JP2610991B2 - Directivity control type speaker system - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a speaker system in which a plurality of speakers are arranged, and more particularly, to a directivity control type speaker system in which a directivity pattern is controlled.

(Prior Art) In a speaker system arranged in a straight line, it is generally known that the directivity pattern becomes sharper than a single speaker unit over the entire reproduction band.
Since the directivity pattern becomes too sharp in the middle and high ranges, there is a demand for broadening the pattern. On the other hand, in the low frequency range, the directivity pattern is somewhat sharper than that of a single speaker unit, but is much broader than the middle and high frequency ranges.

As a conventional loudspeaker system for controlling the directivity pattern, there has been known a loudspeaker system as described in JP-A-63-9300. FIG. 5 is a block diagram showing the concept of this conventional directivity control type speaker system. In this speaker system, a plurality of speaker units 41, 42a to 4na, 42b to 4nb are linearly arranged,
Delay units 21 to 2n are provided in front of the speaker units 41, 42a to 4na and 42b to 4nb. Each speaker unit 42a
4na and 42b to 4nb are supplied with audio signals delayed by a time proportional to the distance from the center speaker unit 41. As a result, a directivity pattern different from that of the speaker system that does not use the delay units 21 to 2n is obtained.

(Problems to be Solved by the Invention) However, in the above-described speaker system, it is possible to sharpen the directivity pattern of a low frequency band by giving a certain delay time, and to provide a middle and high frequency band by giving another delay time. Although the directivity pattern can be broadened, it is not possible to set a delay time that sharpens the directivity pattern in the low frequency range and sharpens the directivity in the middle and high frequency ranges. That is, only by giving the delay time, it is only possible to sharpen the directivity pattern uniformly from low to middle and high frequencies or to broaden the directivity pattern uniformly.

Further, in the above-described speaker system, the directivity pattern can only be sharpened or broadened, but cannot meet the demand for other directivity patterns.

As described above, the conventional directivity control type speaker system has a problem that the directivity pattern lacks arbitrariness.

SUMMARY OF THE INVENTION In view of the above problems, an object of the present invention is to provide a directivity control type speaker capable of controlling directivity for each frequency from a low band to a mid-high band, and having a certain degree of arbitrariness in its directivity pattern. It is to provide a system.

(Means for Solving the Problems) In order to solve the above-mentioned object, a directivity control type speaker system according to the present invention includes a plurality of speaker units and a plurality of finite impulse for inputting an audio signal branched from one signal source. A response filter (hereinafter abbreviated as an FIR filter), wherein the FIR filter and the speaker unit are connected so that an output signal of each of the FIR filters is input to one or more speaker units of the plurality of speaker units. Are connected, and the filter characteristic obtained by using the nonlinear optimization method so that the directivity pattern of the plurality of speaker units as a whole approaches a desired directivity pattern is the FI characteristic.
This is set as the filter characteristic of the R filter.

Here, the plurality of speaker units can be arranged in a straight line, a plane, or a three-dimensional shape, and the plurality of speaker units are arranged symmetrically and arranged at symmetric positions. Can be connected to a common FIR filter.

Further, the desired directivity pattern may be constant over the entire reproduction band of the speaker unit.

(Operation) The filter characteristic of the FIR filter configured as described above is designed by a nonlinear optimization method. Since arbitrary characteristics can be easily set for each frequency in the FIR filter,
A filter characteristic is designed so as to generate a directivity pattern close to a desired directivity pattern for each frequency.

