JP4292327B2 - Speaker array and microphone array - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a speaker array and a microphone array for improving directivity in the low-pitched sound zone of a speaker array or a microphone array without lengthening array length, and for preventing the increase of the level of a side-lobe. <P>SOLUTION: Filter coefficients set in one-dimensional digital filters connected to each of a plurality of speakers forming a speaker array give amplitude characteristics to a two-dimensional digital filter such that when the frequency characteristics of a two-dimensional digital filter formed of one-dimensional digital filters is represented by a two-dimensional frequency plane, a plurality of ripples are provided in a stop band in a cross-section in the spatial frequency direction and also the amplitude of ripples in a non-physical area out of a plurality of ripples is larger than the amplitude of ripples in a physical area. <P>COPYRIGHT: (C)2007,JPO&amp;INPIT

Description

  The present invention relates to a technique for improving the directivity of a speaker array and a microphone array, and more particularly to a technique for improving the directivity in a low sound range.

A sound field is formed only in a specific direction by using a speaker array or microphone array formed by arranging a plurality of transducers such as speakers and microphones linearly at a predetermined interval, or only sound coming from a specific direction is received. Techniques for picking up sound are widely used.
By the way, in this kind of speaker array or microphone array, it is desirable that the same directivity characteristic can be realized over a wide band from a high sound range to a low sound range. However, the directivity characteristics in the low sound range improve as the array length of the speaker array or microphone array (the value obtained by multiplying the number of transducers by the arrangement interval of the transducers) increases (see Non-Patent Document 1). In order to ensure sufficient directivity in the sound range, there has been a problem that the apparatus size of the speaker array and microphone array becomes large.
Thus, various techniques for solving the above-described problems have been proposed in the past, and an example thereof is the technique disclosed in Non-Patent Document 2. This Non-Patent Document 2 describes a Dolph whose cross section in the spatial frequency direction in the two-dimensional frequency plane has a ripple characteristic such as a stop band in the amplitude characteristics of a digital filter connected to each transducer constituting a speaker array or a microphone array. -A technology that expands the band that can give the same directivity to the low frequency range by setting the filter coefficient of each digital filter so that it becomes the amplitude characteristic (or its approximate characteristic) of the Chebyshev filter .
Toshiro Oga, Yoshio Yamazaki, Yutaka Kaneda “Acoustic System and Digital Signal Processing” The Institute of Electronics, Information and Communication Engineers 1993-05 p176-186 Matsumoto Yasushi, Nishikawa Kiyoshi "Design Method of Directional Array Speaker with Constant Sidelobe Amount" IEICE Technical Report 2004-74 p13-18

However, the ripple with the ripple characteristic such as the stop band is a non-physical region (region of | f ₂ |> ρ | f ₁ | in the two-dimensional frequency plane. However, ρ = D / cT, T: sampling interval, D: speaker Since there is usually also an area other than the interval, c: sound speed, f ₁ : normalized time frequency, f ₂ : normalized spatial frequency), a large amplitude is generated in the ripple such as the stop area in order to improve directivity in the low sound range. If given, there is a problem that the amplitude level of the side lobe, which is a directional characteristic that is originally unnecessary, increases.
The present invention has been made in view of the above problems, and improves the directivity of the speaker array and microphone array in the low frequency range without increasing the array length and avoids an increase in the amplitude level of the side lobe. The aim is to provide technology that makes it possible.

  In order to solve the above problems, the present invention is provided with a plurality of speakers arranged linearly at a predetermined interval, and corresponding to each of the plurality of speakers, and a predetermined filter coefficient is preset. And a one-dimensional digital filter that performs a filtering process on the input voice data according to the filter coefficient and outputs the same. The voice data obtained by digitally converting the input voice signal A speaker array for supplying audio signals obtained by subjecting audio data output from each one-dimensional digital filter to analog conversion to the corresponding speakers while supplying the digital filters, and outputting audio corresponding to the audio signals The filter coefficients set in the respective one-dimensional digital filters are formed by the respective one-dimensional digital filters. When the frequency characteristics of the two-dimensional digital filter to be expressed are expressed in a two-dimensional frequency plane, the cross section in the spatial frequency direction has a plurality of ripples in the stop band, and within the non-physical region of the plurality of ripples There is provided a loudspeaker array characterized by a filter coefficient that gives the two-dimensional digital filter an amplitude characteristic in which the amplitude of the ripple is larger than the amplitude of the ripple in the physical region.

