JP2011002704A - Sound signal transmitting device, sound signal receiving device, sound signal transmitting method and program therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem that, in order to obscure audio comprehension nature, a cut-off frequency is lowered to a degree in which a formant structure of voice is not transmitted so that sound of high frequency band may not be transmitted, but in such a case, information of a sound source other than voice (glass cracking sound etc.), which is high in frequency, is not transmitted.SOLUTION: In sound signal transmitting technology, a sound signal collected by a microphone is divided into frames by prescribed time, and a sound signal by frame is divided into a plurality of frequency band sound signals, and the divided sound signal by frame is obscured. When the obscuring processing is performed, frequency smoothed power of the input signal is determined, and variation of the input signal is compressed, and the input signal is time smoothed to longer time than a frame length.

Description

  The present invention relates to an acoustic signal transmission device, an acoustic signal reception device, an acoustic signal transmission method, and a program thereof that are used when an elderly person or the like is monitored remotely using communication.

  In an always-connected communication system that watches over the elderly and the like, privacy considerations are indispensable when transmitting to a side home or the like that watches over an acoustic signal collected by a microphone at the side home or the like. In order to grasp the state of the side to be watched while taking privacy into consideration, it is necessary to process the acoustic signal and make it unclear (without the comprehension of the voice) before transmitting it. Patent document 1 is known as a prior art. Patent Document 1 describes a method of reducing the intelligibility of an uttered voice through a low-pass filter.

JP 2006-238110 A

  However, in order to eliminate the intelligibility of the voice, it is necessary to lower the cut-off frequency to such an extent that the voice formant structure is not transmitted so that the sound in the high frequency band is not transmitted. However, if it does so, the subject that the information (sound etc. which the glass broke) of the sound source in high frequencies other than a voice will not be transmitted remains.

  In order to solve the above problems, an acoustic signal transmission technique according to the present invention divides an acoustic signal collected from a microphone into frames for each predetermined time, and the acoustic signal for each frame is divided into a plurality of frequency band acoustic signals. The sound signal divided for each frame is obscured. When performing the obscuring process, the frequency-smoothed power of the input signal is obtained, the fluctuations in the power of the input signal are compressed, and the input signal is timed longer than the frame length. To smooth.

  The present invention has an effect that the obscuring unit can remove the intelligibility of the voice and transmit the power of the non-voice signal and the time variation of the timbre.

The figure which shows the structural example of the acoustic signal transmitter. The figure which shows the example of a processing flow of the audio equipment transmitter 100. The figure which shows the example of filter bank (weight is the same) design. The figure which shows the example of a filter bank (a weight changes with frequencies) design. The figure which shows the example of a compression characteristic. The figure which shows the structural example of the acoustic signal receiver 200. FIG. The figure which shows the example of a processing flow of the acoustic signal receiver 200. FIG. 8 is a diagram illustrating a configuration example of the acoustic signal transmission device 300. The figure which shows the structural example of the acoustic signal receiver 400. The figure which shows the structural example of the acoustic signal transmitter 500. The figure which shows the example of a processing flow of the acoustic signal transmission apparatus 500. The figure which shows the example of arrangement | positioning of two microphone arrays. The figure which shows the structural example of the gain coefficient calculation part 530. The figure which shows the structural example of the acoustic signal receiver 600. FIG. The figure which shows the structural example of the acoustic signal transmitter 700. FIG. The figure which shows the example of a processing flow of the acoustic signal transmitter 700. The figure which shows the example of arrangement | positioning of the microphone 3 which inputs a signal into an acoustic signal transmitter. The figure which shows the example of a processing flow of the determination part 750. The figure which shows the structural example of the acoustic signal transmitter 800. The figure which shows the example of arrangement _| positioning of microphone 3 _R1 , 3 _L1 , 3 _R2 , 3 _L2 which inputs a signal into an acoustic signal transmitter.

  Hereinafter, embodiments of the present invention will be described in detail.

[Acoustic signal transmitting apparatus 100]
The acoustic signal transmission device 100 according to the first embodiment will be described with reference to FIGS. 1 and 2. FIG. 1 shows a configuration example of the acoustic signal transmission device 100, and FIG. 2 shows a processing flow example of the acoustic device transmission device 100.
The acoustic signal transmission device 100 includes a storage unit 103, a control unit 105, an ADC 107, a frame division unit 109, a division unit 111, an obscuring unit 120, and a transmission unit 130. Further, the obscuring unit 120 includes a frequency direction smoothing unit 121, a dynamic range compression unit 123, and a time direction smoothing unit 125. Each part will be described.

<Storage unit 103 and control unit 105>
The storage unit 103 stores / reads each input / output data and each data of the calculation process one by one. Thereby, each calculation process is advanced. However, the data need not necessarily be stored in the storage unit 103, and data may be directly transferred between the units.
The control unit 105 controls each process.

<ADC 107>
The ADC (Analog to Digital Converter) 107 converts the analog sound signal x (t) collected by the microphone 3 into a digital sound signal x (u) (s107). However, t represents continuous time and u represents discrete time. In addition, when a microphone with a built-in ADC is used, the acoustic signal transmission device 100 does not have to have an ADC. Note that the acoustic signal is a signal including voice information and non-voice information.

<Frame division unit 109>
The frame dividing unit 109 divides the digital acoustic signal x (u) collected from the microphone 3 into frames n every predetermined time (s109), and outputs the acoustic signal X (n). Here, n represents a frame number. The predetermined time is, for example, 100 msec or less. By shortening the predetermined time to some extent, the frequency resolution of the discrete Fourier transform performed by the dividing unit 111 described later is roughened, and the processing amount for obscuring the acoustic signal is reduced.

<Division unit 111>
The dividing unit 111 divides the acoustic signal X (n) for each frame n into a plurality of frequency band acoustic signals X (ω, n) (s111). However, ω represents a frequency. For example, it is realized by discrete Fourier transform or the like.

