JP4850191B2 - Automatic volume control device and voice communication device using the same - Google Patents

Description

  The present invention relates to an automatic volume control device that performs volume control of an input audio signal and a voice communication device using the same.

  In recent years, with the development of networks such as FTTH (Fiber To The Home), public wireless networks, and high-speed mobile communication networks, voice communication using voice communication devices has increased in various places, and multipoints (such as TV voice conference systems) Voice communication systems capable of calling at three or more locations are becoming popular for the purpose of cost reduction and the like.

  In a multipoint call, the volume level varies depending on the speaker depending on the sensitivity and directivity of the microphone to be used and the distance between the microphone and the speaker. This makes it difficult to adjust the volume on the receiving side.

  The volume that can be adjusted by the user on the receiver side is often only the audio after each speaker's audio is mixed, and if the volume is adjusted to match the volume of one speaker, the volume of the other speaker may not be adjusted. The problem of becoming appropriate is likely to occur.

  In order to solve this, an audio communication system is often equipped with an automatic volume control device (AGC) that adjusts the volume so that the received voice has a predetermined target volume level. . As shown in FIG. 1, the automatic volume control device is a technique for adjusting and outputting the volume of an input audio signal to a predetermined target volume level in the AGC unit 1.

  FIG. 2 is a block diagram showing an example of a receiving side portion of a voice communication apparatus used in a multipoint voice communication system. In the figure, an audio signal from a certain speaker is adjusted to a target volume level by the AGC unit 2 and supplied to the mixing unit 4, and an audio signal from another speaker is adjusted to the target volume level by the AGC unit 3. The signal is supplied to the mixing unit 4 and a synthesized voice signal is output from the mixing unit 4.

  In Patent Document 1, the signal is classified into a voice section and a noise section, and the noise section amplification factor is set to be equal to or lower than the voice section amplification ratio in accordance with the SN ratio, thereby suppressing a sense of harshness due to noise amplification. Is described.

In Patent Document 2, one of a plurality of input signals is used to determine a reference gain, and the gains of other signals are automatically finely adjusted based on the above-described reference gain. It is described that all audio signals can be easily adjusted to the same volume by making a determination.
JP 2003-60459 A JP 2006-287716 A

  Consider a case in which there is a voice speaker with a poor signal-to-noise ratio during a multipoint call at three or more points. As shown in FIG. 3, if the input voice signal of the first speaker has a high S / N ratio, the input voice signal of the second speaker has a medium S / N ratio, and the input voice signal of the third speaker has a low S / N ratio. To do.

  Since each speaker's input speech signal is amplified to the target volume level in the AGC unit, the noise level of the speech signal of the third speaker is larger than the noise level of the speech signals of the first and second speakers. Become. For this reason, the noise level in the audio signal obtained by mixing the amplified audio signals of the first to third speakers is increased, the SN ratio of the audio signals of the first and second speakers is deteriorated, and the voice is heard. There is a problem that it becomes difficult.

  The present invention has been made in view of the above points, suppresses an excessive increase in noise level, and reduces the difficulty of hearing other speakers' voices even if there are speakers with poor signal-to-noise ratios. It is an object of the present invention to provide an automatic sound volume control device and a voice communication device using the same.

An automatic volume control device according to an embodiment of the present invention includes a sound determination unit that determines a sound portion and a non-voice portion of an input sound signal;
A sound level calculating means for calculating a sound level in a sound portion of the input sound signal;
Noise level calculation means for calculating a noise level in a non-voice portion of the input voice signal;
SN ratio calculating means for calculating an SN ratio of the input voice signal from the voice level and the noise level;
An amplification factor calculating means for calculating an amplification factor of the input audio signal from the audio level, the noise level, the SN ratio, and a preset target volume level;
Amplifying means for amplifying and outputting the input audio signal at the amplification factor;
I have a,
The amplification factor calculating means calculates an amplification factor so that the sound level becomes the target sound volume level when the SN ratio is equal to or greater than a threshold value, and the noise level is calculated from the target sound volume level when the SN ratio is less than the threshold value. The amplification factor is calculated so as to be lower by the threshold value .

