JP3441112B2 - Multipoint communication controller - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a multipoint communication control apparatus for realizing a multipoint video conference using video conference terminals at three or more points.

[0002]

2. Description of the Related Art FIG. 13 shows, for example, Japanese Patent Laid-Open No. 1-286689.
FIG. 11 is a block diagram showing a conventional multipoint communication control device disclosed in Japanese Patent Publication No. FIG. 14 is a block diagram showing a voice processing unit of a conventional multipoint communication control device. In FIG. 13, 1 is the multipoint communication control device, and 2 is a system control unit that controls and manages the whole. Reference numeral 3 denotes a line interface unit (hereinafter referred to as line I / F unit) for connecting each video conference terminal (not shown).

Denoted at 4a to 4m are demultiplexing units for demultiplexing the multiplexed data received from each of the video conference terminals into audio data, video data and general-purpose data. Reference numerals 5a to 5m are multiplexing units for multiplexing audio data, video data, and general-purpose data transmitted to each video conference terminal.

Reference numeral 6 is an audio processing unit for adding the audio data separated by the separating units 4a to 4m and sending it to the multiplexing units 5a to 5m, and 7 is 1 of the video data separated by the separating units 4a to 4m. A general-purpose data processing unit 8 selects one of the general-purpose data and sends it to the multiplex units 5a to 5m by selecting one of the general-purpose data separated by each of the separation units 4a to 4m. It is a department.

FIG. 14 is a block diagram showing the structure of the voice processing unit 6. In the figure, 9a to 9m are separation parts 4
a voice decoding unit that decodes the non-linear (coded) voice data separated in a to 4 m into linear data.
Reference characters a to 10m are voice coding units that code the linear data to which the voice is added to the non-linear voice data. Reference numeral 11 is a voice detection unit that generates voice detection information for performing gain control in the voice addition unit 12 from each linear voice data sent from the voice decoding units 9a to 9m, and 12 is the decoded voice. It is a voice addition unit that adds data and controls the gain so that the level becomes appropriate. Further, FIG. 15 is a graph showing gain control of output audio data with respect to changes in input audio data in a conventional audio processing unit. In the figure, the voice output target value is a target value to which the gain is controlled when the gain is controlled. The AGC voice detection threshold is a threshold for determining whether to perform gain control on valid voice data.

Next, the operation will be described. In FIG. 13, among the video conference terminals connected to the line I / F unit 3 via the line, the video conference terminal (parent terminal) that intends to hold the conference is connected to the multipoint control unit 1. Instruct whether to hold a multipoint video conference with the video conference terminal.

When the system control unit 2 receives the request, it instructs the line I / F unit 3 to call the specified video conference terminal. When the line between each called video conference terminal and the multipoint communication control device 1 is connected, transmission and reception of multiple data in which audio data, video data and general-purpose data are multiplexed is performed, and a multipoint television is transmitted. The meeting starts.

[0008] The multiplexed data transmitted from the video conference terminal is received by the line I / F unit 3 and then passed to the corresponding separating units 4a to 4m, where voice data,
Video data and general-purpose data are separated, and audio data is transmitted to the audio processing unit 6, video data to the video processing unit 7, and general-purpose data to the general-purpose data processing unit 8.

In the voice processing unit 6, the voice data from the respective separating units 4a to 4m are converted into the corresponding voice decoding units 9a to 9a in FIG.
It is received at 9 m, decoded from non-linear data to linear data, and sent to the voice detection unit 11. The voice detection unit 11 generates voice detection information for performing gain control in the voice addition unit 12, and passes it to the voice addition unit 12. The voice adding unit 12 adds the voice data received from the voice detecting unit, and controls the gain so that the voice data added to the preset voice output target value level approaches based on the received voice detection information. The gain-controlled audio data is sent to the audio encoding units 10a to 10m, encoded from linear data to nonlinear data, sent to the multiplexing units 5a to 5m in FIG. 13, and sent to each video conference terminal. It

Next, the conventional gain control will be described with reference to FIG. In FIG. 15, the section A is the voice detection unit 11
Indicates that the voice level is detected to be higher than the voice detection threshold value for AGC. In this case, the voice detection signal 300 is output from the voice detection unit 11 to the voice addition unit 12. The voice adding unit 12 controls the voice level of the section A so as to approach the voice output target value.