<Obscuring part 120>
The obscuring unit 120 obscures the acoustic signal divided for each frame n (s120). The obscuring unit 120 includes a frequency direction smoothing unit 121, a dynamic range compression unit 123, and a time direction smoothing unit 125, which are connected in cascade. In this embodiment, it is assumed that the frequency direction smoothing unit 121, the dynamic range compression unit 123, and the time direction smoothing unit 125 are connected in this order.

“Frequency direction smoothing unit 121”
The frequency direction smoothing unit 121 obtains the power | F (m, n) | ² that is smoothed in frequency of the input signal X (s121). The frequency direction smoothing unit 121 is a filter bank, for example. Note that m is a filter bank number, and m = 1 to M, where M is the number of filter divisions. | F (m, n) | ² obtains, for example, the power | X (ω, n) | ² of the frequency band acoustic signal X (ω, n) and multiplies them by the weight at the frequency ω of each filter bank. And adding them. 3 and 4 are examples of filter bank design. The value on the vertical axis is the weight. The frequency on the horizontal axis is logarithmically converted. If the filter band is increased as the frequency becomes higher in this way, it may be closer to the auditory sense. Thus, first, smoothing is performed in the frequency direction.

"Dynamic range compressor 123"
The dynamic range compressor 123 compresses power fluctuations of the input signal (s123). FIG. 5 shows an example of compression characteristics.
If the decibel transform value of | F (m, n) | ² is Lv (m, n), the compressed value Lv ′ (m, n) (true value | F ′ (m, n) | ² ) Is as follows.
Lv ′ (m, n) = − ∞ (Lv (m, n) <Np1)
Lv ′ (m, n) = r × Lv (m, n) + Gmin (Np1 ≦ Lv (m, n) ≦ Np2)
Lv ′ (m, n) = r × Np2 + Gmin (= Gmax) (Lv (m, n)> Np2)
However, r represents a compression rate (0 <r ≦ 1). For example, when compressing to half, it is set to 0.5. Np1 represents a point at which the output is reduced to 0 at a lower level. For example, by setting a value slightly higher than the level of background noise (background noise), it is possible to transmit the information only when the sound fluctuates, such as when something sounds. The background noise level can be automatically obtained by, for example, averaging Lv (m, n) for a long time, and therefore Np1 may be set to be, for example, 3 dB larger than the value. Np2 mainly represents a point at which the output is set to a predetermined value (Gmax) so that an excessive sound is not reproduced as it is. Gmin is the lower limit value of the compression value (however, if it is smaller than Np1, the compression value is -∞), and Gmax is the upper limit value of the compression value. The values of r, Gmin, Gmax, Np1 and Np2 may be changed for each m, that is, for each frequency band. For example, if the band (500 Hz to 3000 Hz) representing the characteristics of phonemes is strongly compressed, the voice becomes more unclear.

“Time direction smoothing unit 125”
The time direction smoothing unit 125 smoothes the input signal temporally in a time longer than the frame length (s125). For example, | F ′ (m, n) | ² is smoothed in the time direction to obtain | F ″ (m, n) | ^2. For smoothing in the time direction, for example, a forgetting factor is multiplied. There is a method of memorizing past time series and moving average.

  Each process of the frequency direction smoothing unit 121, the dynamic range compression unit 123, and the time direction smoothing unit 125 is a process for obscuring the collected sound signal. For example, if it is obscured only by smoothing in the frequency direction, it is necessary to reduce the number of filter banks, and it becomes difficult to transmit the frequency characteristics of the original signal. Similarly, if it is obscured only by smoothing in the time direction, it is necessary to convert it into a slow fluctuation with a larger time constant, and it becomes difficult to transmit the time fluctuation characteristic of the original signal. Also, if the power fluctuation is compressed too much, it becomes difficult to distinguish between loud sounds and small sounds. That is, by using three smoothings in combination, it is possible to obscure without increasing the strength of each smoothing too much. In addition, the effect is the same even if the order is changed. For example, it is possible to perform dynamic range compression and time direction smoothing before frequency direction smoothing.

<Transmitter 130>
The transmitting unit 130 transmits the signal | F ″ (m, n) | ² obscured by the obscuring unit 120 to the acoustic signal receiving device (s130). The transmitting unit 130 is, for example, a LAN adapter or the like.
[Acoustic signal receiving apparatus 200]
6 shows a configuration example of the acoustic signal receiving apparatus 200, and FIG. 7 shows a processing flow example of the acoustic signal receiving apparatus 200. The acoustic signal receiving apparatus 200 includes, for example, a reception unit 201, a storage unit 203, a control unit 205, a gain calculation unit 207, a carrier signal generation unit 208, a frame division unit 209, a division unit 211, a multiplication unit 213, and a time domain conversion unit 215. , DAC 217 and amplifier 219.

<Receiving unit 201>
The receiving unit 201 receives the signal | F ″ (m, n) | ² obscured by the obscuring unit 120 of the acoustic signal transmitting apparatus 100 (s201) .The receiving unit 201 is, for example, a LAN adapter or the like. .
<Storage unit 203 and control unit 205>
The storage unit 203 and the control unit 205 have the same configuration as the storage unit 103 and the control unit 105 of the acoustic signal transmission device 100, respectively.

<Gain calculation unit 207>
The gain calculation unit 207 calculates the gain g (ω, n) for each frequency ω using the obscured acoustic signal | F ″ (m, n) | ² (s207).
For example, the gain calculation unit 207, in the case of uniform filter bank as in FIG. 4, all the power of the frequency ω included in the m-th bank equal | F "(m, n) | If ² For example, in the case of a filter bank having an overlap as shown in Fig. 3, when a certain frequency ω is included in the filter bank of m1 and m2, the weights of m1 and m2 are set to W (m1) and W (m2), respectively. Then
| F "(ω, n) | ² = W (m1) × | F" (m1, n) | ² + W (m2) × | F "(m2, n) | ²
What should I do? Then, a normalization gain g (ω, n) is obtained by multiplying an appropriate small value, for example, the reciprocal of the maximum value of the power values of all frequency components.