An automatic volume control device according to another embodiment of the present invention includes a sound determination unit that determines a sound part and a non-sound part of an input sound signal;
A sound level calculating means for calculating a sound level in a sound portion of the input sound signal;
Noise level calculation means for calculating a noise level in a non-voice portion of the input voice signal;
SN ratio calculating means for calculating an SN ratio of the input voice signal from the voice level and the noise level;
The input audio signal is calculated from the audio level, the noise level, the SN ratio calculated by the SN ratio calculating means, the preset target volume level, and the SN ratio supplied from one or more other automatic volume control devices. An amplification factor calculating means for calculating the amplification factor;
Amplifying means for amplifying and outputting the input audio signal at the amplification factor;
I have a,
The amplification factor calculating means is configured such that when the SN ratio calculated by the SN ratio calculating means is equal to or higher than the SN ratio supplied from the one or more other automatic sound volume control devices, the sound level becomes the target sound volume level. When the S / N ratio calculated by the S / N ratio calculating means is less than the S / N ratio supplied from the one or more other automatic volume control devices, the noise level is the highest S / N ratio from the target volume level. The amplification factor is calculated so as to be a low value .

In the automatic volume control device,
The amplification factor calculating means is configured such that when the SN ratio calculated by the SN ratio calculating means is greater than or equal to a threshold value or greater than or equal to an SN ratio supplied from the one or more other automatic volume control devices, the audio level is the target volume. An amplification factor is calculated so as to be a level, and the SN ratio calculated by the SN ratio calculation means is not more than a threshold value, and any of the SN ratios supplied from the one or more other automatic volume control devices is not less than the threshold value. When the amplification factor is calculated so that the noise level is lower than the target volume level by the threshold value, the SN ratio calculated by the SN ratio calculation unit is less than the threshold value and the one or more other automatic volume control devices The noise level is the highest SN ratio among the SN ratios supplied from the target volume level from the one or more other automatic volume control devices when all the S / N ratios supplied from Calculating the amplification factor so that the lower value.

In the automatic volume control device,
The voice level calculation means smoothes and outputs the voice level calculated in the current input voice signal frame with the voice level obtained in the previous input voice signal frame.

In the automatic volume control device,
The noise level calculation means smoothes and outputs the noise level calculated in the current input voice signal frame with the noise level obtained in the previous input voice signal frame.

In the automatic volume control device,
The amplification factor calculating means smoothes and outputs the amplification factor calculated for the current input audio signal frame with the amplification factor obtained for the previous input audio signal frame.

In addition, a voice communication device according to an embodiment of the present invention includes a plurality of the automatic volume control devices,
Mixing means for mixing audio signals output from the plurality of automatic volume control devices;
Have.

  According to the present invention, it is possible to suppress an excessive increase in noise level.

  As a result, even if there is a speaker with a poor S / N ratio, it is possible to reduce the difficulty of listening to the voices of other speakers.

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

<First Embodiment>
FIG. 4 shows a configuration example of the first embodiment of the automatic volume control device. In the figure, an input audio signal from a terminal 11 is supplied to a VAD (Voice Activity Detection) unit 12. Note that the audio signal is a digital audio signal sampled at a sampling frequency of 8 KHz, for example, and 160 samples (corresponding to 20 msec) as one frame.

  The VAD unit 12 is described in, for example, Japanese Patent No. 3849116. For each frame in which an audio signal on which environmental noise is superimposed as an input signal is given in time series, power, zero crossing rate, power spectrum peak frequency, A voice feature value is calculated using a pitch period, etc., a voice feature value is calculated based on only a difference in higher-order components of the peak frequency of the power spectrum, and voice or non-voice (based on the above two feature values) In other words, the determination result is supplied to the audio level update unit 13 and the noise level update unit 14 together with the audio signal.

  When the sound level update unit 13 determines that the sound part is the sound part in the current frame n, the sound level update unit 13 first obtains the sound level V in the current frame by the expression (1).