Next, in section B, the audio level is AG
It is below the C voice detection threshold value, and it is determined that there is no sound. In this case, since the gain used in the section A is retained as it is, the voice addition unit performs an operation to increase the gain also in the section B.

Next, section C shows a case where the voice level is above the AGC voice detection threshold value. Also in this case, the voice detection signal 300 is output to the voice addition unit 12. In this example, the voice level in the section C is the same as the voice output target value, so that the gain is set to 1 in the voice adder.

In section D, the voice level is lower than the AGC voice detection threshold value, but the gain (=
Since 1) is held as it is, the voice level of the section D is not changed in the voice adding unit 12. Next, in the section E, since the voice level is higher than the AGC voice detection threshold value and higher than the voice output target value, a gain of − is given in the voice adding unit, and the voice level is the voice level. It can be matched with the target value.

Further, in the section F, the gain is below the AGC voice detection threshold value. However, since the gain of-is given in the section E, the gain is held as it is in the section F and the noise is kept. The level of can be lowered.

[0015]

Since the conventional multipoint-to-point communication control apparatus has the above-described voice processing unit, no processing is performed on noise, and the voice adding unit 1 is not used.
When a gain of 1 or more is added in order to obtain an appropriate voice level in 2, the noise level becomes high and the voice becomes difficult to hear. For example, in the section B shown in FIG. 15, the voice level is lower than the AGC voice detection threshold value, and the gain is maintained at 1 or more even though it is determined to be noise, so that There was a problem of raising the level.

The present invention has been made to solve the above problems, and adds a gain of 1 or less to a voice signal of a level determined to be noise by a noise control unit, and a voice addition unit. It is an object of the present invention to obtain a multipoint communication control device that keeps the input noise level or lower even if a gain of 1 or more is added.

[0017]

In a multipoint communication control apparatus according to the present invention, a noise control section is provided in a voice processing section together with a voice detection section, and a noise for determining whether or not voice input data is noise. A voice detection threshold for control is provided, and a voice input signal below the voice detection threshold for noise control is added with a gain of 1 or less to lower the voice level.

A multipoint communication control apparatus according to the present invention is
The feature is that the noise level is reduced according to the number of terminals to which voice is added.

The multipoint communication control apparatus according to the present invention is
The feature is that the noise level is reduced to a level divided by the number of terminals for voice addition.

A multipoint communication control apparatus according to the present invention is
It is characterized in that the noise control voice detection threshold value used by the noise control unit is adaptively changed.

The multipoint communication control apparatus according to the present invention is
It is characterized in that the noise control unit sequentially processes the audio data from a plurality of video conference terminals in time division for each channel.

A multipoint communication control apparatus according to the present invention comprises:
A noise control unit compares a voice data level with a noise control voice detection threshold value and a comparison unit that sets a plurality of samples as one block and determines whether the noise is in block units.
And a second determining unit for determining whether or not there is noise across a plurality of blocks.

The multipoint communication control apparatus according to the present invention is
When the decoding unit decodes the voice data into low frequency band data and high frequency band data, the noise control unit uses the low frequency band data to determine whether or not it is noise.

[0024]

Since the voice processing unit according to the present invention performs voice addition with the noise level being lowered by the noise control unit, there is no possibility of adding noise as in the conventional case and raising the noise level. Good audio can be obtained in point video conferences.

[0025]

EXAMPLES Example 1. An embodiment of the present invention will be described below with reference to the drawings. In FIG. 1, reference numeral 6 is a voice processing unit in the multipoint communication control device. Denoted at 9a to 9m are decoding units for decoding the non-linear audio data from each channel into linear data, and denoted at 10a to 10m are coding units for encoding the linear audio data to each channel into non-linear audio data. 1
Reference numeral 1 is voice that generates voice detection information necessary for gain control in the voice adder 12 from the linear voice data for each channel sent from the voice decoders 9a to 9m, and passes the voice detection information to the voice adder 12. It is a detection unit. Reference numeral 12 is a voice addition unit that adds voice data from each video conference terminal and then controls the gain so as to reach the level of the voice output target value for each channel based on the voice detection information from the voice detection unit. Thirteen
Is a noise control unit that has a voice threshold value for determining whether the voice input data is noise or not, determines the noise for each channel, and reduces only the noise.