  In order to solve the above-described problem, the present invention provides a plurality of microphones arranged linearly at a predetermined interval and each of the plurality of microphones, and a predetermined filter coefficient is set in advance. And a one-dimensional digital filter that performs output processing on the input audio data according to the filter coefficient, and performs digital conversion on the audio signal output from each of the plurality of microphones. In the microphone array that supplies the obtained audio data to the corresponding one-dimensional digital filter and outputs the sum signal of the audio data output from each one-dimensional digital filter, the microphone data is set to each one-dimensional digital filter. The filter coefficient is the value of the two-dimensional digital filter formed by each one-dimensional digital filter. When the wave number characteristic is represented by a two-dimensional frequency plane, the cross section in the spatial frequency direction has a plurality of ripples in the stop band, and the amplitude of the ripples in the non-physical region among the plurality of ripples is the physical region There is provided a microphone array characterized in that it is a filter coefficient that gives the two-dimensional digital filter an amplitude characteristic that is larger than the amplitude of the ripples therein.

  According to the present invention, it is possible to improve the directivity in the low sound range of the speaker array and the microphone array without increasing the array length, and to avoid an increase in the level of the side lobe.

Hereinafter, the best mode for carrying out the present invention will be described with reference to the drawings.
(A. First embodiment)
(A-1: Configuration)
FIG. 1 is a block diagram illustrating a configuration example of a speaker array 100 according to the first embodiment of the present invention. As shown in FIG. 1, the speaker array 100 includes transducers (speakers in this embodiment) 110-1 and 110-2 that are linearly arranged at predetermined intervals (in this embodiment, a constant interval D). ... 110-n and the same number of one-dimensional digital filters 120-1, 120-2 ... 120-n as the speakers.

In the speaker array 100 of FIG. 1, an audio signal (analog signal) supplied from an external sound source (not shown) is converted into digital data (hereinafter referred to as audio data) by an A / D converter (not shown). Audio data is supplied to each of the one-dimensional digital filters 120-i (i: natural numbers of 1 to n: the same applies hereinafter).
In each of the one-dimensional digital filters 120-i in FIG. 1, filter coefficients characteristic of the speaker array according to the present invention are set in advance. Each of these one-dimensional digital filters 120-i performs a filtering process corresponding to the filter coefficient on the audio data delivered from the A / D converter and outputs the result.
The audio data output from each one-dimensional digital filter 120-i is converted into an audio signal by a D / A converter (not shown), and the speaker 110- corresponding to the one-dimensional digital filter 120-i. i. As a result, a sound corresponding to the sound signal supplied from the D / A converter is emitted from each of the speakers 110-i.
The above is the configuration of the speaker array 100.

As described above, the hardware configuration of the speaker array 100 according to the present embodiment is not different from the hardware configuration of the conventional speaker array. However, in the speaker array 100 according to the present embodiment, filter coefficients characteristic of the speaker array according to the present invention are preset in each of the one-dimensional digital filters 120-i. A characteristic amplitude characteristic of the speaker array according to the present invention is given to the formed two-dimensional digital filter, and a characteristic directivity characteristic of the speaker array according to the present invention is realized by the amplitude characteristic. .
Hereinafter, the amplitude characteristic of the two-dimensional digital filter formed by the one-dimensional digital filter 120-i and the directivity characteristic realized by the amplitude characteristic will be described with reference to the drawings. In the following, it is assumed that each of the speakers 110-i has an ideal characteristic (that is, a characteristic that the directivity characteristic does not depend on the frequency of the output sound). In the following, it is assumed that the speaker arrangement interval D = 0.068 [m], the sampling frequency f _s = 6087 [Hz], the number of FIR taps = 61, and n (the number of speakers) = 15.