（１）式において、ｉｎｐｕｔ（ｋ）は入力音声信号の振幅を、Ｍはフレーム長をそれぞれ表す。次に、音声レベルＶと、前回フレームの平均音声レベルＶ＿ａｖｅ（ｎ−１）とを、平滑化係数ＣＯＦ１を用いて平滑化を行い、今回フレームの平均音声レベルＶ＿ａｖｅ（ｎ）を求める。なお、今回フレームが非音声部分と判定された場合は、平均音声レベルＶ＿ａｖｅ（ｎ）は更新しない。この今回フレームの平均音声レベルＶ＿ａｖｅ（ｎ）を増幅率決定部１５に供給する。なお、平滑化係数ＣＯＦ１は例えば０．９０〜０．９９程度の値である。

In equation (1), input (k) represents the amplitude of the input audio signal, and M represents the frame length. Next, the audio level V and the average audio level V_ave (n−1) of the previous frame are smoothed using the smoothing coefficient COF1, and the average audio level V_ave (n) of the current frame is obtained. Note that if it is determined that the current frame is a non-voice portion, the average voice level V_ave (n) is not updated. The average voice level V_ave (n) of the current frame is supplied to the amplification factor determination unit 15. The smoothing coefficient COF1 is a value of about 0.90 to 0.99, for example.

V_ave (n) = V_ave (n−1) × COF1 + V × (1.0−COF1)
... (2)
When it is determined that the current frame n is a non-speech part, the noise level updating unit 14 first obtains the noise level N in the current frame by Equation (3).

次に、雑音レベルＮと、前回フレームの平均雑音レベルＮ＿ａｖｅ（ｎ−１）とを、平滑化係数ＣＯＦ１を用いて平滑化を行い、今回フレームの平均雑音レベルＮ＿ａｖｅ（ｎ）を求める。なお、今回フレームが音声部分と判定された場合は、平均雑音レベルＮ＿ａｖｅ（ｎ）は更新しない。この今回フレームの平均雑音レベルＮ＿ａｖｅ（ｎ）を増幅率決定部１５に供給する。

Next, the noise level N and the average noise level N_ave (n−1) of the previous frame are smoothed using the smoothing coefficient COF1, and the average noise level N_ave (n) of the current frame is obtained. Note that if it is determined that the current frame is an audio part, the average noise level N_ave (n) is not updated. The average noise level N_ave (n) of the current frame is supplied to the amplification factor determination unit 15.

N_ave (n) = N_ave (n−1) × COF1 + N × (1.0−COF1)
... (4)
The amplification factor determination unit 15 calculates the gain of the current frame from the average sound level V_ave (n), the average noise level N_ave (n), and the target volume level that is determined in advance and supplied from the host device via the terminal 16. s_gain is determined and supplied to the gain multiplier 17.

  The gain multiplication unit 17 multiplies the input audio signal supplied from the terminal 11 by the amplification factor from the amplification factor determination unit 15 and outputs an output audio signal output (k) from the terminal 18.

output (k) = input (k) × s_gain
(However, k = 1, 2 ... M) (5)
The amplification factor determination unit 15 may be configured to smoothly vary the amplification factor for each sample as follows, for example, in order to prevent the amplification factor from changing suddenly at the frame boundary. Here, the gain at the kth sample of the n frame is gain (n, k), the instantaneous gain obtained only in the n frame is s_gain, and the smoothing coefficient is COF2. The smoothing coefficient COF2 is a value of about 0.90 to 0.99, for example.

gain (n, k) = gain (n, k−1) × COF 2
+ S_gain × (1.0−COF2) (6)
In this case, the output audio signal output (k) is as follows.

output (k) = input (k) × gain (n, k)
(However, k = 1, 2 ... M) (7)
<Configuration of amplification factor determination unit>
FIG. 5 shows a configuration example of an embodiment of the amplification factor determination unit 15. In the figure, the amplification factor determination unit 15 includes an SN ratio calculation unit 21 and an amplification factor calculation unit 22.

  The SN ratio calculation unit 21 calculates the SN ratio from the average voice level V_ave (n) of the current frame from the voice level update unit 13 and the average noise level N_ave (n) of the current frame from the noise level update unit 14. This is supplied to the amplification factor calculation unit 22.

  The amplification factor calculation unit 22 includes the average audio level V_ave (n) of the current frame from the audio level update unit 13, the average noise level N_ave (n) of the current frame from the noise level update unit 14, and the SN ratio calculation unit 21. The gain s_gain of the current frame is calculated from the S / N ratio and the target volume level from the host device.

<Operation of Gain Calculation Unit 22>
FIG. 6 shows a flowchart of an example of processing executed by the amplification factor calculation unit 22. In step S1, the amplification factor calculation unit 22 determines whether or not the SN ratio supplied from the SN ratio calculation unit 21 is equal to or greater than a threshold value. Here, the threshold is set in advance as a value of about 12 dB, for example.