Further, in FIG. 1, reference numerals 100a to 100m are audio data in which 4-channel data output from the decoding units 9a to 9m are multiplexed. Also, 200a-2
00 m is voice data whose noise is reduced by the noise control unit 13. A voice detection signal 300 is output by the voice detection unit 11. 400a to 400m should be at the level of the voice output target value output from the voice addition unit 12.
Is audio data that has been controlled gain.

Decoding sections 9a to 9m and coding sections 10a to 1
0m decodes or encodes 4-channel data. The voice detection unit 11 and the noise control unit 13 input voice data of four channels, detect voices for the respective channels, and control noise. The voice addition unit adds the input voice data and subtracts the voice component of its own channel from the added voice, and then performs gain control so that the gain reaches the target level of voice output, and outputs the gain to the encoding unit.

Next, FIG. 2 is a diagram showing the control of the gain with respect to the fluctuation of the voice input level. This gain control is performed for each channel. Further, in FIG. 2, the voice output target value is a target value when the gain is controlled and the human voice is set to an appropriate level. This voice output target value is the voice addition unit 12
Is the target value of the audio data output to the encoding units 10a to 10m. The AGC voice detection threshold value is a threshold value that determines the voiced detection information to be passed from the voice detection unit 11 to the voice addition unit 12 in FIG. The voice input level is this AG
When it exceeds the C voice detection threshold value, the voice detection unit 11 outputs the voice detection signal 300 to the voice addition unit 12. The noise control voice detection threshold is a threshold for determining whether noise is present or not in the noise controller 13 and reducing only noise.

Next, the operation will be described. The operation up to the start of the multipoint video conference is the same as in the conventional example, and will not be repeated. The voice detection unit 11 determines whether the voice data sent from the decoding units 9a to 9m is voice input data having a level higher than the AGC voice detection threshold value, and sends the information to the voice addition unit 12. . Further, the noise control unit 13 determines whether the voice data received from the voice detection unit 11 is noise lower than the noise control voice detection threshold value or not, and as shown in sections B and G of FIG. If it is noise, a gain of 1 or less is added to lower the level. Also,
If it is not noise, the data is sent to the voice addition unit 12 with a gain of 1 as in the section C. The voice adding unit 12 adds voices based on the voice detection information from the voice detecting unit 11, and controls the gain so that the added voice data has a voice output target value. The gain control is performed only for the voice data whose level is higher than the AGC voice detection threshold, as in the sections A and E. As shown in FIG. 2, a gain of 1 or more is added to the voice data that is lower than the output target value and higher than the AGC voice detection threshold, as in the section A, to obtain the output target value level. In addition, as in the section E, a gain of 1 or less is added to the audio data higher than the output target value to bring it to the level of the output target value. As in sections D and F, A
The voice data having a level lower than the voice detection threshold for GC retains the previous gain to eliminate a sudden change in the gain and reduce the discomfort.

Next, the voice processing unit 6 will be further described with reference to FIG. In FIG. 3, reference numeral 20 is a receiving side voice highway for receiving separated voice, 9a is a decoding unit for decoding the voice received from the receiving side voice highway 20, and 24 is a decoding unit 9a. A serial-parallel converter for converting the voice data into a serial-parallel converter, a dual port memory 25 accessible from two directions, a voice detector 11 for detecting voice, and a noise controller 13 for controlling noise. The decoding unit 9a is composed of a digital signal processor # 1 (DSP # 1) 22, and the voice detection unit 11 and the noise control unit 13 are also composed of a digital signal processor (DSP) 26.
Reference numeral 28 is a division unit for time-dividing 8-bit data into 4-bit data, and 30 is a voice addition highway for outputting voice data to the voice addition unit 12.

