient that can maximize the use of the FIR filter. The purpose is to provide a generation method.

[0007]

The filter coefficient generating method of the present invention which achieves the above object, is a delaying means for generating a plurality of delay signals by delaying an input signal by different predetermined delay amounts, respectively. A filter for obtaining each of the predetermined delay amounts and each of the predetermined coefficients for constructing a filter having an arithmetic means for generating an output signal by multiplying each of a plurality of delay signals by each predetermined coefficient and adding them together. A coefficient generation method comprising: (1) obtaining an impulse response curve for obtaining a desired characteristic, (2) obtaining a large number of extreme values in which the signs of the differential coefficients of the impulse response curve change, and (3) From the large number of data of the impulse response curve corresponding to a large number of extreme values, respectively, extract the desired number of data in descending order of absolute value and the extracted data By extracting the delay amount corresponding to the Re respectively, each data and each of the delay amounts are extracted, respectively, characterized by using as the above predetermined coefficient, and each of the predetermined delay amount.

[0008]

According to the filter coefficient generating method of the present invention, a large number of extreme values of the impulse response curve are obtained, and a desired number of data and the time of occurrence of the data (in the order of increasing absolute value) are selected from the large number of extreme values. The amount of delay) is extracted, the coefficient can be easily obtained mechanically, and the reflected sound can be made as close to the audible natural reflection sound as possible within the range that can be simulated by the filter.

[0009]

EXAMPLES Examples of the present invention will be described below. FIG. 1 is a block diagram showing the overall configuration of an audio device to which the present invention is applied. Multiple microphones 1_1, ..., 1_p, ..., 1_q, ..., 1_n at various positions in the hall
Is installed, and one of them is a microphone 1_
1, ..., 1_p have their arrangement positions, directions, directional characteristics, etc. determined so as to mainly pick up the sound emitted from the sound source, and some microphones 1_p + 1 ,.
The arrangement position, orientation, directional characteristics, etc. of the microphones 1_q + 1, ..., 1_n are determined so as to mainly pick up the early reflection sound and the microphones 1_q + 1, ..., 1_n.

These plurality of microphones 1_1, ...,
The signals obtained at 1_p, ..., 1_q, ..., 1_n are input to the signal processing device 2. The internal configuration of the signal processing device 2 will be described later. The signal processing device 2 is controlled by a control signal output from the controller 3 in conformity with the MIDI standard. By operating the controller 3, the acoustic characteristics of this hall can be adjusted, for example, for lectures or concerts.

A control box 4 is installed on the sleeve of the stage installed in front of the hall. The control box 4 and the controller 3 are, for example, RS232C.
The same adjustment is performed by operating a push button switch of the control box 4 instead of directly operating the controller 3.

From the signal processing device 2 to a plurality of systems (3 in this case)
System) acoustic signals are output, and each acoustic signal is an amplifier 5_
1, 5_2, 5_3 are amplified respectively, and each speaker 6
Sound is emitted into the hall from _1, 6_2, 6_3. Here, the speaker 6_1 is, for example, two cluster speakers provided on both front sides of the hall, the speaker 6_2 is, for example, a total of six wall speakers embedded in both front side walls, and the speaker 6_3 is installed on, for example, a ceiling of a passenger seat. 28 ceiling speakers in total.

FIG. 2 is a block diagram showing an internal configuration of the signal processing device 2 shown as one block in FIG. The plurality of microphones 1_1, ..., 1_p, shown in FIG.
..., 1_q, ..., 1-n obtained by each signal A_1, ...,
, A_q, ..., A_n are input to the A / D converter 21 and have a predetermined sampling frequency, for example, 48 kH.
digital signals D_1, ..., sampled by z
, D_q, ..., D_n.