  If the S / N ratio is equal to or greater than the threshold, the gain s_gain of the current frame is calculated in step S2 such that the average audio level V_ave (n) of the current frame becomes the target volume level.

  On the other hand, if the SN ratio is less than the threshold value, the gain s_gain of the current frame is set so that the average noise level N_ave (n) of the current frame becomes a value obtained by subtracting the threshold value from the target sound volume level (target sound volume level−threshold) in step S3. Is calculated.

<Configuration of voice communication device of multipoint voice communication system>
FIG. 7 shows a block diagram of the first embodiment of the receiving side portion of the voice communication apparatus in the multipoint voice communication system. In the figure, encoded speech signals received from a plurality of speakers are supplied to terminals 21-1 to 21-n of the speech communication apparatus 20, and each encoded speech signal is transmitted to speech decoding units 22-1 to 22-n. Decoded at each. Each decoded audio signal is supplied to each of the AGC units 23-1 to 23-n.

  Each of the AGC units 23-1 to 23-n is an automatic volume control device that performs the operation shown in FIG. 6 with the configuration shown in FIGS. 4 and 5, and controls the volume of each audio signal to mix the output audio signal. 24. The mixing unit 24 mixes the audio signals supplied from the AGC units 23-1 to 23-n and outputs the mixed audio signals from the terminal 25.

<Specific volume control operation>
Here, a specific sound volume control operation when n = 3 in FIG. 7 will be described. The target sound volume level given to each of the AGC units 23-1 to 23-3 is set to -24 dBov (ov: overload, dB display of how much smaller than the maximum value), and the SN ratio threshold = 12 dB.

  As shown in FIG. 8, the voice level of the first speaker input to the AGC unit 23-1 is −30 dBov, the noise level is −54 dBov, the SN ratio is 24 dB, and the input is input to the AGC unit 23-2. The second speaker's voice level is -42 dBov, the noise level is -48 dBov, the S / N ratio is 6 dB, and the third speaker's voice level input to the AGC unit 23-3 is -18 dBov, and the noise level is Is −36 dBov and the S / N ratio is 18 dB.

  In the AGC unit 23-1, since the SN ratio = 24 dB is equal to or greater than the threshold = 12 dB, the amplification factor = 6 dB (amplification) is calculated so that the audio level = −30 dBov is set to the target volume level = −24 dBov. As a result, the voice level of the first speaker output from the AGC unit 23-1 is -24 dBov, the noise level is -48 dBov, and the SN ratio is 24 dB.

  In the AGC unit 23-2, since the SN ratio = 6 dB is less than the threshold value = 12 dB, the amplification factor = 12 dB (amplification) is calculated so that the noise level = −48 dBov is set to the target volume level−12 dB = −36 dBov. . As a result, the voice level of the second speaker output from the AGC unit 23-2 is -30 dBov, the noise level is -36 dBov, and the SN ratio is 6 dB.

  In the AGC unit 23-3, since the SN ratio = 18 dB is equal to or greater than the threshold = 12 dB, the amplification factor = −6 dB (attenuation) is calculated so that the audio level = −18 dBov is set to the target sound volume level = −24 dBov. As a result, the voice level of the third speaker output from the AGC unit 23-3 is -24 dBov, the noise level is -42 dBov, and the SN ratio is 18 dB.

  Thus, the gain is determined so that the voice level becomes the target volume level for the speaker having a good S / N ratio, and the threshold value is subtracted from the target volume level for the speaker having a poor S / N ratio. By determining the gain so that it does not become larger than the value, it is possible to secure a certain S / N ratio for the first speaker and the third speaker, and the voices of other speakers are affected by the influence of the second speaker having a poor S / N ratio. A good call can be made without causing the problem of difficulty in hearing.

  On the other hand, in the conventional technique that only adjusts the input voice to the target volume level, as shown in FIG. 9, the target volume level is set to −24 dBov, and the same as in FIG. 8 from the first to third speakers. When there is an input voice, the amplification factor of the first and third speakers is not different from the example of FIG. 8, but the amplification factor of the second speaker is 18 dB and the noise level is −30 dBov. 1. It becomes difficult to hear the voice of the third speaker.

Second Embodiment
FIG. 10 shows a configuration example of the second embodiment of the automatic volume control device. In the figure, the same parts as those in FIG.