Reference numeral 31 is a voice addition highway for inputting voices added from the voice addition unit 12, 32 is first-out,
First-in-first-out FIFO, 34 is a parallel-to-serial converter that performs parallel-to-serial conversion, and 33 is the FIFO 32 and parallel-to-serial converter 34
Is a transmission control unit that controls the input / output of data with respect to.
35 is a counter that generates timing by inputting a 2 MHz clock and a timing pulse, 2
Reference numerals 3 and 27 are timing generators that generate timing based on the count from the counter 35. Reference numeral 10a is an encoding unit for encoding the audio data to be transmitted, and 36 is a digital signal processor (D
SP # 1). Reference numeral 21 is a transmitting side audio highway for transmitting the audio to be transmitted to the multiplexing unit.

Next, a timing chart of each part shown in FIG. 3 will be described with reference to FIGS. 4 to 7. FIG. 4 is a diagram showing encoded data # 1 input by the decoding unit 9a. Further, FIG. 4 shows the coded data # 1 output from the coding unit 10a at the same time. As shown in FIG. 4, the decoding unit 9a receives the FSR # 1 signal from the timing generation unit 23 and detects the beginning of the encoded data # 1. When the head of the coded data # 1 is detected, the decoding unit 9a inputs the coded data of that channel and performs decoding according to a predetermined rule. There are cases where the encoded data is 56 kbps or 64 kbps, and in either case, the decoding unit 9a decodes the non-linear audio data into linear audio data.

The decoding unit 9a divides the audio data into low-frequency data and high-frequency data for each channel to perform decoding. The low frequency data is data in a low frequency band, and the high frequency data is data in a high frequency band. 16 bits are used for low band data and 16 bits are used for high band data. Therefore, data is decoded using 32 bits for one channel. In this way, the data output from the decoding unit 9a becomes 32-bit linear data for one channel. The 32-bit linear data decoded by the decoding unit is multiplexed with the 32-bit linear data output from the other three decoding units and input to the serial / parallel conversion unit. This serial-parallel converter 24
The data input to is shown in FIG.

FIG. 5 is a diagram showing a case where the 32-bit linear data output from the decoding section is multiplexed four times. This 4
Here, the multiplexed linear data is referred to as linear received data. The linear reception data is composed of low band data from channel 1 to channel 4 and high band data from channel 1 to channel 4. As shown in the figure, low frequency band data from channel 1 to channel 4 continues, and then high frequency data from channel 1 to channel 4 continues. This data is synchronized with the 8K frame palace, and the low frequency data and the high frequency data are output in synchronization with the FSX # 1 signal output from the timing generation unit 23. Further, the data of each channel of low band data and high band data is composed of 16 bits. For example, the low-frequency linear reception data of channel 1 is composed of 16 bits. In this way, the linear received data is converted into 8-bit parallel data via the serial / parallel converter 24 and input to the dual port RAM 25. The voice detection unit and the noise control unit perform voice detection and noise control on the voice data input to the dual port RAM 25, and output the voice data to the voice addition highway 30.

FIG. 6 is a diagram showing the timing of transmitting and receiving data on the voice addition highway. The voice detection unit 11 and the noise control unit 13 shown in FIG. 3 perform voice detection and noise control for channels 1 to 4. Therefore, by providing six circuits as shown in FIG. 3, 24-channel data can be handled. It is assumed that the voice addition highway can transmit and receive data of, for example, a maximum of 24 channels. In such a case, as shown in FIG. 6, low-frequency data from channel 1 to channel 8 is output to the audio addition highway, then low-frequency data from channel 9 to channel 16 is output, and then, The low frequency data of channel 17 is output from channel 17. Channel 1 after a further open time
8 to 8 are output, high channel data of channel 9 to channel 16 are output, and high frequency data of channel 17 to channel 24 are output.

FIG. 7 is a diagram showing data divided into 4 bits by the dividing unit 28. FIG. 7 is a diagram showing the bit configuration of the low frequency band data of channels 1 to 3 shown in FIG. The data of each channel is as shown in FIG.
Both the low-frequency data and the high-frequency data are composed of 16 bits, and are divided into 4-bit units by the dividing unit 28, so that the low-frequency data or high-frequency data of one channel is output to the audio addition highway in 4 clocks. .