(A-2: Amplitude characteristics and directivity characteristics of a two-dimensional digital filter)
2 to 6 are diagrams showing the amplitude characteristics of the two-dimensional digital filter of the speaker array 100 and the directivity characteristics realized by the amplitude characteristics.
FIG. 2 is a diagram showing the amplitude characteristics of the two-dimensional digital filter formed by the one-dimensional digital filter 120-i on a two-dimensional frequency plane. FIG. 3 shows a part of the amplitude characteristics shown in FIG. FIG. 6 is a diagram showing an equiamplitude characteristic diagram in which the normalized time frequency f ₁ is in the range of 0 to 0.5 and the normalized spatial frequency f ₂ is in the range of 0 to 0.5. The normalized time frequency is a value obtained by normalizing the time frequency by the reciprocal of the time sampling interval, and the normalized spatial frequency is obtained by normalizing the spatial frequency by the reciprocal of the speaker arrangement interval D. Value.
As is apparent from reference to FIGS. 2 and 3, in the speaker array 100 according to the present embodiment, the normalized time-frequency f ₁ is lower region of the stop band (e.g., f ₁ is 0 to 0.1 area ) Is provided with a plurality of ripples, and a ripple in the non-physical area among the plurality of ripples is given a large amplitude (in this embodiment, “1”). Is suppressed to be lower than the ripple in the non-physical region. As is clear from FIG. 2 and FIG. 3, since ripples in the non-physical region are equal ripples having substantially the same amplitude, the amplitude characteristics shown in FIG. 2 and FIG. Called a characteristic.

FIG. 4 shows the amplitude characteristics shown in FIG. 2 in the plane including each speaker 110-i and the observation point of the sound output from the speaker array 100, and the direction perpendicular to the arrangement direction of the speakers 110-i is 0 degree. It is the figure shown as a frequency characteristic about the angle (angle (phi) of FIG. 1) of the observation point direction seen from the center of the speaker arrangement | sequence in the case. FIG. 4 shows frequency characteristics for φ = 0 °, 24 °, 40 °, 70 °, and 90 °.
As is apparent from FIG. 4, the amplitude level of the acoustic beam output from the speaker array 100 according to the present embodiment is about φ = 24 ° when the frequency is higher than a certain value. Is approximately 6 [dB] lower than the φ = 0 ° direction, and the φ = 40 °, 70 °, and 90 ° directions have an amplitude level of about 20 compared to the φ = 0 ° direction. [DB] It turns out that it has fallen.
FIG. 5 is a diagram showing the amplitude characteristics shown in FIG. 2 as directional characteristics at several frequencies (202.10742 Hz, 404.214484 Hz, 499.32422 Hz, 998.664844 Hz, 1997.2969 Hz and 2995.9453 Hz). is there.
As apparent from FIGS. 4 and 5, according to the speaker array 100 of the present embodiment, the side lobe level is maintained while keeping the main lobe width of the acoustic beam constant for frequencies above a certain value. It can be seen that it can be kept substantially constant (-20 dB in this case).

