JP2010171513A - Sound reproducing device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To make it easy to distinguish the difference of speakers reproduced in a sound reproducing device used in a voice conference or the like. <P>SOLUTION: The sound reproducing device includes a first loudspeaker and a second loudspeaker comprising a plurality of loudspeaker units and a plurality of filters for setting a filter coefficient to the loudspeaker units. For the directional characteristics of the sound pressure of the first and second loudspeakers, the target direction of the high sound pressure is formed in one of the directions where the loudspeaker units are disposed, the dead angle direction of the low sound pressure is formed in a direction different from the target direction, and the target direction of the second loudspeaker is turned to the dead angle direction of the first loudspeaker. Thus, even when the reproducing sound signals of the same amplitude are input simultaneously to the first loudspeaker and the second loudspeaker, a listener can easily recognize the difference of the sound by the difference of the sound pressure. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a sound reproducing device used for an audio conference or the like for connecting different points by communication.

  For example, stereo reproduction has been used for a sound reproducing apparatus used in a conventional audio conference or the like. An example thereof is shown in Non-Patent Document 1 as a video-linked sound image localization device. FIG. 11 shows the configuration of the image-linked sound image localization apparatus and its operation will be briefly described. The video-linked type sound image localization device includes a display 110 and a pair of speakers 120L and 120R. The listener 130 is positioned approximately at the center of the speakers 120L and 120R, and localizes the voice of the previous speaker by the difference in sound pressure between the speakers 120L and 120R. When the former speaker A located in the left direction in the display 110 speaks, the sound pressure of the speaker 120L is high and the sound pressure of the speaker 120R is low. In the figure, the magnitude of the sound pressure is represented by the magnitude of an arcuate wave. Since the sound pressure of the speaker 120R at this time is small, it is omitted in FIG. In this way, the sound pressure difference between the left and right and the display on the display 110 can grasp who is speaking.

Edited by Nobuhiko Kitawaki "Sound Communication Engineering-Voice and Acoustic Technology in the Multimedia Age", Corona, Inc., August 30, 1996, first edition, page 209

However, in the conventional stereo reproduction method, when two speakers speak and reproduce simultaneously from both channels, it is not easy to distinguish and distinguish between the two speakers. In other words, if the utterances are simultaneously reproduced from the left and right speakers at the same volume, it is difficult to distinguish the two speakers because the sounds of the left and right speakers overlap.
The present invention has been made in view of this point, and an object of the present invention is to provide an acoustic reproduction apparatus that makes it easy to distinguish between different speakers even when reproduced sounds are reproduced simultaneously.

  The sound reproducing device of the present invention includes a first speaker and a second speaker. The first speaker is composed of a plurality of speaker units and a plurality of filters for setting filter coefficients in the speaker unit, and a target direction is formed in a predetermined direction, and a blind spot direction is formed in a direction different from the target direction. is there. The second speaker includes a plurality of speaker units and a plurality of filters that set filter coefficients in the speaker unit, and the target direction is directed to any blind spot direction of the first speaker.

  According to the present invention, since the target direction of the second speaker is directed in the direction of the blind spot direction of the first speaker, the sound pressure can be obtained even when the reproduction sound signal having the same amplitude is simultaneously input to the first speaker and the second speaker. Played with big and small. That is, it sounds as if the sound from the first speaker was emitted from a distance and the sound from the second speaker was emitted from near. Therefore, the listener can easily distinguish between different speakers.

The figure which shows the function structural example of the sound reproduction apparatus 100 of this invention. The figure which shows the precondition of how to obtain | require the filter coefficient h. The figure which shows the specific example of the directional characteristic of the 1st speaker. The figure which shows an example of the directional characteristic of the speaker comprised by three speaker units. The figure which shows an example of the directional characteristic obtained with the filter coefficient calculated by including background noise in a restraint condition. The figure which shows an example of the directional characteristic which made 0 degree of directions of a speaker using three speaker units into the target direction, and made 180 degrees into the blind spot direction. The figure which shows an example of the directional characteristic formed with the filter coefficient calculated using Formula (10). The figure which shows an example of the directional characteristic of the speaker of the modification 2. The figure which shows an example of the speaker 80 which arranged the speaker unit in the perpendicular direction. The figure which shows the function structural example of Example 2 of this invention which made the speaker 80 the 1st speaker. The figure which shows the structure of the video interlocking sound image localization apparatus disclosed by the nonpatent literature 1. FIG.

  Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows an example of a functional configuration of the sound reproducing device 100 of the present invention. The sound reproducing device 100 includes a first speaker 10 and a second speaker 20. The first speaker 10 includes a speaker unit 11, a filter 1 connected to the speaker unit 11, and a speaker unit 12 and a filter 2 connected to the speaker unit 12. For example, the reproduction sound signal L of one channel of a stereo signal transmitted from the other party by communication is input to the filter 1 and the filter 2. Different filter coefficients h ₁ and h ₂ are set for the filters 1 and 2, respectively. The directivity characteristics of the first speaker 10 are set by the filter coefficients h ₁ and h ₂ . In this example, a blind spot direction with a small sound pressure is set on the speaker unit 12 side in the arrangement direction of the speaker units 11 and 12. The directivity characteristics of the first speaker 10 and the second speaker 20 are schematically shown by broken lines.

Similar to the first speaker 10, the second speaker 20 includes a speaker unit 21, a speaker unit 22, and filters 3 and 4 connected to them. For example, the reproduction sound signal R of the other channel is input to the filters 3 and 4. Filter 3 has the same filter coefficient h ₁ as filter ₁ , and filter 4 has the same filter coefficient h ₂ as filter ₂ . In other words, the filter coefficients h ₁ and h ₂ are set in an opposite relationship with respect to the blind spot direction of the first speaker 10. As a result, the target direction in which the sound pressure of the second speaker 20 is large is directed to the blind spot direction of the first speaker 10. In this example, three listeners 33a, 33b, and 33c are positioned ahead of the blind spot direction of the first speaker 10 and the extension direction of the target direction of the second speaker 20.

  When signals having the same amplitude are simultaneously input to the reproduced sound signals L and R of the sound reproducing device 100 configured as described above, the listeners 33a to 33c reduce the reproduced sound L of the reproduced sound signal L and reduce the reproduced sound. The reproduced sound R of the signal R is heard greatly. That is, it sounds as if the sound from the first speaker with the blind spot facing the listeners 33a to 33c was emitted from a distance, and the sound from the second speaker with the target direction emitted from near. Therefore, even if a signal having the same amplitude is input to both channels at the same time, the sound can be easily recognized.

  Further, the first speaker 10 and the second speaker 20 may be arranged in a line in the direction of the listener, or the lines may be changed in parallel as shown in FIG. good. In addition, the first speaker 10 and the second speaker 20 may be arranged in parallel with the same distance between the first speaker 10 and the listeners 33a to 33c and the second speaker 20 and the listeners 33a to 33c. . In FIG. 1, the distance d between the rows of the speaker units 11 and 12 and the rows of the speaker units 21 and 22 is small. By increasing the distance d, it is possible to further facilitate the discrimination of the sounds of both channels. Is possible.

  The first speaker 10 and the second speaker 20 are generally arranged on a table 31 such as a conference desk. Moreover, you may make it arrange | position the display 30 which projects the other party in the position which opposes listener 33a-33c on both sides of the table 31. FIG.

[Setting method of speaker directivity]
A method of obtaining the filter coefficient h that sets the directivity characteristics of the first speaker 10 and the second speaker 20 will be described. Fig. 2 shows the idea of the preconditions. FIG. 2 schematically shows how a white noise plane wave is input to the first speaker 10 from a target direction and a blind spot direction. Here, the reason why white noise is input to the speaker is based on the idea based on the reciprocity theorem in which the same result can be obtained even if the sound source and the receiving point are reversed. According to this reciprocity theorem, even if it is assumed that sound is input to the speaker, a filter coefficient for outputting the sound can be obtained.

To the speaker unit 11 and 12 constituting the first speaker 10, and the target signal vector U _S shown in Equation (1) as a white noise of each plane wave is input, the noise signal vector U _N shown in Equation (2) Set.

ここでＴは転置を、Ｌはサンプリングの個数、ｋはサンプリング時刻を表す。

Here, T represents transposition, L represents the number of samplings, and k represents the sampling time.

The target signal vectors U _S1 (k) to U _S1 (kL) of the speaker unit 11 and the target signal vectors U _S2 (k) to U _S2 (kL) of the speaker unit 12 are white noise. As a result, random values are set with a time lag corresponding to the distance between the speaker unit 11 and the speaker unit 12. For example, the speaker units 11 and 12 are arranged at an interval of 8 cm along the target direction, and the sampling frequency is 8 kHz. Under that condition, the vector set for U _S2 (k) is set with a delay of 250 μs with respect to U _S1 (k), assuming that the speed of sound is 340 m / s. The noise signal vector is set based on the same idea.

The autocorrelation matrices R _S and R _N of the target signal vector and the noise signal vector are calculated by the equations (3) and (4). E [•] represents an expected value / time average.

Then, an average value S shown in Expression (5) obtained by multiplying the target signal vector U _S by a vector component at a specific time in the target signal vector is obtained. The filter coefficients h ₁ of the filter 1 connected to the speaker unit 11, the filter coefficient h ₂ of the filter 2 that is connected to the speaker unit 12 can be expressed by equation (6).