<Carrier signal generator 208>
The carrier signal generator generates a stationary carrier signal z (u) (s208). The carrier signal z (u) may be any sound as long as it has a wider frequency component than the obfuscated signal F ″ (ω, n) and is a stationary sound. Even if it is applied, a sound like water murmur that is not harsh is good.The carrier signal z (u) may be stored in the storage unit 203 or the like, in which case the acoustic signal The receiving apparatus 200 may not include the carrier signal generation unit 208.

<Frame Dividing Unit 209 and Dividing Unit 211>
The frame dividing unit 209 divides a predetermined stationary carrier signal z (u) into frames for every predetermined time (s209), and outputs Z (n). The dividing unit 211 divides the carrier signal for each frame n into a plurality of frequency band signals Z (ω, n) (s211). The frame dividing unit 209 and the dividing unit 211 have the same configuration as the frame dividing unit 109 and the dividing unit 111 of the acoustic signal transmitting apparatus 100, respectively, and divide the frame at the same time interval as the frame dividing unit 109.

<Multiplier 213>
The multiplier 213 multiplies the gain G (ω, n) and the signal Z (ω, n) in the frequency band (s213) to obtain Y (ω, n).
<Time domain conversion unit 215>
The time domain conversion unit 215 converts the value Y (ω, n) obtained by the multiplication unit 213 into a time domain signal y (n) (s215). Further, the frames are combined and converted into a continuous digital signal y (u). For example, the signal is converted into a time domain signal by inverse discrete Fourier transform or the like.

<DAC 217 and amplifier 219>
The DAC (Digital-to-Analog Converter) 217 converts the digital signal y (u) input from the time domain conversion unit 215 into an analog signal y (t) (s217). The amplifier 219 amplifies the analog signal y (t) and outputs it to the speaker 221 (s219). Note that in the case of outputting a signal to a speaker in which a DAC or an amplifier is incorporated, the acoustic signal receiving device 200 does not have to have the DAC 217 or the amplifier 219.

<Effect>
By configuring the acoustic signal transmitting apparatus 100 and the acoustic signal receiving apparatus 200 as described above, it is possible to eliminate the intelligibility of the voice and transmit the power of the non-voice signal and the time variation of the timbre.
That is, with such a configuration, the pitch information of the acoustic signal on the acoustic signal transmitting device 100 side, that is, the watched side is erased. If transmission is performed without smoothing processing, the time change of the external shape of the frequency characteristic (corresponding to the formant structure in speech) remains, so that intelligibility remains (you can tell what you are talking about) By performing the clarification process, intelligibility can be eliminated. As a result, sound power and timbre changes are applied to the carrier signal while eliminating the intelligibility of the sound, so that it is easy to grasp the situation on the side where the sound signal receiving apparatus 200, that is, the watching side is watched over. . It should be noted that by appropriately adjusting the strength of the obscuring process, it is possible to select whether to consider more privacy or to convey more information.

<Others>
Note that the process of obtaining the gain g (ω, n) using | F ″ (m, n) | ² may be performed in the acoustic signal transmitting apparatus 100 before transmission, provided that the smoothing process, that is, non-transmission is performed. If the clarification processing is not performed on the transmission apparatus 100 side (watching side), privacy protection cannot be achieved.
In addition, processing performed by the gain calculation unit 207, the carrier signal generation unit 208, the frame division unit 209, the division unit 211, the multiplication unit 213, and the time domain conversion unit 215 of the acoustic signal reception device 200 is performed in the acoustic signal transmission device 100. The digital signal y (u) may be transmitted to the acoustic signal receiving device 200. However, the amount of data to be transmitted is larger than when transmitting | F ″ (m, n) | ² .

[Modification 1]
[Acoustic signal transmitting apparatus 300]
An acoustic signal transmission device 300 according to the first modification of the first embodiment will be described with reference to FIG. FIG. 8 shows a configuration example of the acoustic signal transmission device 300.
The acoustic signal transmission device 300 includes a storage unit 103, a control unit 105, an ADC 107, a frame division unit 109, an obscuring unit 320, a gain calculation unit 327, and a transmission unit 130. A different part from Example 1 is demonstrated. In this modification, M band limiting filters (filter banks 321) are used instead of the dividing unit 111 that uses discrete Fourier transform.

<Obscuring part 320>
The obscuring unit 320 includes a filter bank 321, a dynamic range compression unit 123, and a time direction smoothing unit 125.
“Filter Bank 321”
An acoustic signal X (n) for each frame n is input to the file bank 321. The filter bank 321 combines the dividing unit 111 and the frequency direction smoothing unit 121 of the first embodiment, and obtains a signal | F (m, n) | ² smoothed in the frequency direction for each frequency band acoustic signal. For example, the acoustic signal X (n) is passed through the filter bank 321 to obtain a signal for each frequency band, and the signal | F (m, n) | ² smoothed in the frequency direction is obtained by squaring and adding the amplitude. be able to. This signal | F (m, n) | ² is output to the dynamic range compression unit 123 or the time direction smoothing unit 125. In this modification, it is output to the dynamic range compression unit 123. The processing contents of the dynamic range compression unit 123 and the time direction smoothing unit 325 are the same as those in the first embodiment.
<Gain calculation unit 327>
The gain calculation unit 327 has the same configuration as the gain calculation unit 207 of the acoustic signal receiving device 200 of the first embodiment.

[Acoustic signal receiving apparatus 400]
FIG. 9 shows a configuration example of the acoustic signal receiving device 400. The acoustic signal receiving device 400 includes, for example, a reception unit 201, a storage unit 203, a control unit 205, a carrier signal generation unit 208, a frame division unit 209, a filter bank 421, a multiplication unit 413, a frame synthesis unit 415, a DAC 217, and an amplifier 219. Have.
<Filter bank 421>
A carrier signal Z (n) for each frame n is input to the file bank 421. For example, the filter bank 421 passes the carrier signal Z (n) and obtains a signal Z (ω, n) for each frequency band. This is output to the multiplier 413.