FIG. 6 shows a comparison between the main lobe width (φ = 0 ° direction) of the speaker array 100 according to the present embodiment and a conventional rectangular in-phase driving (in-phase driving by a signal subjected to rectangular window processing). It is the figure which plotted the angle which represents the width | variety of the area | region where the amplitude of an acoustic beam falls 6 [dB] for every frequency. As apparent from FIG. 6, according to the speaker array 100 of the present embodiment, the main lobe width in the low sound range can be made narrower than the conventional rectangular in-phase drive speaker array. You can see that
Further, as apparent from FIG. 6, in the speaker array 100 according to the present embodiment, for example, when a certain value (for example, 80 °) is defined as the main lobe width, the main lobe width is used. The lower limit of the frequency of the acoustic beam that can be output (that is, the lower end band f _L of the directional speaker array: see Non-Patent Document 2 for details) is reduced as compared with the case where conventional rectangular in-phase driving is performed. You can see that More specifically, when the ripple amplitude of the non-physical region is set to “1” (FIG. 6: gain 1), the lower end of the band of the speaker array is 20 in comparison with the conventional rectangular in-phase drive. When the ripple amplitude of the non-physical region is set to “2” (FIG. 6: gain 2), the frequency is reduced by 32.8%. That is, according to the speaker array 100 according to the present embodiment, it is possible to reduce the lower end of the band (that is, to improve the directivity in the low sound range) as compared with the conventional rectangular in-phase driving speaker array. Yes.

As described above, in the loudspeaker array 100 according to the present embodiment, when the frequency characteristics are represented in the two-dimensional frequency plane in the two-dimensional digital filter, in the cross section in the spatial frequency direction, Amplitude characteristics having a plurality of ripples, and the amplitude of the ripples in the non-physical region among the plurality of ripples is larger than the amplitude of the ripples in the physical region (in this embodiment, the blocking characteristic shown in FIG. 2) (Ripple characteristics such as 2 stages), it is possible to improve the directivity in the low frequency range of the speaker array and microphone array without increasing the array length, and to avoid an increase in the level of the side lobe. It has become.
Next, a description will be given of the design of a two-dimensional digital filter (that is, calculation of filter coefficients to be set for each one-dimensional digital filter 120-i) for realizing a two-step stopband equal ripple characteristic as shown in FIG.

(A-3: Two-dimensional digital filter design)
In Non-Patent Document 2 described above, when the amplitude characteristics of a two-dimensional digital filter formed by a group of one-dimensional digital filters connected to each speaker are viewed on a two-dimensional frequency plane, the output of the speaker array is sufficient. It is disclosed that the frequency characteristic when observed at a distant observation point is an amplitude characteristic distributed on a straight line represented by the following Equation 1 in a two-dimensional frequency plane.
(Expression 1) f ₂ = f ₁ · D · sin (φ) / (c · T)
In Equation 1, f ₁ is a normalized time frequency, f ₂ is a normalized spatial frequency, D is a transducer interval, T is a time sampling period, and c is a sound velocity.

Therefore, the directivity characteristic of the speaker array at a certain non-standardized time frequency f is on a straight line defined by the standardized time frequency f ₁ = f · T corresponding to the non-standardized time frequency f in the two-dimensional frequency plane. It can be said that they are distributed according to the relationship represented by the following formula 2.
(Expression 2) φ = sin ⁻¹ {(f ₂ · c · T) / (f ₁ · D)}

That is, if the two-dimensional digital filter can be designed so that the desired directivity at each non-standardized time frequency f is distributed on the straight line f ₁ = f · T in the relationship shown by the above equation 2, A desired directivity characteristic can be obtained. Non-Patent Document 2, the normalized spatial frequency direction of the two-dimensional frequency plane as described above (i.e., f ₂ direction) side by side one-dimensional filter characteristic section of setting the target characteristic of the two-dimensional digital filter, its A method for obtaining FIR filter coefficients by applying a two-dimensional Fourier series approximation to a target characteristic is disclosed.

More specifically, Non-Patent Document 2 describes, as design conditions for a speaker array formed of (N ₂ +1) speakers, an acoustic beam center φ ₀ and beam end angles (φ _{s +} , φ _s− ) and A design procedure for a two-dimensional digital filter in the case where a size (amplitude) δ of an equal ripple side lobe is given is disclosed. In the following, φ ₀ = 0 °, φ _{s +} = φ _s , φ _s− = −φ _s (that is, the acoustic beam is symmetric with respect to its center (φ ₀ = 0 °) axis). To do.