  In FIG. 10, an input voice signal from the terminal 11 is supplied to a VAD (Voice Activity Detection) unit 12. Note that the audio signal is a digital audio signal sampled at a sampling frequency of 8 KHz, for example, and 160 samples (corresponding to 20 msec) as one frame.

  The VAD unit 12 is described in, for example, Japanese Patent No. 3849116. For each frame in which an audio signal on which environmental noise is superimposed as an input signal is given in time series, power, zero crossing rate, power spectrum peak frequency, A voice feature value is calculated using a pitch period, etc., a voice feature value is calculated based on only a difference in higher-order components of the peak frequency of the power spectrum, and voice or non-voice (based on the above two feature values) In other words, the determination result is supplied to the audio level update unit 13 and the noise level update unit 14 together with the audio signal.

When the sound level update unit 13 determines that the sound part is the sound part in the current frame n, the sound level update unit 13 first obtains the sound level V in the current frame by the expression (1).
Next, the audio level V and the average audio level V_ave (n−1) of the previous frame are smoothed using the smoothing coefficient COF1, and the average audio level V_ave (n) of the current frame is obtained. Note that if it is determined that the current frame is a non-voice portion, the average voice level V_ave (n) is not updated. The average voice level V_ave (n) of the current frame is supplied to the amplification factor determination unit 35. The smoothing coefficient COF1 is a value of about 0.90 to 0.99, for example.

V_ave (n) = V_ave (n−1) × COF1 + V × (1.0−COF1)
... (2)
When it is determined that the current frame n is a non-speech part, the noise level updating unit 14 first obtains the noise level N in the current frame by Equation (3).
Next, the noise level N and the average noise level N_ave (n−1) of the previous frame are smoothed using the smoothing coefficient COF1, and the average noise level N_ave (n) of the current frame is obtained. Note that if it is determined that the current frame is an audio part, the average noise level N_ave (n) is not updated. The average noise level N_ave (n) of the current frame is supplied to the amplification factor determination unit 35.

N_ave (n) = N_ave (n−1) × COF1 + N × (1.0−COF1)
... (4)
The amplification factor determination unit 35 receives the average sound level V_ave (n), the average noise level N_ave (n), the target volume level that is determined in advance and supplied from the host device via the terminal 16, and the terminal 36. The gain s_gain of the current frame is determined from the S / N ratio of the other speakers supplied from the other automatic volume control device supplied in this way, and supplied to the gain multiplier 17.

  The gain multiplying unit 17 multiplies the input audio signal supplied from the terminal 11 by the amplification factor from the amplification factor determining unit 35 and outputs the output audio signal output (k) from the terminal 18.

output (k) = input (k) × s_gain
(However, k = 1, 2 ... M) (5)
Note that the amplification factor determination unit 35 may be configured to smoothly vary the amplification factor for each sample as follows, for example, in order to prevent the amplification factor from changing suddenly at the frame boundary. Here, the gain at the kth sample of the n frame is gain (n, k), the instantaneous gain obtained only in the n frame is s_gain, and the smoothing coefficient is COF2. The smoothing coefficient COF2 is a value of about 0.90 to 0.99, for example.

gain (n, k) = gain (n, k−1) × COF 2
+ S_gain × (1.0−COF2) (6)
In this case, the output audio signal output (k) is as follows.

output (k) = input (k) × gain (n, k)
(However, k = 1, 2 ... M) (7)
<Configuration of amplification factor determination unit>
FIG. 11 shows a configuration example of an embodiment of the amplification factor determination unit 35. In the figure, the same parts as those in FIG. In FIG. 11, the amplification factor determination unit 35 includes an SN ratio calculation unit 21, an SN ratio comparison unit 37, and an amplification factor calculation unit 38.

  The SN ratio calculation unit 21 calculates the SN ratio from the average voice level V_ave (n) of the current frame from the voice level update unit 13 and the average noise level N_ave (n) of the current frame from the noise level update unit 14. The signal is supplied to the SN ratio comparison unit 37.

  The S / N ratio comparison unit 37 compares the S / N ratio of the speaker of the own device calculated by the S / N ratio calculation unit 21 with the threshold value and the S / N ratio of the other speakers, and compares the comparison result with the S / N ratio of the speaker of the own device. The threshold value and the SN ratio of other speakers are supplied to the amplification factor calculation unit 38.