Thus, the filter coefficients h ₁ and h ₂ are the inverse matrix (R _S + R _N ) of the sum of the autocorrelation matrices respectively corresponding to the target signal vector U _S and the noise signal vector U _N reaching the plurality of speaker units. ^-1 multiplied by the average value S = E [U _Si (n) · U _S ] obtained by multiplying the target signal vector U _S by the vector component at a specific time in the target signal vector. Value.

The above calculation was performed by a computer to obtain the filter coefficients h ₁ and h ₂ . FIG. 3 shows the directivity characteristic of the sound pressure of the first speaker 10 obtained with the filter coefficients h ₁ and h ₂ . Concentric circles represent sound pressure and represent 15, 10, 5 dB from the outside. In this example, the target direction is the 180 ° direction and the sound pressure is about 10 dB. The opposite 0 ° direction is the blind spot direction, and the sound pressure is about 0 dB. In the figure, ● represents the approximate position of the speaker unit. The directivity characteristic has a frequency of 500 Hz. The directivity characteristics shown below all have a frequency of 500 Hz.

[Modification 1]
The first speaker 10 and the second speaker 20 are each composed of two speaker units, and each speaker unit is arranged in a row. However, each speaker of the present invention is limited to this configuration. Not. Two or more speaker units may be provided, and each speaker may not be arranged linearly.

  FIG. 4 shows an example of directivity characteristics of a speaker constituted by three speaker units. In this example, three speaker units were arranged in a line, the direction of the line was set to 0 °, and the target direction was calculated, and the 90 ° direction was calculated as the blind spot direction. Concentric circles represent the true value of the amplitude and are 25, 20, 15, 10, 5 from the outside.

In FIG. 4, only constraint conditions of the target direction 0 ° and the blind angle direction 90 ° are given. As a result, since the constraint condition in the target direction is 1, a steep valley appears in the amplitude in the target direction. In order to improve this directivity, background noise is incorporated into the constraint conditions. Background noise is diffusive noise that exists in the real environment. For example, air-conditioning sound that is reflected multiple times in the room.
How to incorporate this background noise, a noise vector U _n of the speaker units, as uncorrelated noise vector with each other, Equation (7), equation (8), the noise component as shown in Equation (9) U _{n (k} ) ... U _n (k−L) (n = 1, 2, 3) and 2 (L + 1) 0 components are added.

The noise vector U _n of each speaker unit adds to the noise signal vector U _N shown in Formula (2) forming a self-correlation matrix R _N, calculates the filter coefficients. In other words, the autocorrelation matrix R _N of the noise signal vector U _N, and each of a plurality of speaker units conditions plus uncorrelated noise. When the filter coefficients thus obtained are used, the directivity shown in FIG. 4 is improved as shown in FIG. The concentric circles in FIG. 5 indicate values normalized by setting the maximum amplitude to 1. The amplitude in the target direction 0 ° is 0.8, and the amplitude in the blind angle direction 90 ° is 0.1 or less. When there are two or more speaker units, the number of blind spots can be n-1.

  In FIG. 5, the blind spot direction is set to 90 °, but the directivity characteristics similar to those in FIG. 3 can be formed even if three speaker units are used. FIG. 6 shows directivity characteristics obtained with filter coefficients calculated by incorporating background noise into the constraint conditions, with 0 ° as the target direction and 180 ° as the blind spot direction. Thus, a characteristic having a single blind spot direction of 180 ° can be formed.

  In addition to the method of incorporating background noise into the constraint condition, an equation (10) obtained by adding a product of a small positive number λ and a unit matrix I of n (L + 1) rows and n (L + 1) columns to Equation (6) The calculation method shown in FIG. Here, n is the number of speaker units.

  FIG. 7 shows the directivity characteristic formed by the filter coefficient calculated by Expression (10). The directivity shown in FIG. 7 is slightly different from the directivity shown in FIG. 5 in that the directivity in the target direction is flattened. This is because the characteristic shown in FIG. 7 is a simple method in which only the diagonal component is added in the expression (10), whereas the characteristic shown in FIG. 5 is obtained by adding the expressions (7) to (9). This is due to the difference in the calculation method for obtaining the average value of the autocorrelation matrix. When the calculation is performed by adding Expressions (7) to (9), a minute component exists around the diagonal component.