In the design procedure disclosed in Non-Patent Document 2, first, as shown in FIG. 7B, a two-dimensional frequency plane is represented by M ₁ (this embodiment) in the range of f ₁ = −0.5 to 0.5. in, _{M 1} is assumed to be an even number) is divided, the Dolph-Chebyshev characteristic cross-section at each frequency _{f 1 = k 1 / M 1} (k 1 = -M 1 / 2~M 1/2 integer) Design and target fan filter characteristics by juxtaposing them in the following procedure. The following description is based on the premise of a region where f ₁ ≧ 0.

Specifically, first, the characteristics of a Dolph-Chebyshev filter of order N ₂ and stop band ripple size δ are designed, and the stop band end frequency f _st is a straight line with φ = φ _s (ie, f ₂ = f ₁ · D · sin (φ _s ) / (c · T)), the frequency f ₁ is obtained. Next, at the cross-sectional position of f ₁ ≧ f _l, the cross-sectional characteristics at f ₁ = f _l are shown in FIG. 7A so that the stop band edge is positioned on the straight line of φ = φ _s. Arranged in _two directions.
On the other hand, for the cross-sectional position of f ₁ <f ₁ , as shown in FIG. 8A, a Dolph-Chebyshev filter in which the stop band ripple is gradually increased from δ to a predetermined allowable value δ _L as shown in FIG. Put the characteristic (f2 = −0.5 to 0.5). However, the stop band ripple is determined so that the stop band end f _st is located on the straight line φ = φ _s in any cross section. Then, as shown in FIG. 8 (b), the value of the frequency _{f 1} the characteristics of the stop band ripple [delta] _L is first placed as _{f u,} the cross section of the f 1 _{<f u} in _f 1 = _{f u} Put the same property as the property. Note that f _L in FIG. 8B is the lower end of the band of the speaker array, and is a value determined by the following Equation 3.
(Number _{3) f L = c · T} · f c / (Dsin (φ s))
However, f _c in Equation 3, the amplitude half the frequency of the Dolph-Chebyshev filter characteristic of the stop band ripple δL shown in Figure 8 (a).
Thereafter, a filter coefficient to be set for each one-dimensional digital filter is calculated by performing a two-dimensional inverse discrete Fourier transform on the target amplitude characteristic of the fan filter set in this way.

In contrast, in the two-dimensional digital filter design of the speaker array 100 according to this embodiment, the M ₁ divided 2-dimensional frequency plane, as shown in FIG. _{9 (b), f 1 ≧} f l in the stopband place one-dimensional filter having a small ripple across the while keep the cross-sectional characteristics, only a one-dimensional filter of a large ripple in cross section f 1 _{<f l} in the non-physical area (hatched portion in the figure). Two amplitude characteristics shown in FIG. 9A are amplitude characteristics of a one-dimensional filter placed on each cross section. As can be seen from the comparison of the two amplitude characteristics, by increasing the ripple in the non-physical region, the frequency range occupied by the ripple is widened, and conversely the passband is narrowed. Therefore, in the design of the two-dimensional digital filter according to the present embodiment, it is placed at the cross-sectional position of the time frequency in the lower sound range until the ripple amplitude in the non-physical region reaches a predetermined maximum value.

  In this embodiment, in order to design a one-dimensional filter having two-step stopband equiripple characteristics as shown in FIG. 9A, a program for designing a filter according to the Parks & McClellan equiripple filter design algorithm is used. . Here, the equiripple filter design algorithm of Parks & McClellan is an algorithm for designing a filter so that a desired frequency response and an actual frequency response are optimized by using a Remez exchange algorithm and a weighted Chebyshev approximation theory. A filter designed according to this algorithm is sometimes referred to as a minimax filter because it is optimal in minimizing the maximum error between the desired frequency response and the actual frequency response. A filter designed according to this algorithm is also known as an equiripple filter because it exhibits equiripple in its frequency response. In this embodiment, the case where the Parks & McClellan equiripple filter design algorithm is used for the design of a one-dimensional filter with two-step stopband equiripple characteristics will be described. However, other FIR filter design algorithms may be used. Of course.