  The amplification factor calculation unit 38 includes the average audio level V_ave (n) of the current frame from the audio level update unit 13, the average noise level N_ave (n) of the current frame from the noise level update unit 14, and the SN ratio calculation unit 21. , The target volume level from the host device, the comparison result from the SN ratio comparison unit 37, the SN ratio of the speaker of the own device, the threshold value, and the SN ratio of one or more other speakers. From this, the gain s_gain of the current frame is calculated.

<Operation of Gain Calculation Unit 38>
FIG. 12 shows a flowchart of an example of processing executed by the amplification factor calculation unit 38. In the figure, in step S11, the amplification factor calculation unit 38 determines that the SN ratio of the speaker of the own device is one of the SN ratios of one or more other speakers based on the comparison result supplied from the SN ratio calculation unit 21. If the S / N ratio of the speaker of the apparatus is the highest, the gain s_gain of the current frame so that the average voice level V_ave (n) of the current frame becomes the target volume level in step S12. Is calculated.

  On the other hand, if the S / N ratio of the other speaker is higher than the S / N ratio of the speaker of the own device, in step S13, the average noise level N_ave (n) of the current frame is the other story having the highest S / N ratio from the target volume level. The gain s_gain of the current frame is calculated so as to be a value obtained by subtracting the SN ratio of the speaker (the target sound volume level-the SN ratio of the speaker having the maximum SN ratio).

<Configuration of voice communication device of multipoint voice communication system>
FIG. 13 is a block diagram of a second embodiment of the receiving side portion of the voice communication apparatus in the multipoint voice communication system. In the figure, the same parts as those in FIG.

  In FIG. 13, encoded speech signals received from a plurality of speakers are supplied to terminals 21-1 to 21-n of the speech communication device 20, and each encoded speech signal is transmitted to speech decoding units 22-1 to 22-n. Decoded at each. Each decoded audio signal is supplied to each of the AGC units 43-1 to 43-n.

  Each of the AGC units 43-1 to 43-n is an automatic volume control device that performs the operation shown in FIG. 12 (or FIG. 15) with the configuration shown in FIGS. 10 and 11, and controls the volume of each audio signal and outputs it. The audio signal is supplied to the mixing unit 24, and the SN ratio of the speaker of the own apparatus is obtained and supplied to all other automatic volume control apparatuses. The mixing unit 24 mixes the audio signals supplied from the AGC units 43-1 to 43-n and outputs the mixed audio signals from the terminal 25.

<Specific volume control operation>
Here, a specific sound volume control operation when n = 3 in FIG. 13 will be described. It is assumed that the target volume level given to each of the AGC units 43-1 to 23-3 is -24 dBov (ov: overload, dB display of how much smaller than the maximum value), and the SN ratio threshold = 12 dB.

  As shown in FIG. 14, the voice level of the first speaker input to the AGC unit 43-1 is −30 dBov, the noise level is −40 dBov, the SN ratio is 10 dB, and the input is input to the AGC unit 43-2. The second speaker's voice level is -42 dBov, the noise level is -48 dBov, the SN ratio is 6 dB, and the third speaker's voice level input to the AGC unit 43-3 is -18 dBov, and the noise level Is −26 dBov and the S / N ratio is 8 dB.

  In this case, the SN ratio of the first speaker output from the AGC unit 43-1 is the highest. In the AGC unit 43-1, since the SN ratio is the highest, the amplification factor = 6 dB (amplification) is calculated so that the audio level = −30 dBov is set to the target sound volume level = −24 dBov. As a result, the voice level of the first speaker output from the AGC unit 43-1 is -24 dBov, the noise level is -34 dBov, and the SN ratio is 10 dB.

  In the AGC unit 43-2, since the SN ratio (10 dB) of the first speaker is higher than the SN ratio (6 dB) of the second speaker of the own apparatus, the noise level = −48 dBov is set to the target volume level (−24 dBov) − Amplification factor = 14 dB (amplification) is calculated so that the SN ratio of one speaker (10 dB) = − 34 dBov. As a result, the voice level of the second speaker output from the AGC unit 43-2 is -28 dBov, the noise level is -34 dBov, and the SN ratio is 6 dB.

