JP5082541B2 - Loudspeaker - Google Patents

Description

The present invention relates to a loudspeaker device that emits sound loud the desired audio signal to a predetermined loudspeaker area.

  A device that emits sound to a public place where a large number of people gather, such as a hospital lobby, has been put to practical use, and an array type is used to emit sound in a beam shape toward a specific direction in the public place. A speaker apparatus has also been proposed (for example, Patent Document 1).

JP 2005-173137 A

  Since the voice output in the lobby of the hospital is, for example, a call for a patient's examination, it is not preferable that the voice is heard by the entire lobby from the viewpoint of protecting personal information.

  As described above, an array speaker type loudspeaker that concentrates sound in one direction and emits it in the form of a beam has also been put into practical use, but the sound emitted in the form of a beam can be heard clearly only at one point. If the whole area of the hospital lobby, etc. is quiet, the sound can be heard clearly throughout the area even if the beam is emitted. There was a problem that it was.

An object of the present invention is to provide a loudspeaker that prevents a target voice from being clearly heard except at a specific target point.

The present invention relates to a plurality of speakers arranged at a plurality of points in or near a loudspeaker area, a sound signal input unit for inputting a target sound signal that is a sound signal to be amplified, and the target sound signal to each of the plurality of speakers. A delay for matching the arrival timings of the sound waves of the target audio signals emitted from the plurality of speakers at a specific target point by delaying the timing to be supplied to each of the speakers and then supplying them to each speaker. And a mask audio signal supply unit for supplying a mask audio signal for masking the target audio signal at a place other than the target point to the plurality of speakers .

  In the present invention, the supply timing is delayed to each of the plurality of speakers so that the arrival timing of the sound wave of the target audio signal matches at the target point. As a result, the target sound signal coming from each speaker at the target point has the same sound wave timing, so the target sound can be heard clearly with a high signal level, and the target sound coming from each speaker at a point other than the target point. Since the timing of the sound wave of the signal is shifted, the signal level does not increase and the sound is heard shifted and the intelligibility is lowered, making it difficult to hear the content of the voice. Thereby, the target voice can be heard with high gain and high clarity only at the target point.

In the above invention, the mask audio signal supply unit, said as a mask speech signal, to each of the plurality of speakers may be one, respectively without different signals correlated to test paper. Moreover, as the pre-Symbol no different signal correlated respectively, may use different random signals, respectively.

  Signals that are not correlated with each other, even if they are combined, do not have the same phase and increase in amplitude, or the phase is shifted and the amplitude does not decrease. Therefore, by outputting such uncorrelated signals from each speaker, an almost constant level steady sound with small signal level fluctuations depending on location can be made to flow through the entire loudspeaker area, and this can be used as a target point as a mask sound. The intelligibility of the target speech signal at a point other than can be reduced. A random signal has the highest decorrelation between a plurality of signals.

In the above invention, the mask audio signal supply unit supplies a periodic signal as the mask audio signal, and supplies the periodic signals to the speakers so that the signal level of the periodic signal becomes a minimum value at the target point. Processing means for processing the signal may be further provided .

  Although uncorrelated signals can be used as the mask sound signals, uncorrelated sounds are often harsh to the listener. Therefore, in the present invention, the same audio signal is sent from each speaker. As the same audio signal, for example, an environmental sound such as a wave sound may be used, so that even if the mask audio signal for masking the target audio signal flows at a predetermined level throughout the loudspeaker area, the listener may be disturbed. You don't feel anymore.

  Also, since the same audio signal has complete correlation, the signal level at a specific point (target point) can be compared with other points by adjusting the timing and frequency characteristics of the sound emitted from each speaker. It can be controlled to a small value (minimum value), and the signal level at other points can be controlled to a predetermined value that is a signal level that functions as mask sound. As a result, it is possible to make the target voice easier to hear by lowering the signal level of the mask voice at the target point to be lower than the signal level at other points while flowing the mask voice at a predetermined signal level throughout the loudspeaker area.

In the above invention, a delay time storage unit that stores the delay time corresponding to a plurality of target points is provided, and when the target point is designated, the delay unit determines the arrival timing of the sound wave at the designated target point. The target audio signal may be delayed by a delay time to be matched .