Next, the configuration and operation of the voice detection unit 11 and the noise control unit 13 will be described with reference to FIG. In FIG. 8, 700 is a first detection unit, 800 is a second detection unit, 130 is a first comparison unit, 140 is a first determination unit,
Reference numeral 50 denotes a second determination unit, which constitutes a first detection unit 700. Furthermore, 160 is a second comparison unit, 1
Reference numeral 70 denotes a third determination unit and 180 denotes a fourth determination unit, which form a second detection unit 800. Reference numeral 910 is a noise control voice detection threshold value, and 920 is an AGC voice detection threshold value. A noise removal unit 930 removes noise when the input voice data is determined to be noise.

Next, the operation will be described. First, the input signal to the voice detector is input to the first detection unit 700 and the second detection unit 800, respectively. Of these, the first detection unit 700 includes a first comparison unit 130, a first determination unit 140,
The second determination unit 150 is provided and outputs the determination result with the noise control voice detection threshold value 910. Further, the second detection unit 800 includes a second comparison unit 160, a third determination unit 170,
The fourth determination unit 180 is provided, and outputs the determination result of voiced or silent.

Next, the operation of the first detector 700 will be described in more detail. The input signal is first subjected to noise control voice detection threshold value 91 in first comparison section 130.
Compared to 0. The first determination unit 140 determines whether there is sound or noise in block units for a certain fixed time based on the comparison result of the first comparison unit 130. For example, when 16-bit data as shown in FIG. 5 is set as one sample, eight samples are set as one block, and at least one of these eight samples is used as a noise control voice detection threshold 910.
If there is something that exceeds the threshold, this block is determined to be sound. On the contrary, if none of the 8 samples exceeds the noise control voice detection threshold value 910, this block is determined as noise. The second determination unit 150 determines noise only when the first determination unit 140 continuously determines noise for a predetermined number of blocks (for example, 10 blocks), and determines that there is sound otherwise. Then, the result of this determination is
It is output as the determination result of the detection unit 700 of.

The noise removing unit 930 inputs the determination result from the first detecting unit 700, and when the result indicates that it is noise, a gain of 1 or less is applied to the input voice data. The multiplication reduces the noise level. Sections B and G shown in FIG. 2 show the case where the first detection unit 700 determines that the voice data in this section is noise in this way, and noise is added by adding a gain of 1 or less. Shows the case of decreasing.

Next, the operation of the second detector 800 will be described. The input signal is first compared with the AGC voice detection threshold value 920 in the second comparison section 160. The AGC voice detection threshold 920 is a predetermined value, which is higher than the applicable range of the noise control voice detection threshold 910, and is usually higher than the maximum expected background noise level. It is set. Third determination unit 1
Reference numeral 70 determines whether or not there is sound in a block unit for a certain fixed time based on the comparison result in the second comparison unit 160. For example, as described above, 16-bit data is taken as one sample, one block is made up of eight samples, and if even one of these eight samples exceeds the AGC voice detection threshold value 920, It is determined to be voiced, and otherwise it is determined to be silent. Fourth determination unit 1
The reference numeral 80 determines that the sound is present only when the third determination unit 170 determines that the sound is continuous for a certain number of blocks (for example, 10 blocks), and otherwise determines that the sound is not present. It is output as the determination result of the second detection unit 800.

Sections A to G shown in FIG. 2 respectively indicate sections in the case where it is continuously determined to be voiced in this way, or the case where it is continuously determined to be noise. The above-described first detection unit 700 and second detection unit 800
Is performed using only the low frequency data shown in FIG. That is, the comparison determination is performed without referring to the high frequency data. The use of low-frequency data in this way is because the audio level is largely influenced by the low-frequency data, whereas the audio level is not influenced by the high-frequency data. That is, comparison determination can be easily performed by not referring to the high frequency data. As described above, the first detection unit 700 and the second detection unit 800 use low frequency data, whereas the noise removal unit 930 uses both low frequency data and high frequency data in order to remove noise. This is because it is necessary to reduce the gain of both the low band data and the high band data with respect to the section determined to be noise using only the low band data.