Then, such filter characteristics are set in the FIR filter, and the audio signal filtered by the FIR filter is supplied to the speaker unit, and emitted from the speaker unit as a sound wave. Sound waves output from the plurality of speaker units interfere with each other, and as a result, a directivity pattern close to a desired directivity pattern is generated for each frequency.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of a speaker system showing one embodiment of the present invention. In FIG. 1, 14 is a signal source,
16 is an input terminal, 70 to 74 are FIR filters, and 80 to 84 are amplifiers. 90, 91a to 94a and 91b to 94b are speaker units. Reference numerals 100, 101a to 109a, and 101b to 109b denote microphones, which are provided to measure the characteristics of the entire speaker system and the individual speaker units 90, 91a to 94a, and 91b to 94b. All microphones 100, 101
a to 109a, 101b to 109b are installed at a distance of 6 m from the central speaker unit 90, and are -90 ° with respect to the front axial direction.
It is installed at intervals of 10 ° from the direction of 90 ° to 90 °.
The audio signal generated from the signal source 14 branches into five via an input terminal 16 and is input to FIR filters 70 to 74, respectively. F
Output signals of the IR filters 70 to 74 are input to amplifiers 80 to 84. The output signal of the amplifier 80 is
Supplied to Output signal of amplifier 81 is speaker unit
The signals output from the amplifiers 82 to 84 are supplied to speaker units 92a to 94a and 92b to 94b. Each of the speaker units 90, 91a to 94a, 91b to 94b includes a circular diaphragm having a diameter of 95 mm, and is linearly arranged at intervals of 111 mm.

The audio signal generated from the signal source 14 and branched via the input terminal 16 is filtered by FIR filters 70 to 74,
After being amplified by the amplifiers 80 to 84, each speaker unit 90,
Supplied to 91a-94a, 91b-94b, the speaker unit 90,
It is emitted as sound waves from 91a to 94a and 91b to 94b. The sound waves emitted from each of the speaker units 90, 91a-94a, 91b-94b interfere with each other, resulting in a different directivity pattern than when only a single speaker unit is driven.
How the sound waves interfere at each frequency depends on the arrangement state of the speaker units 90, 91a to 94a, 91b to 94b, and the speaker units 90, 91a to 94a generated between the speaker units 90, 91a to 94a, 91b to 94b. , 91b-94b, and the amplitude of the speaker units 90, 91a-94a, 91b-94b. The phase and amplitude of the input signals of the speaker units 90, 91a to 94a and 91b to 94b are FI
It can be set freely for each frequency by the R filters 70-74.

Assuming that the arrangement of the speaker units 90, 91a to 94a and 91b to 94b is fixed, the directivity pattern of the entire speaker system is determined by how to set the characteristics of the FIR filters 70 to 74.

Now, if there is a desired directivity pattern to be realized for the speaker system connected as shown in FIG. 1, the directivity pattern of the speaker system should be closest to the desired directivity pattern. , FIR filters 70-74. Such a filter characteristic is called an optimum filter characteristic. Then, such optimum filter characteristics can be obtained by a non-linear optimization method. Although the non-linear optimization method is a well-known calculation method, an outline of a calculation procedure using the non-linear optimization method will be described below with reference to this embodiment.

Here, a desired directivity pattern for the speaker system is a target in the calculation of the nonlinear optimization method. Then, a target value Td as shown in FIG. 2 is determined. The target value Td is a numerical value representing a desired directivity pattern at intervals of 10 °. For example, Td = 0.400 in the 30 ° direction, and Td = 0.200 in the 40 ° direction. The ratio between the two is 2/1, which is a 6 dB difference when expressed in decibels. Here, as shown in FIG. 1, d is a suffix representing a direction at every 10 °. The target value Td is constant for the frequency f. Wd is a weight for weighting in the evaluation function of the nonlinear optimization method, and an appropriate value as shown in FIG. 2 can be used here. This weight Wd is also constant for the frequency f.

The degree of approximation of the directivity pattern of the speaker system to the target value Td can be expressed by the following equation (1).
(F) is the evaluation function in the nonlinear optimization method.

Here, f represents a frequency. Yd (f) is the output characteristic of the speaker system in the direction of d, that is, the observation point in the direction of d from the input terminal 16 common to the FIR filters 70 to 74, that is, the microphones 100, 101a to 109a, and 101b to 109.
This is the transfer function up to the installation point of b. And this Yd
(F) can be expressed by the following equation (2).

Here, Fn (f) is a transfer function of the n-th FIR filter 7n. Therefore, the relationship between Fn (f) and the amplitude characteristic Rn (f) and the phase characteristic θn (f) of the FIR filter is as shown in the following equations (3) and (4).

Re [Fn (f)] = Rn (f) · cos (θn (f)) (3) Im [Fn (f)] = Rn (f) · sin (θn (f)) (4) where Re [Fn (f)] represents the real part of Fn (f) as Im [Fn
(F)] means the imaginary part of Fn (f).