  In the AGC unit 43-3, since the SN ratio (10 dB) of the first speaker is higher than the SN ratio (8 dB) of the third speaker of the own apparatus, the noise level = −26 dBov is set to the target volume level (−24 dBov) − Amplification factor = −8 dB (attenuation) is calculated so that the SN ratio of one speaker (10 dB) = − 34 dBov. As a result, the voice level of the third speaker output from the AGC unit 43-3 is -26 dBov, the noise level is -34 dBov, and the SN ratio is 8 dB.

  As described above, the gain is determined so that the voice level becomes the target volume level for the speaker having a good S / N ratio, and the noise level is the highest from the target volume level to the speaker for the speaker having a poor S / N ratio. By determining the gain so that it does not become larger than the value obtained by subtracting the SN ratio of the high speaker, the SN ratio of the speaker having the highest SN ratio can be maintained. A good call can be made without causing a problem that it is difficult to hear the voice of the speaker.

<Other operations of amplification factor calculation unit 38>
FIG. 15 shows a flowchart of another example of processing executed by the amplification factor calculation unit 38. In the figure, in step S <b> 21, the amplification factor calculation unit 38 determines whether or not the SN ratio is equal to or greater than a threshold value from the comparison result supplied from the SN ratio calculation unit 21. Here, the threshold is set in advance as a value of about 12 dB, for example.

  If the S / N ratio is equal to or greater than the threshold, the gain s_gain of the current frame is calculated in step S22 such that the average sound level V_ave (n) of the current frame becomes the target volume level.

  On the other hand, if the SN ratio is less than the threshold value, it is determined in step S23 whether or not the SN ratio of the speaker of the own device is the highest among the SN ratios of one or more other speakers. If the S / N ratio of the user is the highest, the gain s_gain of the current frame is calculated in step S22 so that the average sound level V_ave (n) of the current frame becomes the target volume level.

  If the SN ratio of the other speaker is higher than the SN ratio of the speaker of the own device in step S23, it is determined in step S24 whether or not there is another speaker whose SN ratio is greater than or equal to the threshold. If there is another speaker, the gain s_gain of the current frame such that the average noise level N_ave (n) of the current frame becomes a value obtained by subtracting the threshold value from the target sound volume level (target sound volume level−threshold) in step S25. Is calculated.

  If there is no other speaker exceeding the threshold value in step S24, in step S26, the average noise level N_ave (n) of the current frame is obtained by subtracting the SN ratio of the other speaker having the highest SN ratio from the target volume level. The gain s_gain of the current frame is calculated so as to be a value (target volume level-SN ratio of the speaker having the maximum SN ratio).

  That is, when the S / N ratio of all the speakers is less than the threshold, the S / N ratio of the speaker having the highest S / N ratio is regarded as the threshold, and the S / N ratios of all automatic volume control devices are matched to the S / N ratio of the highest speaker. I am doing so.

  As described above, the gain is determined so that the voice level becomes the target volume level for the speaker having a good S / N ratio, and the noise level is the highest from the target volume level to the speaker for the speaker having a poor S / N ratio. By determining the gain so as not to be larger than the value obtained by subtracting the SN ratio of a high speaker, an SN ratio of a certain level or more can be ensured for a speaker having a good SN ratio, and other factors are affected by the influence of the speaker having a poor SN ratio. A good call can be made without causing a problem that it is difficult to hear the voice of the speaker.

It is a block diagram of an example of the conventional automatic volume control apparatus. It is a block diagram of an example of the receiving side part of the conventional voice communication apparatus. It is a figure for demonstrating the problem of the conventional voice communication apparatus. It is a figure which shows the structural example of 1st Embodiment of an automatic volume control apparatus. 3 is a diagram illustrating a configuration example of an embodiment of an amplification factor determination unit 15. FIG. 5 is a flowchart of an example of processing executed by an amplification factor calculation unit 22; It is a block diagram of 1st Embodiment of the receiving side part of the audio | voice communication apparatus in a multipoint audio | voice communication system. It is a figure for demonstrating the mode of automatic volume control of 1st Embodiment. It is a figure for demonstrating the mode of the conventional automatic volume control. It is a figure which shows the structural example of 2nd Embodiment of an automatic volume control apparatus. 4 is a diagram illustrating a configuration example of an embodiment of an amplification factor determination unit 35. FIG. 5 is a flowchart illustrating an example of processing executed by an amplification factor calculation unit. It is a block diagram of 2nd Embodiment of the receiving side part of the audio | voice communication apparatus in a multipoint audio | voice communication system. It is a figure for demonstrating the mode of automatic volume control of 2nd Embodiment. 10 is a flowchart of another example of processing executed by an amplification factor calculation unit 38.