  According to the present invention, since the target audio signal (target audio signal) is emitted from the speakers arranged two-dimensionally or three-dimensionally so that the timing is matched only at the target point, the phase is matched at the target point. In addition to the target point, the phase / timing is shifted and the degree of intelligibility is reduced at points other than the target point. Thereby, the target voice signal can be heard only at the target point.

  Further, in the present invention, by filling the entire loudspeaker area with a random signal or mask audio signal, the random signal or mask is used in areas other than the target point where the timing of the target audio signal coincides and the sound pressure level is high. Since the level of the target voice signal is lowered with respect to the voice signal, that is, the S / N ratio is lowered, the target voice signal can be made more difficult to hear except at the target point.

Embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram of a loudspeaker according to a first embodiment of the present invention. Moreover, FIG. 2 is a figure which shows the example of arrangement | positioning of the speaker of the same loudspeaker.

  As shown in FIG. 2, in this loudspeaker, each of the speakers 20-1 to 20-n is not in a line shape like an array speaker, but an area in which a loud voice such as a hospital lobby is desired to reach (hereinafter referred to as a loudspeaker area). .) Is widely arranged in two or three dimensions. The speakers 20-1 to 20-n may be arranged in any direction regardless of the arrangement, but may be arranged around the loudspeaker area 7 and installed toward the center of the loudspeaker area 7. preferable. The speakers 20-1 to 20-n may be embedded in the ceiling.

  In FIG. 1, the loudspeaker units 10-1 to 10-n of the loudspeaker 1 are connected to the speakers 20-1 to n, respectively. Each loudspeaker unit 10 includes an FIR filter 11, a filter coefficient table 12, a random noise generator 13, an adder 14, and an output amplifier 15.

  The target audio signal is input from the target audio signal input unit 2 to the FIR filter 11. The target voice signal is a voice signal that the target person wants to listen to, such as a hospital call announcement. Further, a target point designation signal is input from the target point designation signal input unit 3 to the filter coefficient table 12. The filter coefficient table 12 sets a filter coefficient corresponding to the target point in the FIR filter 11 based on the input target point designation signal.

  Setting of the filter coefficient based on the target point designation signal to the FIR filter 11 is performed by all the loudspeaker units 10-1 to 10-n. Each of the sound amplifying units 10-1 to 10-n filters the target voice signal with the FIR filter 11 and outputs the target voice signal to the speakers 20-1 to 20-n, whereby the target voices emitted from the speakers 20-1 to 20-n are designated as target points. Since the timing is aligned at the target point specified by the signal and the phases of all frequency components are aligned, the frequency characteristics are flat, and the user at the target point can hear the target voice clearly.

  Further, since the sound signals emitted from the speakers 20-1 to 20-n arrive at positions other than the target points with a phase (timing) shifted, the intelligibility of the synthesized speech is low and the contents are heard. I can't. It should be noted that the frequency characteristics cannot be evaluated because the signals other than the target point are shifted in place timing.

  In addition, since the speakers 20-1 to 20-n are arranged in a two-dimensional or three-dimensional manner instead of an array (one-dimensional), the timings of the sound signals emitted from the speakers 20-1 to 20-n are consistent and clear. The place where high-quality speech is synthesized is a narrow dot-like region (a linear region in the array shape). As a result, a clear voice whose contents can be understood can be reached only in this dotted narrow area, and a low-clarity voice whose contents cannot be understood can be reached in the other areas.

  Note that a delay circuit may be provided in place of the FIR filter 11 if only the timings of the voices reaching the target points from the speakers 20-1 to 20-n are matched (if the frequency characteristics are not controlled). In this case, a delay time table is provided instead of the filter coefficient table 12.

  Further, in this embodiment, random noise is added to the target audio signal that has passed through the FIR filter 11 in each of the sound amplifying units 10-1 to 10-n. Random noise is generated by a random noise generator 13, and signals are added by an adder 14.