Further, Snd (f) in the equation (2) is the distance from the output terminal of the nth FIR filter 7n to the observation point in the d direction when only the speaker unit connected to the nth FIR filter 7n is driven. It is a voltage function. Snd (f) can be obtained by measurement or by calculation. In this example, an experimental apparatus having the configuration shown in FIG.
(F) was determined by measurement.

In order to use the nonlinear optimization method, it is necessary to define a variable vector, but here, the FIR filters 70 to 74
It is necessary to construct a variable vector having the filter characteristics as elements. Therefore, a variable vector as shown in the following equation (5)
v0 can be considered.

v0 = (R0 (f) R1 (f) ... R4 (f) θ0 (f) θ1
(F)... Θ4 (f) (5) However, when the nonlinear optimization algorithm is driven by this variable vector v0, the absolute value of Rn (f) may converge at a very large value. , So Rn
In (f), it is necessary to set an upper limit value. The reason why it is necessary to set an upper limit for the amplitude characteristic Rn (f) of the filter characteristics is as follows. That is, when designing the filter characteristics of the FIR filter, the limitation on the amplitude coming from the FIR filter element as an electric component must be satisfied. It is obvious to those skilled in the art to design a filter characteristic by setting an upper limit value for the amplitude characteristic.

Therefore, in order to give an upper limit value Rmax (f) to the amplitude characteristic Rn (f), this is also a well-known calculation procedure of the non-linear optimization method. .

Rn (f) = Rmax (f) · sin (φn (F)) (6) As a variable vector, a variable vector v1 represented by the following equation (7) can be considered.

v1 = (φ0 (f) φ1 (f) ... φ4 (f) θ0
(F) θ1 (f)... Θ4 (f)) (7) In this case, no matter how much the absolute value of φn (f) converges to a large value, the amplitude characteristic Rn (f) is obtained by the equation (6). ), The amplitude characteristic Rn (f) does not exceed Rmax (f).

Since the variable vector v1 and the evaluation function E (f) can be related by the above equations (1) to (4), (6) and (7), the algorithm for nonlinear optimization is driven. Can be.

Here, Davidon-Fletcher-Powel, a well-known algorithm as one of the nonlinear optimization algorithms, is used.
Calculated by l-method. In the calculation of the nonlinear optimization, the optimum filter characteristics were obtained for each frequency at intervals of about 20 Hz from 20 Hz to 2500 Hz.

FIG. 3 shows the optimum filter characteristics obtained in this manner. In FIG. 3, (R0) to (R4) are amplitude characteristics of optimal filter characteristics to be set in the FIR filters 70 to 74, respectively. (Θ0) to (θ4) are the phase characteristics of the optimum filter characteristics to be set in the FIR filters 70 to 74, respectively. In FIG. 3 (R0), what is indicated by a dotted line is Rmax (f) of the equation (6) appropriately determined for calculation.
It is. Although not shown in FIGS. 3 (R1) to (R4), the amplitude characteristics of all the filters are calculated in FIG.
The same Rmax (f) as in (0) was set. As shown in FIG. 3, the optimum filter characteristic is completely different from the characteristic of the delay means used in the prior art, and both the amplitude characteristic and the phase characteristic are very complicated.

FIG. 4 shows the results obtained by setting the optimum filter characteristics shown in FIG. 3 to the FIR filters 70 to 74 of the experimental apparatus and measuring the directivity pattern. In FIG. 4, (f1) to (f6) show the measurement results of the directivity patterns at frequencies of 80 Hz, 125 Hz, 250 Hz, 500 Hz, 1000 Hz, and 1600 Hz, respectively.
The directivity pattern shown by the solid line is a measured value obtained from the experimental apparatus. For reference, the target value Td shown in FIG. 2 is plotted with a cross. FIG. 4 shows only the directivity pattern in the range of 0 ° to 90 ° with respect to the front axial direction. However, from the symmetry of the speaker system of FIG.
In the range of -90 °, a directivity pattern almost symmetrical to FIG. 4 should be obtained.