  FIG. 10 is a diagram showing parameters given to the program and characteristics of the design results. As shown in FIG. 10, in the present embodiment, three approximate bands (pass band, stop band 1, stop band 2) are set, and the target amplitude (1 for each approximate band) is set as a parameter for defining each approximate band. , 0, 0), error ripple (respectively δ1≈0, δ2 = δ, δ3 = δn) and weight (respectively w1, w2, w3), and repeated approximation under the condition of δ1w1 = δ2w2 = δ3w3 By defining the filter coefficient, a one-dimensional filter is designed.

FIG. 11 is a diagram in which a one-dimensional filter designed in accordance with Parks &McClellan's equiripple filter design algorithm and a design example of a one-dimensional filter having a Dolph-Chebyshev characteristic are shown. As is apparent from FIG. 11, in the former one-dimensional filter, it is understood that the width of the pass band is narrower than that of the latter by increasing the ripple in the stop band 2. Thus, the effect of narrowing the width of the pass band becomes more remarkable as the number of ripples in the non-physical area occupies the total number of ripples in the stop area. Theoretically, for ripples in the non-physical region, it is possible to set the amplitude as much as possible, but in practice it is necessary to set the upper limit appropriately, for example, as the upper limit “1” (that is, a value equal to the amplitude of the pass band), “2” (a value twice the amplitude of the pass band), or the like may be set.
By performing a two-dimensional inverse discrete Fourier transform on the target amplitude characteristic of the two-dimensional digital filter designed in this way, a filter coefficient to be set for each of the one-dimensional digital filters forming the two-dimensional digital filter is calculated. Is done. By setting the filter coefficient thus calculated in each one-dimensional digital filter 120-i, the amplitude characteristics shown in FIG. 2 are given to the two-dimensional digital filter formed by the one-dimensional digital filter. Become.

(A-4: Effects of the first embodiment)
As described above, the characteristics in the physical region directly affect the directivity, while the characteristics in the non-physical region do not directly affect the directivity. Therefore, in the speaker array 100 according to the present embodiment, By using a one-dimensional filter having a two-stage stopband ripple characteristic, the width of the main lobe can be reduced as the final filter coefficient characteristic while keeping the sidelobe level low particularly in the low band. Is possible.

In addition, according to the present embodiment, by adjusting the one-dimensional filter optimally according to f ₁ , the width of the main lobe is kept constant while keeping the influence of the side lobe low even in a band lower than the conventional band. Is possible. As described above, since the width of the main lobe depends on the number of ripples in the non-physical region and the amplitude thereof, the amplitude set for the ripple in the non-physical region so that the necessary directivity can be obtained according to f _1. By adjusting the number, it becomes possible to make the width of the main lobe constant in a lower band than in the prior art.

Moreover, a relatively high region time-frequency (e.g., the region in the non-patent document 2 defined by f _l ≦ f ₁₎ for, without increasing the amplitude of the ripples in the non-physical area the width of the main lobe sufficient Since it can be narrowed, for example, the Dolph-Chebyshev characteristic disclosed in Non-Patent Document 2 may be used in place of the ripple characteristic of two stages of the stop band. In addition, for such a region, if the width of the main lobe is set so as not to depend on the time frequency as disclosed in Non-Patent Document 2, the conventional method is combined with the improvement in characteristics in the low region according to the present embodiment. It becomes possible to obtain directivity characteristics that do not depend on time frequency in a wider band.

(B. Second Embodiment)
Next, a microphone array 200 according to the second embodiment of the present invention will be described.
FIG. 12 is a diagram illustrating a configuration example of a microphone array 200 according to the second embodiment of the present invention. As apparent from a comparison between FIG. 12 and FIG. 1, the configuration of the microphone array 200 is different from the configuration of the speaker array 100 in that the speaker 110-i (i: natural number of 1 to n) is replaced. The microphone 210-i (i: a natural number from 1 to n) that outputs an audio signal corresponding to the collected audio is provided.
In the microphone array 200, the audio signal output from the microphone 210-i is converted into audio data by an A / D converter (not shown) and input to the one-dimensional digital filter 120-i. Then, the filter processing described above is performed by each one-dimensional digital filter 120-i, and the filtered audio data output from each one-dimensional digital filter is added by an adder (not shown), and a sum as a result of the addition is added. A signal is output.