The digital signals D_1, ..., D_
, D_q, ..., D_n are input to the mixer 22. This mixer 22 receives a plurality of output signals D_ according to a command from a control means 25 that controls the signal processing device 2 based on a control signal input from the controller 3.
, ..., D_p, ..., D_q, ..., D_n are appropriately mixed to generate three mixed signals M_1, M_2, M_3. Three-system mixed signal M generated by the mixer 22
_1, M_2 and M_3 are input to the signal processing device 23. In this signal processing device 23, sound is emitted from a sound source,
An initial reflected sound generation unit 231 simulating the initial reflected sound from the time the sound reaches the listening position until approximately 150 msec elapses.
And a rear reverberant sound generator 232 that simulates a reflected sound (reverberant sound) subsequent to the initial reflected sound. Of these, the initial reflected sound generator 231 includes the first initial reflected sound generator 231_1 and Second initial reflected sound generation means 231_
It consists of two.

The first initial reflected sound generating means 231_1
Is an initial reflected sound generation unit 231_1_1, an initial reflected sound diffusion unit 231_1_2, and an equalizer 231_1_.
It consists of three. Second initial reflected sound generating means 23
1_2 is also configured similarly to the first initial reflected sound generation unit 231_1. Second initial reflected sound generation means 231_2
Is different from the first initial reflected sound generating means 231_1 only in the set parameters and the like, and therefore is not shown here, and the following description is based on the description of the first initial reflected sound generating means 231_1. Let them be represented.

The rear reverberation sound generating section 232 is composed of a rear reverberation sound generating means 232_1 and an equalizer 232_2. Each mixed signal M output from the mixer 22
_1, M_2 and M_3 are the first initial reflected sound generating means 231_1 and the second initial reflected sound generating means 231_, respectively.
2, input to the rear reverberation sound generation unit 232. These first
In the initial reflected sound generation unit 231_1, the second initial reflected sound generation unit 231_2, and the rear reverberation sound generation unit 232, the input mixed signals M_1, M_2, and M_3 are respectively based on the command from the control unit 25. , Speaker 6_1 (cluster speakers installed on both sides in front of the hall), speaker 6_2 (wall speakers embedded in front both side walls) and speaker 6_3 (ceiling speakers installed on the ceiling of the audience) shown in FIG. Suitable acoustic signals SD_ each representing the first early reflections
1, an acoustic signal SD_2 representing the second initial reflected sound, and an acoustic signal SD_3 representing the rear reverberation sound. These acoustic signals SD_1, SD_2, SD_3 are digital signals, and are input to the D / A converter 24 to be supplied to the D / A converter 24.
Each of the A converters 24 outputs an analog acoustic signal SA_1,
It is converted into SA_2 and SA_3. These analog acoustic signals SD_1, SD_2, SD_3 pass through the amplifiers 5_1, 5_2, 5_3 shown in FIG.
Sound is emitted into the hall by _1, 6_2, 6_3.

In FIG. 3, the initial reflected sound generating means 231_1_1 and the initial reflected sound diffusing means 2 shown in each block in FIG.
It is a circuit block diagram which showed the structure of 31_1_2. The initial reflected sound generation means 231_1_1 has a so-called FIR filter structure, receives the mixed signal M_1,
A plurality of (here, m) delay units 51_1, 51_2, ..., 51_m_1, 51_ that sequentially delay by each delay amount set based on a command from the control unit 25 (see FIG. 2)
The delay means 51 composed of m and the delay units 51_1 and 51_
2, ..., 51_m_1, 51_m, each output signal (each delay signal) is multiplied by each predetermined coefficient set based on a command from the control means 25.
2, ..., 52_m_1, 52_m, and a large number of unit adders 53_1, ..., 53_m-2, 53_m-1, which add together the signals output from the respective multipliers 52_1, 52_2, ..., 52_m_1, 52_m. It comprises an adder 53.