<Multiplier 413>
The multiplier 413 receives the gain g (ω, n) received by the receiver 201 and the carrier signal Z (ω, n) for each frequency, and multiplies these values to obtain Y (ω, n). The data is output to the frame composition unit 415.
<Frame synthesis unit 415>
The frame synthesizer 415 synthesizes signals for each frame and frequency, obtains a digital signal y (u), and outputs it to the DAC 217. The processing of the DAC 217 and the amplifier 219 is the same as that in the first embodiment.
By configuring the acoustic signal transmitting device 300 and the acoustic signal receiving device 400 as described above, the same effects as in the first embodiment can be obtained.

  Not only is the obscured acoustic information always transmitted, but when the watched side desires, it is not only effective in an emergency, but also more convenient if a voice call is possible. Patent Document 1 describes a method in which, for example, an obfuscation process is canceled when an elderly person or the like to be watched turns on a call switch. However, there are also concerns that the operation may be troublesome, forgetting to switch the switch after a call, voice that will not be obscured is transmitted to the other party, and privacy is infringed. In addition, it is conceivable to give psychological anxiety to the user that such a specification is always picked up by a microphone. In view of this, an apparatus is provided that automatically cancels the voice obscuring process and enables transmission of clear voice. An embodiment will be described in which speech uttered at a specific location is transmitted clearly and other sounds are obscured. By doing this, the user (the side to be watched) does not need to perform any special operation, and only needs to remember the specific place, and the risk of transmitting unintended voice to the other party is reduced. For this purpose, it is necessary to determine the presence or absence of a sound source at a specific location. As a method therefor, an embodiment applying Japanese Patent Application Laid-Open No. 2009-25490 (hereinafter referred to as “reference document 1”) will be described.

[Acoustic signal transmission device 500]
An acoustic signal transmission device 500 according to the second embodiment will be described with reference to FIGS. 10 and 11. FIG. 10 shows a configuration example of the acoustic signal transmission device 500, and FIG. 11 shows a processing flow example of the acoustic signal transmission device 500. Only parts different from the first embodiment will be described. The two microphone arrays 3R and 3L are arranged as shown in FIG.
The acoustic signal transmission device 500 includes a determination unit 550 that determines whether sound is generated from a specific position. Then, when no sound is generated from a specific position, the obscuring unit 120 generates an obscured acoustic signal.

  The acoustic signal transmission apparatus 500 includes a sound collection unit 4, a frame division unit 109, a division unit 111, a processing target signal generation unit 540, a power spectrum estimation unit 507, a gain coefficient calculation unit 530, an obscuration unit 120, and a multiplication unit 509. A determination unit 550 and a transmission unit 540 are included. The processing contents of the sound collection unit 4, the processing target signal generation unit 540, the power spectrum estimation unit 507, the gain coefficient calculation unit 530, and the multiplication unit 509 are described in detail in Reference Document 1. In this embodiment, an outline will be described.

<Sound collecting unit 4>
The six sound collection units 4-1, 4-2, 4-3, 4-4, 4-5 and 4-6 use the output signals of the microphone arrays 3 </ b> R and 3 </ b> L configured by mounting a plurality of microphones. Then, sounds of different areas are collected (s4). Digital acoustic signals x _SL (u), x _SR (u), x _NL (u), x _NR (u), x _SC (u), and x _NC (u) are output, respectively. The sound collection unit may be 6 or more.

<Frame Dividing Unit 109 and Dividing Unit 111>
The frame dividing unit 109 and the dividing unit 111 have the same configuration as that of the first embodiment. The frame dividing unit 109 converts the digital acoustic signals x _SL (u), x _SR (u), x _NL (u), x _NR (u), x _SC (u), and x _NC (u) into frames at predetermined time intervals. divided into n (s109), the audio signal _{X SL (n), X SR} (n), X NL (n), X NR (n), X SC (n), and outputs the _X NC (n). The dividing unit 111 converts the acoustic signal for each frame n into a plurality of frequency band acoustic signals X _SL (ω, n), X _SR (ω, n), X _NL (ω, n), X _NR (ω, n). , X _SC (ω, n) and X _NC (ω, n) (s111) and output. Note that the frequency domain conversion unit 5 described in Reference 1 includes the functions of the frame dividing unit 109 and the dividing unit 111.

<Processing Target Signal Generation Unit 540>
The processing target signal generation unit 540 generates a processing target signal Y _s (ω, n) from signals from one or more predetermined microphones or sound collection units (s540).
<Power Spectrum Estimator 507>
The power spectrum estimation unit 507 calculates the desired sound source from the signal amount of each collected sound signal obtained by each sound collecting unit 4-1, 4-2, 4-3, 4-4, 4-5, 4-6. The signal amount and the signal amount of other sound sources are estimated for each frequency (s507), and an estimated signal power vector X _opt (ω, n) is output.

<Gain coefficient calculation unit 530>
FIG. 13 shows a configuration example of the gain coefficient calculation unit 530. The gain coefficient calculation unit 530 calculates the signal amount of the desired sound source | S (ω, n) | ² , the signal amounts of all sound sources including the signal amount of the desired sound source | S (ω, n) | ² + | N _LL (ω , N) | ² + | N _L (ω, n) | ² + | N _C (ω, n) | ² + | N _R (ω, n) | ² + | N _RR (ω, n) | ² , Then, a gain coefficient is obtained for each frequency from the processing target signal Y _s (ω, n) (s530). The gain coefficient calculation unit 530 includes a vector element extraction unit 81, a first gain calculation unit 531, a second gain calculation unit 532, and a gain multiplication unit 533.