  Now, in the microphone array, the time-frequency characteristic of the plane wave arriving from the angle φ direction shown in FIG. 12 is one-dimensionally connected to each microphone (microphone 210-i in the present embodiment) constituting the microphone array. When the amplitude characteristics of the digital filter group are viewed on a two-dimensional frequency plane, it is generally known that the digital filter group is distributed on the straight line represented by Equation 2 described above. Therefore, by setting the filter coefficient described in the first embodiment to each of the one-dimensional digital filters 120-i, the directivity characteristics of the microphone array 200 are the same as those in the first embodiment. It is possible to obtain an effect (that is, an effect of improving the directivity in the low sound range of the microphone array and avoiding an increase in the sidelobe level without increasing the array length).

(C. Modification)
Although the embodiment of the present invention has been described above, it is needless to say that the embodiment described above may be modified as described below.
(1) In the above-described embodiment, the case where an acoustic beam symmetric with respect to the central axis of the passband is described, but an asymmetric acoustic beam with respect to the symmetric axis can also be formed.

(2) In the above-described embodiment, the case where the speaker 110-i and the microphone 210-i have ideal characteristics has been described. However, since transducers such as speakers and microphones generally have directivity characteristics according to frequency, amplitude characteristics to be applied to the two-dimensional digital filter (that is, set to each one-dimensional digital filter 120-i). The power filter coefficient) may be determined in consideration of the directivity characteristics for each frequency of the transducer. This can be achieved by applying a method similar to the method disclosed in “Kiyoshi Nishikawa, Takaya Osaki“ Directive Array Speaker Using Two-Dimensional Digital Filter ”(1995)”, for example.

(3) In the above-described embodiment, an equal ripple having a large amplitude is provided in the non-physical region in the stop band, and the physical region has an amplitude smaller than the ripple in the non-physical region (in the above-described embodiment, “δ = A case has been described in which a two-dimensional digital filter is provided with a ripple characteristic having a ripple characteristic such as a two-step stopband in which a ripple having 0.1 ″) is provided. However, the ripples in the non-physical area are not necessarily equal ripples. For example, as shown in FIG. 13, a ripple having a larger amplitude than the passband in the stopband (stopband 2 in FIG. 13) in the non-physical region, the amplitude smaller than the ripple, and in the physical region The stop band multi-stage ripple characteristic may be provided, which has a ripple having a larger amplitude than the ripple in the stop band (stop band 1 in FIG. 13). In short, the frequency characteristics of the two-dimensional digital filter of the speaker array or the microphone array according to the present invention are that a plurality of ripples are provided in the stop band, and the amplitude of the ripple in the non-physical region is larger than the amplitude of the ripple in the physical region. Any frequency characteristic that is increasing may be used.

(4) In the above-described embodiment, a case has been described in which filter coefficients characteristic of the speaker array according to the present invention are preset in each one-dimensional digital filter forming the two-dimensional digital filter. The filter coefficient may be sequentially calculated and set each time such a speaker array or microphone array is used. In this way, for example, when the speaker array or the microphone array according to the present invention is installed and used in an acoustic space such as a concert hall, the directivity characteristics according to the acoustic characteristics of the acoustic space such as the size and shape of the acoustic space. Can be set appropriately.
Further, the filter coefficient set for each one-dimensional digital filter may be given from the outside of the speaker array or the microphone array. Specifically, a filter that sets a communication means such as a NIC (Network Interface Card) in the speaker array or the microphone array and a filter coefficient acquired via the communication network using the communication means in each one-dimensional digital filter. Coefficient setting means may be provided, and for example, reading means for reading data from a computer-readable recording medium such as a CD-ROM (Compact Disk-Read Only Memory) is provided instead of the communication means. Of course, the filter coefficient may be written and distributed on the recording medium, and the filter coefficient read by the reading means may be set in each one-dimensional digital filter by the filter coefficient setting means.