Example 2. Next, another embodiment of the present invention will be described with reference to FIG. 9 is different from FIG. 8 in that the noise detection voice detection threshold value 910 is adaptively determined. In FIG. 9, 110 is a noise level calculation unit that calculates the background noise level based on the input voice data and the output of the second detection unit 800, and 120 is the noise detection voice detection threshold value 91 based on the background noise level.
This is a threshold setting unit that sets 0. Further, reference numeral 940 is a hangover addition unit that performs a hangover process for a voiced time.

In this embodiment, the noise level calculation unit 1 calculates the background noise level of the input voice data when there is no sound.
10 is calculated according to the output from the second detection unit 800, and the threshold value setting unit 120 sets the noise detection voice detection threshold value 910 according to the calculation result. Even when the level changes, the noise control voice detection threshold 910 can be adaptively changed. The hangover addition unit 940 uses the second
After the output of the detection unit 800 changes from voiced to silence,
By performing a hangover addition process for holding the voiced judgment during a certain period of time and using this as an output signal 300 for voice detection, disconnection of the tail of the voice or the like is prevented.

Further, the noise level calculation unit 110 calculates the addition average value of the samples of the audio signal in a certain block interval and then calculates the addition average value thereof, and the comparison result in the second comparison unit 160. The background noise level is calculated based on and. That is, when the voice signal never exceeds the AGC voice detection threshold value 920 in the block section, the background noise level is calculated using the calculated average value. However, when the input signal exceeds the AGC voice detection threshold value 920 at least once in the block section, the calculation of the background noise level is stopped and the previously calculated background noise level value is maintained. Then, subsequently, the threshold setting unit 120 is used by the first comparing unit 130 based on the output result of the noise level calculating unit 110. In this way, the noise control voice detection threshold value 910 is adaptively determined.

In the example shown in FIG. 9, the operation of the voice level calculating section 110 is controlled based on the output of the second comparing section 160, but the third judging section 170 or the fourth judging section 180 is controlled. Noise level calculation unit 110 based on the output of
The operation of may be controlled. In addition, in this embodiment, the noise level calculation unit 110 calculates the average value of the samples of the audio signals in the block, and the background noise level is calculated from this, but the root mean square value in the block is used instead of the average value. Alternatively, the maximum peak value or the like may be used.

Example 3. Next, the noise reduction operation in the noise control unit 13 will be further described. Noise control unit 1
3 reduces noise by adding a gain of 1 or less, but does not make the noise level 0. Rather, it is more natural to leave noise to some extent than to make it to human hearing. If the noise is set to 0, it feels strange and becomes unnatural. Therefore, when reducing noise, it is necessary to consider how to reduce the noise. Here, a case where noise is reduced according to the number of terminals (the number of channels) will be described below.

When the noise control unit 13 reduces the noise level lower than the noise control voice detection threshold value,
By changing the level to be reduced according to the number of terminals to which voice is added, it is possible to leave some noise in the voice-added data. In this way, the noise-controlled data will not be completely silent, so there will be no sudden change in the level between the noise-controlled data and the non-noise-controlled data. Can be reduced.

FIG. 10 is a block diagram showing a voice processing unit in this embodiment. Reference numeral 13 denotes a voice control threshold value for noise control, and the voice data of a level below the threshold value is added to the number of terminals (voice addition). It is a 1 / n noise control unit that lowers the level (1 / n) divided by n), and is otherwise the same as the first embodiment. FIG. 11 shows output voice data in which gain control is performed with respect to changes in input voice data in the circuit of the third embodiment. The voice output target value, the AGC control voice detection threshold value, and the noise control voice detection threshold value are the same as those in the first embodiment.

Next, the operation will be described. The 1 / n noise control unit 13 has a noise detection voice detection threshold value, determines whether the voice data received from the voice detection unit 11 is greater than or equal to the threshold value, and outputs the data below the threshold value as voice. It is lowered to the level (1 / n) divided by the number of terminals (n) to be added. Information on the number of terminals to be added is received from the system control unit. In this way, the noise in the added voice data is (1 / n) × n = 1, which is the level of one terminal, so that the noise level is the same as the level before the addition, and The silent state disappears. As a result, there is no sudden change in the level between the data for which noise control has been performed and the data for which noise control has not been performed, so that it is possible to reduce the discomfort of voice during a conference.