  Since random noise is generated independently for each of the speakers 20-1 to 20-n in each of the loudspeaker units 10-1 to 10-n, the random noise is completely correlated with the random noise input from each of the speakers 20-1 to 20-n. The RMS value of the amplitude is √n times when it is assumed that the RMS values are added at the same volume. For example, if the number of speakers is 8, it becomes √8 times. On the other hand, since the same sound is emitted from all the speakers 20-1 to 20 -n, the amplitude becomes a sound pressure ratio n times when the phases completely coincide, and the number of speakers is 8 when the number of speakers is 8. It becomes. In this way, when n (> 1) speakers output a target audio signal and a random noise whose phases coincide with each other at the target point, one speaker is provided by 20 Log (n / √n) = 20 Log (√n). The S / N ratio can be improved over time. That is, if there are eight speakers, 20 Log (√8) = 10 Log 8≈9, and the S / N ratio can be improved by 9 dB.

  In the filter coefficient table 12 of FIG. 1, filter coefficients corresponding to a plurality of target points 8 are stored, and these filter coefficients are measured by the method shown in FIG. As shown in FIG. 2, the speakers 20-1 to 20-n are installed in the same arrangement as when the loudspeaker of FIG. 1 is operated. Then, the microphone 5 is installed at the target point 8. In each of the sound amplifying units 10-1 to 10 -n, an active filter 18 is connected instead of the FIR filter 11, an input amplifier 16 that amplifies the input of the microphone 5, a sound signal input from the input amplifier 16, and a speaker 20. A calculation unit 17 is provided for calculating the filter coefficient of the active filter 18 based on the difference from the audio signal. That is, the arithmetic unit 17 and the active filter 18 constitute an adaptive filter. In this state, the target voice signal is input from the target voice signal input unit 2. Since the target audio signal at this time is an audio signal for determining the filter coefficient, any audio signal may be used as long as the audio signal lasts for a certain period of time. During this measurement, the random noise generator 13 stops the operation so as not to generate random noise. When the target voice signal is input, the target voice is emitted from each of the speakers 20-1 to 20-n for a certain period, and when the voice is collected from the microphone 5 installed at the target point 8, each sound collection unit 10-1 In ~ n, the calculation unit 17 calculates a filter coefficient based on the difference between the collected audio signal and the audio signal that passes through the active filter 18 and is supplied to the speaker 20. The filter coefficients calculated by the calculation units 17 of the sound amplification units 10-1 to 10-n are the filter coefficients of the speakers 20-1 to 20-n corresponding to the target point 8 at that time. This filter coefficient is registered in the filter coefficient table 12 of each of the loud sound units 10-1 to 10-n.

  The above processing is executed for a plurality of target points 8, and the calculated filter coefficient is registered in the filter coefficient table 12 in association with the position of the target point 8 at that time. Thereby, in the configuration of FIG. 1, when the target point designation information is input from the target point designation information input unit 3, the filter coefficient corresponding to the target point designation information is read from the filter coefficient table 12, and the FIR filter 11 is set.

  In this embodiment, each of the loudspeaker units 10-1 to 10-n is provided with a random noise generating unit 13 so as to emit random noise from each of the speakers 20-1 to 20-n in addition to the target voice. Since the intelligibility of speech is reduced at timings other than the target point 8 due to a timing shift, the configuration of the random noise emission and the random noise generator 13 is not essential.

<< Second Embodiment >>
In the above embodiment, random noise is output as the masked sound, but periodic sound may be output.
A second embodiment of the present invention will be described with reference to FIGS. In this embodiment, a periodic voice is output as a mask voice instead of random noise for the loudspeaker area.

  In other words, in the first embodiment, random noise is emitted from each of the speakers 20-1 to 20-n in addition to the target voice, and the target voice is masked outside the target point 8. In this configuration, since the entire loudspeaking area 7 is filled with random noise that is a mask sound, it is not necessarily comfortable. In addition, the target speech has an increased S / N ratio at the target point 8 due to an increase in the sound pressure level at the target point 8 due to synthesis by synchronous addition, but the random noise level at the target point 8 decreases. I'm not doing it.