“Vector Element Extraction Unit 81”
The vector element extraction unit 81 converts the input estimated signal power vector X _opt (ω, n) into an estimated signal power | S (ω, n) | ² and an estimated left-side noise power | N _LL (ω, n) | ² , estimated left noise power | N _L (ω, n) | ² , estimated front noise power | N _C (ω, n) | ² , estimated right noise power | N _R (ω, n) | ² , The estimated right side noise power | N _RR (ω, n) | ² is output (s81).
“First Gain Calculation Unit 531”
The first gain calculating unit 531, the estimated signal power ^| S (ω, n) | ² and the processing signal _{Y S} (ω, n), the first gain factor _{G S} a (omega, n) as follows Calculate and output (s531).

“Second Gain Calculation Unit 532”
The second gain calculation unit 532 obtains the ratio of the signal amount of the desired sound source to the signal amounts of all sound sources including the signal amount of the desired sound source (hereinafter referred to as “second gain coefficient”) (s532). For example, the second gain calculation unit 532 includes the estimated signal power | S (ω, n) | ² , the estimated left-side noise power | N _LL (ω, n) | ² , and the estimated left-side noise power | N _L (ω, n) | ² , estimated front noise power | N _C (ω, n) | ² , estimated right noise power | N _R (ω, n) | ² , estimated right noise power | N _RR (ω, n) From | ² , the second gain coefficient G _SNR (ω, n) is calculated as shown in the following equation and output.

| N _LL (ω, n) | ² + | N _L (ω, n) | ² + | N _C (ω, n) | ² + | N _R (ω, n) | ² + | N _RR ( If ω, n) | ² is the power | N (ω, n) | ² of the signal amount from a sound source other than the desired sound source, it can be expressed as the following equation.

The second gain calculation unit 532 outputs the second gain coefficient to the gain multiplication unit 533 and also outputs it to the determination unit 550.

“Gain multiplier 533”
The gain multiplication unit 533 outputs a product of the first gain coefficient G _S (ω, n) and the second gain coefficient G _SNR (ω, n) as a gain coefficient R (ω, n) as shown in the following equation ( s533).
R (ω, n) = G _S (ω, n) · G _SNR (ω, n)

<Determining unit 550>
It is determined using the second gain coefficient G _SNR (ω, n) whether or not sound is generated from a predetermined region (s550). Then, when no sound is generated from a specific position, the obscuring unit 120 generates an obscured acoustic signal. If there is no sound in the target area S of FIG. 12, the value of G _SNR (ω, n) in Expression (1) is close to 0 for all frequency components. On the other hand, if there is sound in the target area S, only the frequency component G _SNR (ω, n) of the sound source in the target area S becomes a value close to 1.
Therefore, for example, the determination unit 550 determines whether sound is generated from a predetermined region (target area S) by the following method.

The sound source in the target area S is limited to voice, and G _SNR (ω, n) is added in the frequency direction at the frequency of the voice band (for example, about 100 Hz to 3 kHz), and the value is determined in advance. When the value is exceeded, it is determined that there is sound in the target area S. In addition, when G _SNR (ω, n) sums the number of components exceeding a predetermined threshold value and it exceeds another predetermined threshold value instead of adding. Alternatively, it may be determined that there is sound in the target area S. The threshold value is experimentally determined because it varies depending on the characteristics of the microphone array and the acoustic state of the room. Furthermore, since the above value (added value of G _SNR (ω, n) or the added value of the number of components) changes every moment in units of frame division time, clarification and obscuration is accelerated due to determination errors and the like. Sometimes it changes. In order to prevent this, the obtained values may be smoothed by moving average in the time direction or by a forgetting factor.

By the above method, the determination unit 550 outputs the processing target signal Y _S (ω, n) generated by the processing target signal generation unit 540 to the multiplication unit 509 when sound is generated from a specific position. If the sound is not generated from a specific position, the processing target signal Y _S (ω, n) is controlled to be output to the obscuring unit 120. Further, an additional signal or a control signal is transmitted to the transmission unit 540.

<Obscuring part 120>
The obscuring unit 120 has the same configuration as that of the first embodiment. Therefore, the obscuring unit 120 generates the obscured acoustic signal | F ″ (m, n) | using the processing target signal Y _S (ω, n) (s120).

<Multiplier 509>
The multiplier 509 multiplies the signal Ys (ω, n) whose main component is the signal of the desired sound source given from the processing target signal generator 540 by a gain coefficient R (ω, n) for each frequency domain (s509). ), The background noise component contained in the signal whose main component is the signal of the desired sound source 1 can be suppressed. Multiplier 509 receives signal Ys (ω, n) and gain coefficient R (ω, n), and outputs signal Y (ω, n) in which background noise components are suppressed.

<Transmitter 540>
The transmitting unit 540 suppresses the additional signal or control signal from the determining unit 550, the acoustic signal | F ″ (m, n) | obfuscated from the obscuring unit 120, or the signal Y from which the background noise component is suppressed from the multiplying unit 509. (Ω, n) is input, and | F ″ (m, n) | or Y (ω, n) is transmitted to the acoustic signal receiving device 600 (s540). For example, the additional signal is a signal for identifying a signal to be transmitted | F ″ (m, n) | or Y (ω, n), and the transmission unit 540 adds the received additional signal to each signal for transmission. For example, the control signal is a signal for controlling which channel is used for transmission when | F ″ (m, n) | and Y (ω, n) each have a transmission channel.

[Acoustic signal receiving apparatus 600]
FIG. 14 shows a configuration example of the acoustic signal receiving device 600. The receiving unit 601 of the acoustic signal receiving device 600 determines whether the additional signal or the signal received through the receiving channel is | F ″ (m, n) | or Y (ω, n). In the case of “(m, n) |”, it is transmitted to the gain calculation unit 207. On the other hand, when the received signal is Y (ω, n), it is transmitted to the time domain conversion unit 215.