The initial reflected sound diffusion means 231_1_2
Has a so-called all-pass filter configuration,
An initial reflected sound having a sufficiently finer density than the initial reflected sound obtained by the initial reflected sound generation unit 231_1_1 is obtained. For example, each delay unit 5 of the delay unit 51
1_1, 51_2, ..., 51_m-1, 51_m, delay means 62 for delaying by a delay amount equal to or less than delay amount, multiplier 63 for multiplying the output of the adder 61 by a coefficient based on a command from the control means 25, and this multiplier It has a second adder 64 for adding the output of 63 and the output of the delay means 62, and a second multiplier 65 for multiplying the output of the delay means 62 by a coefficient based on a command from the control means 25. In the adder 61, the output of the initial reflected sound generating means 231_1_1 and the output of the second multiplier 65 are added. The second adder 6
4 is input to the equalizer 231_1_3 (see FIG. 2).

The equalizer 231_1_3 shown in FIG.
The equalizer 231_1_ is a filter that adjusts the frequency characteristics of the entire output signal from the initial reflected sound diffusing means 231_1_2 and various configurations are known.
The illustration and description of 3 are omitted. Initial reflected sound generation means 2
In 31_1_1, the mixed signal M_1 is input, and the delay means 51 generates m delay signals having different delay amounts, and the m delay signals are multiplied by the multipliers 52_1 and 52_.
, ..., 52_m_1, 52_m and the adder 53 perform sum-of-products calculation. Up to this point, it is the same as the conventional initial reflection sound generation means, but if the present invention is carried out and the number of filter coefficients is too small, it may be heard as an unnatural initial reflection sound such as a flutter echo. It may not be possible to obtain an early reflection sound with excellent performance.

The initial reflected sound diffusing means 231_1_2 is provided for the purpose of compensating for such a defect, and the delaying means 6 is provided.
2 and the feedback means (the adder 61 and the multiplier 65), the initial reflected sound diffusing means 231_
When an impulse signal is input to 1_2, the impulse signal is decomposed into a large number of theoretically infinite impulse sequences. Therefore, the initial reflected sound generating means 231_1_1
By passing the output signal of the first reflected sound through the initial reflected sound diffusing means 231_1_2.
The output signal of _1 is temporally diffused and converted into a signal carrying an initial reflected sound having a natural timbre. The output signal of the initial reflected sound diffusing means 231_1_2 is input to the equalizer 231_1_3 (see FIG. 2) and its frequency characteristic is adjusted as described above, and then passes through the D / A converter 24, the amplifier 5_1, and the speaker 6_1. Sound is emitted into the hall.

Here, the initial reflected sound generating means 231_1_
1, the delay units 51_1 and 51_ of the delay unit 51
2, ..., 51_m-1, 51_m, each delay amount and each multiplier 52_1, 52_2, ..., 52_m-
How to obtain each coefficient set to 1,52_m will be described. FIG. 4 is a diagram showing an impulse response curve described below and extracted coefficients.

Here, first, the impulse response curve of the characteristic to be obtained by the initial reflected sound generating means 231_1_1 is obtained. This impulse response curve may be obtained by measuring an actual sound field or by a computer simulation equivalent to this, but here, the case of measuring and obtaining the actual sound field will be described.

Here, the "difference" between the acoustic characteristics of the hall in which the acoustic device of this embodiment is installed and the desired acoustic characteristics.
To obtain and compensate for the acoustic characteristic corresponding to the "difference" by this acoustic characteristic device,
It is the impulse response curve of that "difference" that needs to be determined here. Therefore, a sound is emitted toward a reflection wall that can obtain desired acoustic characteristics, the sound reflected from the reflection wall is recorded by a microphone, and the impulse response is obtained based on the emitted sound and the recorded sound. . Since the calculation for obtaining the impulse response is usually performed by a digital calculation by a computer, the obtained signal is sampled at a sampling frequency such as 48 kHz, A / D converted, and taken into the computer.

Next, similar measurements are performed toward the wall material used for the hall where the acoustic device is to be installed, and based on these measurements, the acoustic device as shown in FIG. The impulse response curve of the acoustic characteristic of "difference" is obtained. The rod-shaped array shown in FIG. 4A represents time-series digital data, and the envelope thereof is an impulse response curve obtained.