  Therefore, in the second embodiment, a sound signal having periodicity that is not random noise is used as the mask sound, and the sound emission area 7 is filled with (comfortable) mask sound that is not random noise. At the target point 8, as in the first embodiment, the target sound emitted from each of the speakers 20-1 to 20-n is synthesized so as to be superimposed and the sound pressure level is increased. Furthermore, since the mask sound is a periodic sound signal, the sound of the mask sound is output at the target point by outputting the mask sound from each of the speakers 20-1 to 20n at a timing and volume that are canceled at the target point. The pressure level can be lowered to make it easier to hear the target sound.

4 to FIG. 6, parts having the same configuration as those of the first embodiment shown in FIG. 1 to FIG.
In FIG. 4, in this embodiment, loudspeaker units 30-1 to 30-n are connected to the respective speakers 20-1 to 20-n. Each of the loud sound units 30-1 to 30-n is provided with a mask sound signal FIR filter 31 and a filter coefficient table 32 in place of the random noise generator 13 of the loud sound units 10-1 to 10-n shown in FIG. A common mask audio signal is input from the mask audio signal input unit 4 to each of the sound amplifying units 30-1 to 30-n. The filter coefficient table 32 stores filter coefficients of the FIR filter 31 corresponding to the plurality of target points 8.

  The mask sound signal is filtered by the FIR filters 31 of the sound amplifying units 30-1 to 30-n so that the sound pressure level at the target point 8 becomes a minimum value. The minimum value is preferably the minimum within the sound emission area 7, but need not be smaller than the area where the sound in the corner of the sound area 7 is difficult to reach, and is locally minimum in the main part of the sound area 7. The idea is that there should be.

  In the filter coefficient table 32 of each of the sound amplifying units 30-1 to 30-n, filter coefficients measured by the following measurement method are stored in order to control the mask audio signal as described above.

FIG. 5 is a diagram showing a configuration for measuring the filter coefficient of the FIR filter 31. FIG. 6 is a diagram illustrating an arrangement example of error microphones 6-1 to n installed at the time of measurement.
As shown in FIG. 6, speakers 20-1 to 20-n are installed in the same arrangement as when the loudspeaker of FIG. 4 is operated. Then, the error microphones 6-1 to 6-m are installed almost evenly in the loudspeaker area 7. In each of the sound amplifying units 30-1 to 30-n, an active filter 35 is connected instead of the FIR filter 31, and the adaptive filter 35 is controlled based on the collected sound signal and the mask sound signal collected by the error microphones 6-1 to 6-1m. A calculation unit 34 for calculating the filter coefficient is provided. That is, the calculation unit 34 and the active filter 35 constitute an adaptive filter. The calculation unit 34 uses an adaptive algorithm such as an LMS (Least Mean Square) algorithm or an RLS (Recursive Least Square) algorithm, and among the error microphones 6-1 to m, the sound collection level of the microphone installed at the target point is a minimum value. Then, the filter coefficient is calculated so that the microphones at the other points have a certain target value (the volume to be heard).

  In this state, a mask sound signal is input from the mask sound signal input unit 4. Since the target audio signal at this time is an audio signal for determining the filter coefficient, any audio signal may be used as long as the audio signal lasts for a certain period of time. In this measurement, an audio signal is not input from the target audio signal input unit 2. By inputting the mask sound signal, the mask sound is emitted from each of the speakers 20-1 to 20-n for a predetermined time, and the sound is picked up by each error microphone 6-1 to m. The collected sound signal is input to the calculation unit 34 of each sound collection unit 30-1 to 30-n. The calculation unit 34 of each of the sound collection units 30-1 to 30-n performs calculation using an adaptive algorithm such as an LMS algorithm or an RLS algorithm based on the collected sound signal and the mask sound signal input from the mask sound signal input unit 4. The filter coefficient is calculated so that the sound pickup level of the microphone installed at the target point among the error microphones 6-1 to m becomes the minimum value, and the microphones at the other points have a certain target value (the volume to be heard). To do. The filter coefficient calculated by the calculation unit 34 of each of the sound amplifying units 30-1 to 30-n is the filter coefficient of the mask sound signal corresponding to the target point 8 at that time. This filter coefficient is registered in the filter coefficient table 32 of each of the sound amplifying units 30-1 to 30-n.