1 is a block diagram illustrating an electrical configuration of a speaker array 100 according to a first embodiment of the present invention. 3 is a diagram illustrating an example of amplitude characteristics of a two-dimensional digital filter of the speaker array 100 using a two-dimensional frequency plane. FIG. It is the figure which showed a part of the same amplitude characteristic with the equal amplitude characteristic figure. It is the figure which plotted the amplitude characteristic of the speaker array 100 for every predetermined angle. It is the figure which plotted the directional characteristic of the speaker array 100 for every predetermined frequency. It is the figure which showed the relationship between the frequency of the acoustic beam output from the speaker array 100, and the main lobe width of the acoustic beam. design method sectional characteristic at f ₁ ≧ f _l is disclosed in Non-Patent Document 2 is a diagram for explaining the. f 1 _<is a diagram for explaining a method of designing the cross-sectional characteristics at f _l is disclosed in Non-Patent Document 2. It is a figure for demonstrating the design method of the cross-sectional characteristic which concerns on this embodiment. It is a figure for showing the characteristic of the design result of the one-dimensional filter by the equiripple filter design program of Parks & McClellan. It is a figure which shows the design example of the one-dimensional filter by the equiripple filter design program of Parks & McClellan, and the design example of the one-dimensional filter of Dolph-Chebyshev characteristic. It is a figure which shows the electrical structure of the microphone array 200 which concerns on 2nd Embodiment of this invention. It is a figure which shows the frequency characteristic concerning a modification (3).

Explanation of symbols

  DESCRIPTION OF SYMBOLS 100 ... Speaker array, 110-i (i: Natural number of 1-n) ... Speaker, 120-i ... One-dimensional digital filter, 200 ... Microphone array, 210-i ... Microphone.

Claims (2)

A plurality of speakers arranged in a straight line at a predetermined interval, and provided corresponding to each of the plurality of speakers, a predetermined filter coefficient is set in advance, and the input voice data includes the filter coefficient. A one-dimensional digital filter that performs a corresponding filtering process and outputs the audio data obtained by subjecting the input audio signal to digital conversion to each one-dimensional digital filter, In a speaker array that supplies an audio signal obtained by subjecting audio data output from a filter to analog conversion to the corresponding speaker and outputs audio corresponding to the audio signal,
The filter coefficient set for each one-dimensional digital filter is:
When the frequency characteristics of the two-dimensional digital filter formed by each of the one-dimensional digital filters are represented by a two-dimensional frequency plane, the cross section in the spatial frequency direction has a plurality of ripples in the stop band, and the plurality The speaker array, wherein the two-dimensional digital filter has an amplitude characteristic in which the amplitude of the ripple in the non-physical region is larger than the amplitude of the ripple in the physical region. A plurality of microphones arranged in a straight line at a predetermined interval and a plurality of microphones are provided corresponding to each of the plurality of microphones, and a predetermined filter coefficient is set in advance. A one-dimensional digital filter that performs a corresponding filtering process and outputs the result, and supplies audio data obtained by performing digital conversion on the audio signal output from each of the plurality of microphones to the corresponding one-dimensional digital filter On the other hand, in the microphone array that outputs the sum signal of the audio data output from each one-dimensional digital filter,
The filter coefficient set for each one-dimensional digital filter is:
When the frequency characteristics of the two-dimensional digital filter formed by each of the one-dimensional digital filters are represented by a two-dimensional frequency plane, the cross section in the spatial frequency direction has a plurality of ripples in the stop band, and the plurality The microphone array, wherein the two-dimensional digital filter has an amplitude characteristic in which the amplitude of the ripple in the non-physical region is larger than the amplitude of the ripple in the physical region.