When the impulse response curve for obtaining the desired characteristic is obtained in this manner, a large number of extreme values D _{i in which} the sign of the differential coefficient of this impulse response curve changes next.
And the time t _i when D _i occurs is determined. Since there are usually a great number of such extreme values, a desired number of data values are selected in descending order of absolute value | D _i | from a large number of data values of the impulse response curve corresponding to each of these many extreme values. The time t _i at which the data occurred is determined. Initial reflection sound generation means 231_1_1 shown in FIG.
The delay means 51 of the m-stage delay units 51_1, 51_2,
, 51_m−1, 51_m exist, m data values D _i and respective times t _i corresponding to the respective data values are obtained here (FIG. 4B). Each time t _i corresponds to each predetermined delay amount according to the present invention. Specifically, the respective times t _i are arranged in order from the earliest in time, and the respective differences between the times adjacent to each other are obtained, and the respective differences are respectively the delay units 51_1, 51_2, ..., 51_m−.
The delay amounts are set to 1,51_m. Also arranged in a manner m pieces of data values D _i the occurrence time sequence determined as described above, each data value of the m ordered thus D _i
Are multipliers 52_1, 52_2, ..., 52_m-1,5
Each predetermined coefficient is set to 2_m.

Here, as described above, a large number of extreme values of the impulse response curve are obtained, and a predetermined number is extracted from the large number of extreme values in descending order of absolute value of the data value. Each delay unit 51_1, 51_2,
..., each delay amount set to 51_m-1, 51_m, and each multiplier 52_1, 52_2, 52_m-1, 52
Each coefficient set to _m can be easily mechanically extracted, and the initial reflected sound generation means 231_1_1 is maximized by making the best use of the initial reflected sound generation means 231_1_1 composed of a predetermined number of steps (m steps). Thus, the impulse response curve is simulated with the maximum accuracy that can be simulated.

FIG. 5 shows the rear reverberation sound generator 23 shown in FIG.
3 is a circuit block diagram showing an example of a circuit configuration of a rear reverberation sound generating means 232_1 which constitutes No. The rear reverberation sound generation means 232_1 is the initial reflection sound diffusion means 23 shown in FIG.
Similar to 1_1_2, adders 71 and 74, delay means 7
2. All-pass filter 70 including multipliers 73 and 75
And a plurality of (here, z) comb filters 80_1, 80_2, ..., 8 following the all-pass filter 70.
0_z and a plurality of comb filters 80_1 and 80_
, ..., 80_z outputs are added to each other by an adder 84. Each comb filter 80_1, 80_
2, ..., 80_z are adders 81_1, 81_2 ,.
81_z and delay means 8 for delaying each predetermined delay amount
2_1, 82_2, ..., 82_z and delay means 82_
1, 82_2, ..., 82_z are composed of multipliers 83_1, 83_2 ,.
, ..., 83_z outputs are added to delay means 82_
1, 82_2, ..., 82_z. Comb filters 80_1, 80_2, ..., 80
_Z is a comb-tooth-shaped frequency characteristic having a plurality of band elimination bands as a whole, and a rear reverberation is obtained by passing a musical tone signal through a filter having such comb-shaped frequency characteristic. It is known.

In the rear reverberation sound generating means 232_1,
First, the all-pass filter 70 adds the first reverberant sound having no timbre change to the signal input to the rear reverberant sound generating means 232_1, and further outputs the output signal through the comb filters 80_1, 80_3, ..., 80_z. By adding each other, the second reverberation sound is added, whereby a more natural audible rear reverberation sound is generated.

The entire frequency characteristic of the signal output from the rear reverberation sound generating means 232_1 is adjusted by the equalizer 232_2 (see FIG. 2), and the D / A converter 24,
Sound is emitted from the ceiling of the hall seats by the speaker 6_3 via the amplifier 5_3.