  The above processing is executed for a plurality of target points 8, and the calculated filter coefficients are registered in the filter coefficient table 32 in association with the positions of the target points 8 at that time. Note that the above-described processing for a plurality of target points may be executed by fixing the positions of the error microphones 6-1 to m and replacing the microphones serving as the target points 8, and each error microphone 6-1 for each target point. ~ M may be re-installed. Moreover, the microphone used as the target point 8 may be one, or plural.

  Thus, in the configuration of FIG. 4, when target point designation information is input from the target point designation information input unit 3, the filter coefficient corresponding to the target point designation information is read from the filter coefficient table 32, and the FIR filter 31 is set. Thereby, the sound pressure level of the mask sound at the target point 8 can be minimized.

  In this case as well, for the filter coefficient for the target speech, the filter coefficient corresponding to the target point designation information is read from the filter coefficient table 12 and set in the FIR filter 11. Thereby, the sound pressure level of the target voice at the target point 8 can be maximized.

  In this embodiment, a comfortable environmental sound such as a wave sound or a river sound can be used as the mask sound signal, so that the mask sound does not become uncomfortable even if the mask sound always flows in the entire loudspeaker area 7.

  In the first and second embodiments described above, on the assumption that the acoustic condition of the loudspeaker area 7 does not change greatly, the filter coefficient is measured once before operation, and the filter coefficient is stored in the filter coefficient table. However, when the acoustic conditions of the loudspeaker area 7 change, the loudspeaker can be operated with the configuration including the adaptive filter and the computation unit for performing adaptive computation shown in FIGS. 3 and 5. Good.

  Further, any method may be used for detecting where the target point 8 is, that is, where the user who wants to hear the target voice clearly is located. For example, in the case of a hospital lobby or the like, a target point may be detected by incorporating a wireless tag into an examination ticket possessed by a patient and detecting the position of the tag.

The block diagram of the loudspeaker which is 1st Embodiment of this invention The figure which shows an example of the installation form of the speaker of the same loudspeaker The figure explaining the measuring method of the filter coefficient of the same loudspeaker Block diagram of a loudspeaker according to a second embodiment of the present invention The figure explaining the measuring method of the filter coefficient of the same loudspeaker The figure which shows an example of the installation form of the error microphone at the time of the measurement of the said filter coefficient

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Loudspeaker 2 ... Target audio | voice signal input part 3 ... Target point designation | designated information input part 5 ... Microphone 6 (6-1 to m) ... Error microphone 7 ... Loudspeak area 8 ... Target point 10 (10-1 to n) ... Loudspeaker unit 11 ... FIR filter 12 ... Filter coefficient table 13 ... Random noise generator 14 ... Adder 15 ... Output amplifier 16 ... Input amplifier 17 ... Calculating unit 18 ... Adaptive filter 20 (20-1 to n) ... Speaker 30 (30 -1 to n) ... Loudspeaker unit 31 ... FIR filter 32 ... Filter coefficient table 34 ... Calculation unit 35 ... Adaptive filter

Claims (5)

A plurality of speakers arranged at multiple points in or near the loudspeaker area;
An audio signal input unit for inputting a target audio signal which is an audio signal to be amplified;
The timing of supplying the target audio signal to each of the plurality of speakers is delayed by an independent delay time and then supplied to each speaker, whereby the sound waves of the target audio signal emitted from the plurality of speakers are transmitted. A delay unit that matches the arrival timing at a specific target point ;
A mask audio signal supply unit for supplying a mask audio signal for masking the target audio signal at a place other than the target point to the plurality of speakers;
Loudspeaker with The mask audio signal supply unit, said as a mask speech signal, to each of the plurality of speakers, public address system according to claim 1, provided feeding each free different signal correlation.   The loudspeaker according to claim 2, wherein the different uncorrelated signals are different random signals. The mask audio signal supply unit supplies a periodic signal as the mask audio signal, and supplies the periodic signal to each speaker so that the signal level of the periodic signal becomes a minimum value at the target point. The loudspeaker according to claim 1, further comprising processing means for processing . A delay time storage unit that stores the delay time corresponding to a plurality of target points, and when the target point is designated, the delay unit matches the arrival timing of the sound wave at the designated target point The loudspeaker according to any one of claims 1 to 4, wherein the target voice signal is delayed.

Images