<Effect>
By configuring the acoustic signal transmitting device 500 and the acoustic signal receiving device 600 in this manner, the same effects as in the first embodiment can be obtained. Furthermore, the voice obscuration process and the cancellation of the process are automatically performed by determining the position of the sound source, thereby preventing forgetting to switch and avoiding complications such as switch operation. It has the effect of realizing natural telecommunications.
In addition to the automatic determination by the determination unit 550 described above, if a switch for canceling the obscuring process is provided, the labor of the user is increased, but the reliability can be increased.

  Further, although the enhancement processing is performed on the processing target signal Ys (ω, n), the processing target signal Ys (ω, n) may be transmitted as it is without performing the enhancement processing. In this case, the multiplication unit 509 may not be provided, and the first gain calculation unit 531 and the gain multiplication unit 533 in the gain calculation unit 530 may not be provided. In the case of this embodiment, the sound outside the area cannot be completely muted (unclear), but if the speaker is in a state where he / she is talking, It is the same as a telephone, and even if a little ambient sound is transmitted to the other party, it is not an unnatural call.

<Others>
Furthermore, in this method, a sound source having the same frequency band as the sound in the target area S reacts, so that a determination method using the sound feature amount may be added thereto. For example, there is a method using the presence or absence of periodicity of the sound source. In this case, output signals from the microphone arrays 3R and 3L are used as input signals to the determination unit 550. For example, Kenzo Ito, “Steady noise suppression method based on speech and non-speech discrimination processing”, Journal of the Acoustical Society of Japan, Von. 61, No. 8, P431-440, 2005 (hereinafter referred to as “Reference 2”), The determination unit 550 calculates an autocorrelation value for the residual signal obtained by linear prediction analysis of the input signal, searches for the maximum value within the range of the fundamental frequency of the speech (for example, about 50 Hz to about 300 Hz), and the value is When a predetermined threshold value (approximately 0.3 or more) is exceeded, it is determined that there is periodicity. In addition, the determination unit 550 determines that the sound is speech when the power of the residual signal exceeds a predetermined threshold value. Note that this threshold value may be set automatically, for example, by continuously measuring the long-term average of the residual power, and then increasing the threshold value by, for example, about 6 dB.

An embodiment using two microphones based on Japanese Patent No. 3670562 (hereinafter referred to as “Reference 3”) will be described as a method of automatic determination when obscuring or clarifying.
[Acoustic signal transmitter 700]
FIG. 15 shows a configuration example of the acoustic signal transmission device 700, and FIG. 16 shows a processing flow example of the acoustic signal transmission device 700. The acoustic signal transmission apparatus 700 includes a stereo signal input unit 701, a frame division unit 109, a division unit 111, a similarity calculation unit 704, an attenuation coefficient calculation unit 705, a multiplication unit 716, addition units 717 and 760, an obscuring unit 120, A determination unit 750 and a transmission unit 540 are included.

FIG. 17 shows an arrangement example of the microphones 3 for inputting signals to the acoustic signal transmission device. The two microphones 3 are arranged at positions symmetrical to the sound source (speaker) position to be clarified.
The processing contents of the stereo signal input unit 701, the similarity calculation unit 704, the attenuation coefficient calculation unit 705, the multiplication unit 716, and the addition unit 717 are described in detail in Reference Document 3. The frame dividing unit 109, the dividing unit 111, and the obscuring unit 120 have the same configurations as those in the first embodiment, and the transmission unit 540 and the acoustic signal receiving device 600 have the same configurations as those in the second embodiment. In this embodiment, the outline will be described.

<Stereo signal input unit 701>
The acoustic signal transmitting apparatus 700 receives a stereo signal via the stereo signal input unit 701 (s701). This stereo signal is processed for each of the left and right channels.
<Frame Dividing Unit 109 and Dividing Unit 111>
The frame dividing unit 109 and the dividing unit 111 have the same configuration as that of the first embodiment. The frame dividing unit 109 divides the digital acoustic signals x _R (u) and x _L (u) into frames n every predetermined time (s109), and outputs the acoustic signals X _R (n) and X _L (n). . The dividing unit 111 divides the acoustic signal for each frame n into a plurality of frequency band acoustic signals X _R (ω, n) and X _L (ω, n) (s111) and outputs the result. Note that the left channel frequency band division unit 103 and the right channel frequency band division unit 104 described in Reference 3 have the functions of the frame division unit 109 and the division unit 111.

<Similarity calculation unit 704>
In the similarity calculation unit 704, the similarity a (ω, n) is calculated for each of the same frequency band for X _R (ω, n) and X _L (ω, n) (s704). For example, a (ω, n) is composed of ai (ω, n) and ap (ω, n), and ai (ω, n) and ap (ω, n) are obtained by the following equations.

<Attenuation coefficient calculation unit 705>
The attenuation coefficient calculation unit 705 calculates an attenuation coefficient g (ω, n) for each frequency band based on the similarity a (ω, n) calculated for each frequency band (s705).
<Multiplier 716>
The multiplier 116 multiplies the attenuation coefficient g (ω, n) by X _R (ω, n) and X _L (ω, n) of each frequency band (s716), and generates an echo signal other than the target area S. Are output as signals X ′ _R (ω, n) and X ′ _L (ω, n).

<Adding units 717 and 760>
The adder 717 adds the left and right channel signals X ′ _R (ω, n) and X ′ _L (ω, n) (s717), converts them to monaural, and calculates Y (ω, n).
On the other hand, the adding unit 760 adds the left and right channel signals X _R (ω, n) and X _L (ω, n) (s760), monauralizes, and calculates X (ω, n).
<Obscuring part 120>
The obscuring unit 120 has the same configuration as that of the first embodiment. Therefore, the obscuring unit 120 generates the obscured acoustic signal | F ″ (m, n) | using X (ω, n) (s120).