Explanation of symbols

12 VAD unit 13 Audio level update unit 14 Noise level update unit 15, 35 Gain determination unit 17 Gain multiplication unit 21 SN ratio calculation unit 22, 38 Gain calculation unit 23-1 to 23-n, 43-1 to 43- n AGC unit 24 mixing unit 37 SN ratio comparison unit 12 service area

Claims (7)

Voice determination means for determining a voice portion and a non-voice portion of an input voice signal;
A sound level calculating means for calculating a sound level in a sound portion of the input sound signal;
Noise level calculation means for calculating a noise level in a non-voice portion of the input voice signal;
SN ratio calculating means for calculating an SN ratio of the input voice signal from the voice level and the noise level;
Amplification factor calculating means for calculating an amplification factor of the input audio signal from the audio level, the noise level, the SN ratio, and a preset target volume level;
Amplifying means for amplifying and outputting the input audio signal at the amplification factor;
I have a,
The amplification factor calculating means calculates an amplification factor so that the sound level becomes the target sound volume level when the SN ratio is equal to or greater than a threshold value, and the noise level is calculated from the target sound volume level when the SN ratio is less than the threshold value. The amplification factor is calculated so as to be a value lower by the threshold value.
An automatic volume control device characterized by that. Voice determination means for determining a voice portion and a non-voice portion of an input voice signal;
A sound level calculating means for calculating a sound level in a sound portion of the input sound signal;
Noise level calculation means for calculating a noise level in a non-voice portion of the input voice signal;
SN ratio calculating means for calculating an SN ratio of the input voice signal from the voice level and the noise level;
The input audio signal is calculated from the audio level, the noise level, the SN ratio calculated by the SN ratio calculating means, the preset target volume level, and the SN ratio supplied from one or more other automatic volume control devices. An amplification factor calculating means for calculating the amplification factor;
Amplifying means for amplifying and outputting the input audio signal at the amplification factor;
I have a,
The amplification factor calculating means is configured such that when the SN ratio calculated by the SN ratio calculating means is equal to or higher than the SN ratio supplied from the one or more other automatic sound volume control devices, the sound level becomes the target sound volume level. When the S / N ratio calculated by the S / N ratio calculating means is less than the S / N ratio supplied from the one or more other automatic volume control devices, the noise level is the highest S / N ratio from the target volume level. Calculate the amplification factor so that the value is as low as possible,
An automatic volume control device characterized by that. The automatic volume control device according to claim 2 ,
The amplification factor calculating means is configured such that when the SN ratio calculated by the SN ratio calculating means is greater than or equal to a threshold value or greater than or equal to an SN ratio supplied from the one or more other automatic volume control devices, the audio level is the target volume. An amplification factor is calculated so as to be a level, and the SN ratio calculated by the SN ratio calculation means is not more than a threshold value, and any of the SN ratios supplied from the one or more other automatic volume control devices is not less than the threshold value. When the amplification factor is calculated so that the noise level is lower than the target volume level by the threshold value, the SN ratio calculated by the SN ratio calculation unit is less than the threshold value and the one or more other automatic volume control devices The noise level is the highest SN ratio among the SN ratios supplied from the target volume level from the one or more other automatic volume control devices when all the S / N ratios supplied from The amplification factor is calculated as a lower value,
An automatic volume control device characterized by that. The automatic volume control device according to any one of claims 1 to 3 ,
The sound level calculating means smoothes and outputs the sound level calculated in the current input sound signal frame with the sound level obtained in the previous input sound signal frame,
An automatic volume control device characterized by that. The automatic volume control device according to any one of claims 1 to 4 ,
The noise level calculation means smoothes and outputs the noise level calculated in the current input voice signal frame with the noise level obtained in the previous input voice signal frame,
An automatic volume control device characterized by that. The automatic volume control device according to any one of claims 1 to 5 ,
The amplification factor calculating means smoothes and outputs the amplification factor calculated in the current input audio signal frame by the amplification factor obtained in the previous input audio signal frame,
An automatic volume control device characterized by that. A plurality of automatic volume control devices according to any one of claims 1 to 6 ,
Mixing means for mixing audio signals output from the plurality of automatic volume control devices;
A voice communication device comprising:

Images