[Modification 2]
The arrangement of the speaker units is not limited to a straight line. For example, a speaker in which five speaker units are arranged like a dice five may be used. The filter coefficient was obtained with + 15 ° and −15 ° (315 °) of the speaker as the blind spot direction and 0 ° as the target direction. FIG. 8 shows the directivity characteristic formed by the filter coefficient.
A blind spot direction with a small sound pressure is formed at ± 15 °, and a target direction with a large sound pressure is formed in the 0 ° direction. When the listener is positioned in the extension direction of the target direction of the speaker having such directivity characteristics, the sound from the speaker is recognized as a sound that can be heard from the listener's own shoulder.
The sound reproducing device 100 may be configured by using a speaker having such directivity characteristics instead of the first speaker 10.

The speaker unit may be arranged in the vertical direction. FIG. 9 shows a speaker 80 configured by arranging two speaker units 81 and 82 in the vertical direction. As described above, the speaker units are arranged in the vertical direction, and the filter coefficients h ₃ and h ₄ of the filter 83 connected to the speaker unit 81 and the filter 84 connected to the speaker unit 82 are appropriately determined by the method described above. When set, it is possible to form a blind spot in a predetermined elevation angle direction. When the speaker 80 is placed on a conference desk, for example, assuming that the directivity characteristic forms a blind spot in the elevation direction corresponding to the direction of the ear of the listener 90, as described in the example using the above five speaker units. The auditory sensation characteristic that can be heard from the shoulder and mouth direction of the listener is obtained.

  This speaker 80 may be used instead of the first speaker 10 of the first embodiment. A configuration example of the sound reproducing device 200 is shown in FIG. In FIG. 10, a filter and the like are omitted. As described above, the effects of the present invention can be obtained even if the first speakers are arranged in a line perpendicular to the second speaker 20 in which the speaker units are arranged in a line.

  As described above, the target direction is directed to the blind spot direction of the first speaker in which the target direction is formed in a predetermined direction and the blind spot direction is formed in a direction different from the target direction, and the first speaker. According to the sound reproducing device of the present invention including the second speaker, even if reproduced sound signals having the same amplitude are input to both speakers at the same time, the reproduced sounds can be easily recognized.

  In addition, although arrangement | positioning of the speaker unit was demonstrated in the linear form or the five examples of the dice, this invention is not limited to arrangement | positioning of a speaker unit. For example, a speaker unit may be arranged at the apex of a triangle, and the speaker unit can be arranged arbitrarily. In short, it is possible to form directional characteristics according to the conditions of the target signal vector and the noise signal vector set for the arrangement of the speaker units.

  Further, the example in which the column of the speaker units of the first speaker and the column of the speaker units of the second speaker are parallel and orthogonal has been described, but this parallel and orthogonal does not mean a strict relationship in mathematics. . As is clear from the example of directivity shown in FIG. 8, a variation in the range of about ± 15 ° is allowed. That is, when the speaker having the directivity shown in FIG. 8 is the first speaker, the target direction of the second speaker to be combined may be, for example, the + 15 ° direction or the −15 ° direction. Even if there is such a difference in directionality between the first speaker and the second speaker, the effect of the present invention can be obtained.

Claims (6)

A first speaker having a plurality of speaker units and a plurality of filters for setting filter coefficients in the speaker units, wherein a target direction is formed in a predetermined direction, and a blind spot direction is formed in a direction different from the target direction;
A plurality of speaker units; and a plurality of filters for setting a filter coefficient in the speaker unit; a second speaker having a target direction directed to any blind spot direction of the first speaker;
A sound reproducing device comprising: The sound reproducing device according to claim 1,
The sound reproducing apparatus according to claim 1, wherein the first speaker and the second speaker have the speaker units arranged in a row and the rows are parallel to each other. The sound reproducing device according to claim 1,
The first speaker has the speaker units arranged in a line,
The sound reproducing device according to claim 1, wherein the second speakers are arranged in a line in a direction orthogonal to the line. The sound reproducing device according to any one of claims 1 to 3,
The filter coefficient is
An inverse matrix (R _S + R _N ) ⁻¹ of the sum of autocorrelation matrices respectively corresponding to the target signal vector U _S and the noise signal vector U _N reaching the plurality of speaker units;
An average value S = E [U _Si (n) · U _S ] obtained by multiplying the target signal vector U _S by a vector component at a specific time in the target signal vector;
Is a value obtained by multiplying by H = (R _S + R _N ) ⁻¹ S. In the sound reproducing device according to claim 4,
The inverse matrix (R _S + R _N ) ⁻¹ is
(R _S + λI + R _N ) ⁻¹ obtained by further adding a product of a small positive number λ and a unit matrix I of n (L + 1) rows n (L + 1) columns (n is the number of speaker units). Sound reproducing device. In the sound reproducing device according to claim 4,
Autocorrelation matrix R _N of the noise signal vector, an audio reproducing device, wherein each of the plurality of speaker units is a value of condition plus uncorrelated noise.

Images