<Determining unit 750>
The determination unit 750 determines whether or not sound is generated from a specific region using the similarity a (ω, n) (s750). FIG. 18 shows a processing flow example of the determination unit 750.
An acoustic signal in a position symmetrical to the two left and right microphones, that is, in the target area S, is input to the two microphones 3 with the same phase and the same power. First, initial setting is performed (s751), and the value of ai (ω, n) obtained by Equation (2) is close to a predetermined threshold value close to 1 (for example, 1−k1 ≦ ai ( ω, n) ≦ 1 + k1, for example, k1 = 0.05) (s752), and the value of ap (ω, n) obtained by the equation (3) is close to 1 in advance. (For example, 1−k2 ≦ ap (ω, n) ≦ 1 + k2, for example, k2 = 0.05) (s753), the frequency component of the sound source near the center judge. The above determination is made for all frequencies (s755, s756).

Then, the number of frequency components subjected to the determination is counted (s754), and when the number cnt is equal to or greater than a predetermined threshold value k3 (for example, 30% of all frequency components), a specific region is used. It is determined that sound is generated (s757, s758). If it is less than k3, it is determined that no sound is generated from the specific area (s759). The processing after the determination is the same as that of the determination unit 550.
<Transmitter 540>
The transmission unit 540 uses the additional signal or control signal from the determination unit 550, the obscured acoustic signal | F "(m, n) | output from the obscuring unit 120, or the signal Y output from the multiplication unit 717. (Ω, n) is transmitted to the acoustic signal receiving device 600 (s761, s762).

<Effect>
By adopting such a configuration, the same effects as those of the first and second embodiments can be obtained. In the case of the third embodiment, only two microphones are required, and there is an advantage that the amount of calculation is small. However, in principle, the position of the sound source cannot be determined pinpointed as in the second embodiment. That is, the sound source located at a symmetrical position from the two microphones is determined as a sound that is clearly transmitted regardless of the distance from the microphone. However, the sound source signal at a position far from the microphone is added with the reflected sound from the wall, etc., and the disturbance of the left and right phase difference increases, and the sound power gradually decreases due to distance attenuation. The above is a range in which the position of the ellipse as shown in FIG. 17 is clarified.

<Others>
It is the same that the audio feature amount described in Reference 2 may be added to the determination. Moreover, you may transmit to the acoustic signal receiver 600, without suppressing the sound emitted from other than a specific area | region. In that case, the multiplication unit 716 and the attenuation coefficient calculation unit 705 need not be provided. In addition, the determination unit 750 controls the signals X _R (ω, n) and X _L (ω, n) output from the dividing unit 111 to be output to one of the multiplication unit 716 and the addition unit 760. Also good.

  Example in which two microphone pairs, that is, four microphones, and two similarity calculation units described in Example 3 are used to determine whether sound is generated in a specific area narrower than Example 3. explain.

[Acoustic signal transmission device 800]
FIG. 19 shows a configuration example of the acoustic signal transmission device 800. The acoustic signal transmission apparatus 800 includes a first similarity calculation unit 810, a second similarity calculation unit 811, adders 817 and 860, a determination unit 850, an obscuring unit 120, and a transmission unit 540. The obscuring unit 120 has the same configuration as that of the first embodiment, and the transmission unit 540 and the acoustic signal receiving device 600 have the same configurations as those of the second embodiment.
FIG. 20 shows an arrangement example of the microphones 3 _R1 , 3 _L1 , 3 _R2 , and 3 _L2 for inputting signals to the acoustic signal transmission device. The overlapping area at the center of each microphone pair 3 _R1 , 3 _L1 and 3 _R2 , 3 _L2 is defined as a target area S.

<First Similarity Calculation Unit 810 and Second Similarity Calculation Unit 811>
The first similarity calculation unit 810 includes a stereo signal input unit 701, a frame division unit 109, a division unit 111, and a first similarity calculation unit 804. The second similarity calculation unit 811 has the same configuration as the first similarity calculation unit 810, and includes a second similarity calculation unit (not shown) instead of the first similarity calculation unit 804. The frame dividing unit 109 and the dividing unit 111 have the same configurations as those in the first embodiment, and the transmission unit 540 and the acoustic signal receiving device 600 have the same configurations as those in the second embodiment. The first similarity calculation unit 804 and the second similarity calculation unit have the same configuration as the similarity calculation unit 704 of the third embodiment.

“First similarity calculation unit 804”
The first similarity calculation unit 804 receives the frequency band acoustic signals X _R1 (ω, n) and X _L1 (ω, n) obtained using the output signals of the microphones 3 _R1 and 3 _L1 and receives the first similarity. a1 (ω, n) is obtained. Similarly, the second similarity calculation unit receives the frequency band acoustic signals X _R2 (ω, n) and X _L2 (ω, n) obtained using the output signals of the microphones 3 _R2 and 3 _L2 , and receives the second similarity. The degree a2 (ω, n) is obtained.

<Determining unit 850>
The determination unit determines whether sound is generated from a specific region using the first similarity a1 (ω, n) and the second similarity a2 (ω, n). Specifically, the determination process illustrated in FIG. 18 is performed using the similarities a1 and a2, and only when it is determined that sound is generated from a predetermined region for any similarity. (S757), it is determined that sound is generated from the target area S of FIG. The processing after the determination is the same as that of the determination unit 550.

<Effect>
By adopting such a configuration, it is possible to obtain the same effects as those of the first embodiment, the second embodiment, and the third embodiment. Compared with the second embodiment, the calculation processing is lighter and four microphones are required. On the other hand, compared with the third embodiment, there is an advantage that a specific area can be narrowed.
<Adding unit 817 and adding unit 860>
The adder 817 and the adder 860 are input with frequency band acoustic signals X _R1 (ω, n), X _L1 (ω, n), X _R2 (ω, n), and X _L2 (ω, n), respectively. The values Y (ω, n) and X (ω, n) obtained by adding the values are output. In this case, Y (ω, n) and X (ω, n) are the same.