As can be seen from FIG. 4, the measured value closely approximates the target value Td. This suggests that it can be obtained from various directivity patterns depending on the setting of the target value Td. The experimental results ranged from low to mid-high frequencies (80
Hz to 1600 Hz), indicating a substantially uniform directivity pattern.

Further, the directivity pattern can be sharpened in a low frequency range, which has been considered particularly difficult. 4th
The dotted line in the figure (f2) is the directivity pattern of the speaker system at a frequency of 125 Hz when the FIR filters 70 to 74 of this experimental device are bypassed. In the directivity pattern indicated by the dotted line, the direction forming the front axis (direction of d = 0) and the direction forming the front axis 90 ° (direction of d = 9) are as follows.
There is only a level difference of 2-3dB. On the other hand, by using the FIR filters 70 to 74 with the optimum filter characteristics set, a level difference of 26 dB could be generated. On the other hand, FIG. 2 (B) of the document showing the prior art, Japanese Patent Application Laid-Open No. 63-9300, does not specify the specific frequency, but the directivity pattern depends on the presence or absence of the delay means in the low frequency range. Is shown. Even without the delay means, the directivity pattern had a level difference of as much as 8 dB between the front axis direction and the 90 ° direction, but even with the delay means, only a 16 dB level difference was generated. Also in comparison with this conventional technique, it can be seen that the present invention is effective for sharpening the directivity pattern in the low frequency range.

In the present embodiment, the loudspeaker system is configured symmetrically, but it is not always necessary to have such a symmetrical configuration.

Further, in the present embodiment, the speaker units 90, 91a to 94a, 91
Although b to 94b were arranged linearly, even if they were arranged in a plane and three-dimensionally, in order to obtain a filter characteristic to obtain a directivity pattern closest to a desired directivity pattern, It is obvious that the nonlinear optimization method is effective.

Furthermore, this embodiment has a constant target
Although the optimum filter characteristic is obtained by Td, a different target value Td may be provided for each frequency.

(Effects of the Invention) As described above, according to the present invention, an optimum filter characteristic is obtained by a non-linear optimization method, and a speaker unit is driven by an FIR filter in which the optimum filter characteristic is set, so that the low- to mid-high range is obtained. The directivity can be controlled for each frequency, and the directivity pattern can have some degree of arbitrariness.

[Brief description of the drawings]

FIG. 1 is a block diagram showing one embodiment of the present invention.
FIG. 2 is a diagram showing target values and the like for calculation by the nonlinear optimization method. FIG. 3 is a characteristic diagram showing optimum filter characteristics. FIG. 4 is a characteristic diagram showing a directivity pattern. FIG. 5 is a block diagram showing the prior art. 14 signal source, 16 input terminal, 70-74 FIR filter, 80-84 amplifier, 90, 91a-94a, 91b-94b speaker unit, 100, 101a-109a, 101b-109b …… a microphone.

Continuation of the front page Examiner Yasushi Takahashi (56) References JP-A-63-9300 (JP, A) JP-A-63-234699 (JP, A)

Claims (5)

(57) [Claims] 1. A loudspeaker unit comprising: a plurality of speaker units; and a plurality of finite impulse response filters for inputting audio signals branched from one signal source, wherein an output signal of each of the finite impulse response filters is one of the plurality of speaker units. The finite impulse response filter and the speaker unit are connected so as to be input to one or more speaker units, and nonlinear optimization is performed so that the directivity pattern of the plurality of speaker units as a whole approaches a desired directivity pattern. A directional control type loudspeaker system, wherein a filter characteristic obtained by using a technique is set as a filter characteristic of the finite impulse response filter. 2. The directivity control type speaker system according to claim 1, wherein said plurality of speaker units are linearly arranged. 3. The directivity control type speaker system according to claim 1, wherein said plurality of speaker units are arranged in a plane or a three-dimensional shape. 4. The directivity control type loudspeaker system according to claim 1, wherein said plurality of loudspeaker units are symmetrically arranged, and said two loudspeaker units arranged at symmetric positions are a common finite impulse response filter. Directivity control type speaker system connected to. 5. A directivity control type speaker system according to claim 1, wherein said desired directivity pattern is constant over the entire reproduction band of said speaker unit.