[0030]

As described above, the filter coefficient generation method of the present invention is a method for obtaining coefficients to be set in an FIR filter, and it is possible to obtain a large number of extreme values of an impulse response curve and obtain a large number of these extreme values. Data in the order of increasing absolute value and the time of occurrence (delay amount)
Is used to extract the coefficient, the coefficient to be set in the FIR filter can be easily obtained mechanically.
It is possible to obtain a coefficient that can maximize the use of the R filter.

[Brief description of drawings]

FIG. 1 is a block diagram showing an overall configuration of an audio device to which the present invention is applied.

FIG. 2 is a block diagram showing an internal configuration of a signal processing device shown by one block in FIG.

FIG. 3 is a circuit block diagram showing a configuration of an initial reflected sound generation unit and an initial reflected sound diffusion unit.

FIG. 4 is a diagram showing an example of an impulse response curve and extracted coefficients.

FIG. 5 is a circuit block diagram showing a configuration example of a rear reverberation sound generation unit that constitutes a rear reverberation sound generation unit.

[Explanation of symbols]

1_1, ..., 1_p, ..., 1_q, ..., 1_n Microphone 2 Signal processing device 3 Controller 4 Control box 5_1, 5_2, 5_3 Amplifier 6_1, 6_1, 6_3 Speaker 23 Signal processing means 51 Delay means 51_1, 51_2, ..., 51_m- 1,51_m
Delay unit 52_1, 52_2, ..., 51_m-1, 52_m
Multiplier 53 Adder 61,64 Adder 63,65 Multiplier 62 Small delay means 70 All-pass filter 80_1, 80_2, ..., 80_z Comb filter 231 Initial reflected sound generation unit 231_1 First initial reflected sound generation means 231_1_1 Initial reflected sound Generating means 231_1_2 Initial reflected sound diffusing means 231_1_3 Equalizer 231_2 Second initial reflected sound generating means 232 Rear reverberation sound generating section 232_1 Rear reverberation sound generating means 232_2 Equalizer

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁵ Identification code Internal reference number FI Technical indication location H03H 17/02 P 7037-5J L 7037-5J 17/06 Z 7037-5J (72) Inventor Minami Hideaki, 3-7-13 Shin-Kitajima, Suminoe-ku, Osaka City, ROLAND Co., Ltd. (72) Inventor Tomotaka Hiramatsu 1-25-1, Nishishinjuku, Shinjuku-ku, Tokyo Taisei Construction Co., Ltd. (72) Inventor, Masahiro Tokyo 1-25-1 Nishishinjuku, Shinjuku-ku, Tokyo Within Taisei Construction Co., Ltd. (72) Inventor Mizuho Tanaka 1-25-1 Nishishinjuku, Shinjuku-ku, Tokyo Within Taisei Construction Co., Ltd. (72) Yukio Hamada Tokyo Shinjuku 1-25-1 Nishi-Shinjuku, Ward Taisei Construction Co., Ltd.

Claims (1)

[Claims] 1. A delay means for generating a plurality of delay signals by delaying an input signal by different predetermined delay amounts, and multiplying each of the plurality of delay signals by a predetermined coefficient and adding them together. In the filter coefficient generation method for obtaining each of the predetermined delay amounts and each of the predetermined coefficients for configuring a filter having an arithmetic means for generating an output signal by the method, an impulse response curve for obtaining a desired characteristic is obtained. , A large number of extreme values in which the signs of the differential coefficients of the impulse response curve change are obtained, and a desired number is selected from the large number of data of the impulse response curve corresponding to each of the large number of extreme values in descending order of absolute value. By extracting the data and the delay amount corresponding to each of the extracted data, each extracted data value and each delay amount can be calculated. And a filter coefficient generating method, wherein the filter coefficient generating method uses the predetermined coefficient and the predetermined delay amount, respectively.