<Others>
Note that an attenuation coefficient calculation unit may be provided as in the third embodiment. In this case, in order to obtain the attenuation coefficient from the similarity calculation means 810 and 811, two attenuation coefficient calculation sections are provided, and the output signal of each similarity calculation means is multiplied by the corresponding attenuation coefficient.
The above-described acoustic signal transmitting apparatus and acoustic signal receiving apparatus can be functioned by a computer. In this case, the program for causing the computer to function as a target device (the device having the functional configuration shown in the drawings in various embodiments) or each process of the processing procedure (shown in each embodiment) is processed by the computer. A program to be executed by the computer may be downloaded from a recording medium such as a CD-ROM, a magnetic disk, or a semiconductor storage device or via a communication line into the computer, and the program may be executed.

100, 300, 500, 700, 800 Acoustic signal transmitting device 200, 400, 600 Acoustic signal receiving device 103, 203 Storage unit 105, 205 Control unit 107 ADC 109, 209 Frame dividing unit 111, 211 Dividing unit 120, 320 Unclear Conversion units 321 and 421 Filter bank 121 Frequency direction smoothing unit 123 Dynamic range compression unit 125 Time direction smoothing unit 130, 540 Transmission unit 201, 601 Reception unit 207, 327 Gain calculation unit 213 Multiplication unit 215 Time domain conversion unit 217 DAC 219 Amplifier 415 Frame synthesis unit 4-1 First sound collection unit 4-2 Second sound collection unit 4-3 Third sound collection unit 4-4 Fourth sound collection unit 4-5 Fifth sound collection unit 4-6 Sound collection unit 540 Processing target signal generation unit 507 Power spectrum estimation unit 530 Gain coefficient calculation unit 509 Multiplication unit 550, 750, 850 determination unit 704 similarity calculation unit 804 first similarity calculation unit 810 first similarity calculation unit 811 second similarity calculation unit

Claims (10)

A frame dividing unit that divides the acoustic signal collected from the microphone into frames at predetermined intervals;
A dividing unit for dividing the acoustic signal for each frame into a plurality of frequency band acoustic signals;
An obscuring unit for obscuring an acoustic signal divided for each frame;
The obscuring part is
A frequency direction smoothing unit for obtaining a frequency-smoothed power of an input signal;
A dynamic range compression unit that compresses fluctuations in the power of the input signal;
A time direction smoothing unit that temporally smoothes the input signal in a time longer than the frame length is connected in cascade.
An acoustic signal transmitter characterized by the above. A frame dividing unit that divides the acoustic signal collected from the microphone into frames at predetermined intervals;
An obscuring unit for obscuring an acoustic signal divided for each frame;
The obscuring part is
A filter bank that divides an acoustic signal divided for each frame into a plurality of frequency band acoustic signals and obtains a frequency-smoothed power,
A dynamic range compression unit that compresses fluctuations in the power of the input signal;
A time direction smoothing unit that temporally smoothes the input signal in a time longer than the frame length;
An acoustic signal transmitter characterized by the above. The acoustic signal transmission device according to claim 1 or 2,
A determination unit that determines whether sound is generated from a specific position;
When no sound is generated from a specific position, the obscuring unit generates an obscured acoustic signal.
An acoustic signal transmitter characterized by the above. The acoustic signal transmission device according to claim 3,
6 or more sound collection units for collecting sounds in different areas using output signals of a microphone array configured with a plurality of microphones;
A processing target signal generation unit that generates a processing target signal from one or more predetermined microphones or signals from the sound collection unit;
From the signal amount of each collected sound signal obtained by each sound collecting unit, the signal amount of the desired sound source and the signal amount of the other sound source are estimated for each frequency, and the power spectrum estimation unit,
A second gain calculation unit for obtaining a ratio of the signal amount of the desired sound source to the signal amount of all sound sources including the signal amount of the desired sound source (hereinafter referred to as “second gain coefficient”);
The determination unit
Using the second gain coefficient to determine whether sound is generated from a predetermined area;
An acoustic signal transmitter characterized by the above. The acoustic signal transmission device according to claim 3,
Using an acoustic signal collected from a two-channel microphone, and having a similarity calculator for calculating the similarity between channels for each frequency band;
The determination unit determines whether or not sound is generated from a specific region using the similarity;
An acoustic signal transmitter characterized by the above. The acoustic signal transmission device according to claim 3,
A first similarity calculator that calculates the similarity between channels for each frequency band using an acoustic signal collected from a two-channel microphone;
A second similarity calculation unit that calculates a similarity between channels for each frequency band using an acoustic signal collected from a two-channel microphone different from the two-channel microphone of the first similarity calculation unit; Have
The determination unit determines whether sound is generated from a specific region using the similarity obtained from the first similarity calculation unit and the second similarity calculation unit.
An acoustic signal transmitter characterized by the above. An acoustic signal receiving device for receiving an obscured acoustic signal,
Using the obscured acoustic signal, a gain calculation unit that calculates a gain for each frequency;
A frame dividing unit for dividing a predetermined stationary carrier signal into frames every predetermined time;
A dividing unit for dividing the carrier signal for each frame into signals of a plurality of frequency bands;
A multiplier that multiplies the gain and the signal of the frequency band;
A time domain conversion unit that converts a value obtained by the multiplication unit into a time domain signal;
An acoustic signal receiving device. A frame dividing step for dividing the acoustic signal collected from the microphone into frames at predetermined intervals;
A dividing step of dividing the acoustic signal for each frame into a plurality of frequency band acoustic signals;
Comprising an obscuring step for obscuring the acoustic signal divided for each frame;
The obscuring step includes
A frequency direction smoothing step for obtaining a frequency-smoothed power of the input signal;
A dynamic range compression step for compressing fluctuations in the power of the input signal;
A time direction smoothing step for smoothing the input signal in time longer than the frame length is cascaded;
An acoustic signal transmission method characterized by the above. An acoustic signal transmission method according to claim 8,
A determination step for determining whether or not sound is generated from a specific position;
If no sound is generated from a specific position, the obscuring step generates an obscured acoustic signal.
An acoustic signal transmission method characterized by the above.   A program for causing a computer to function as the acoustic signal transmitting device or the acoustic signal receiving device according to any one of claims 1 to 7.

Images