Example 4. In the above embodiment, the case has been described in which the voice data determined to be noise is reduced to the level divided by the number of terminals (n) to which voice is added, but it is not necessarily limited to the case of dividing by the number of terminals. After adding the voices, the voice adding unit outputs the remaining voices after subtracting the voices of its own channel. For example, three terminals A,
When voice data is being input from B and C, terminal A
, The audio data of terminals B and C are added and output. Further, the voice data of the terminals A and C are added to the terminal B and output. Further, the audio data of the terminals A and B are added to the terminal C and output. Therefore, when three terminals are engaged in a video conference, each terminal has another
The voice data of the terminal will be output. In such a case, the voice data determined as noise may be reduced to ½ instead of ⅓. That is, the number of terminals participating in the video conference may differ from the number of terminals that actually add voice. In addition, a certain range is set for the number of terminals for voice addition. For example, 1 to 3 is 1/3, 4 to 6 is 1/5, and 7 to 10 is 10 minutes. Prepare a table like No. 1
Noise may be reduced according to the table.

Example 5. In the above-described first embodiment, the case where the voice level is made to approach the voice output target value in the voice adding unit 12 has been described, but the voice detection unit 11 or the noise control unit 13 sets the voice data level from each channel to the voice output target. You may make it match with a value. FIG. 12 shows a case where each voice detection unit 11 or noise control unit 13 controls the gain of voice output. The voice data from each channel is output to the voice adder 12 after the level is adjusted to the voice output target value in the voice detector 11 or the noise controller 13. In the voice adding unit 12, after adding these voices, the level is adjusted again to the voice output target value and the voice output target value is output.

Example 6. In the above embodiment, the case where the first detecting unit 700 and the second detecting unit 800 compare the threshold value and the level of the audio data by using the same configuration as shown in FIG. 8 has been described. The first detector 700 and the second detector 800 do not have to have the same configuration. Further, even if the configurations are the same, the number of samples forming one block may be different, or the number of blocks may be different.

Example 7. In the first embodiment described above, the case where the voice detection unit 11 inputs the voice data and the noise control unit 13 outputs the voice data to the voice addition unit has been described. On the contrary, the noise control unit 13 inputs the voice data. The voice detection unit 11 may output the voice data to the voice addition unit. That is, it does not matter which of the operation of detecting the voice by the voice detection unit and the operation of controlling the noise by the noise control unit is performed first.

[0055]

As described above, according to the present invention, by providing the noise control section, the noise level does not increase even if the voices are added and the gain is controlled so as to obtain an appropriate voice level. A multipoint video conference using good voice can be realized.

[Brief description of drawings]

FIG. 1 is a block diagram of a voice processing unit according to the present invention.

FIG. 2 is a diagram showing a change in output voice data accompanying a change in input voice data of a voice processing unit according to the present invention.

FIG. 3 is a more detailed block diagram of a voice processing unit in the present invention.

FIG. 4 is a timing chart of input data of the voice processing unit according to the present invention.

FIG. 5 is a timing chart of data from the encoding unit of the audio processing unit to the audio detection unit according to the present invention.

FIG. 6 is a transmission / reception timing chart of voice addition highway data of the voice processing unit according to the present invention.

FIG. 7 is a transmission / reception timing diagram of voice addition highway data of the voice processing unit according to the present invention.

FIG. 8 is a block diagram of a voice detection unit and a noise control unit of the voice processing unit according to the present invention.

FIG. 9 is a block diagram of a voice detection unit and a noise control unit of the voice processing unit according to the present invention.

FIG. 10 is a block diagram of a voice processing unit in the present invention.

FIG. 11 is a diagram showing a change in output voice data according to a change in input voice data of the voice processing unit according to the present invention.

FIG. 12 is a diagram showing an example of a gain control operation in the present invention.

FIG. 13 is a block diagram of a conventional multipoint control controller.

FIG. 14 is a block diagram of a conventional audio processing unit.

FIG. 15 is a diagram showing a change in output voice data associated with a change in input voice data in a conventional voice processing unit.

[Explanation of symbols]

1 Multipoint communication controller 2 System control unit 3 line interface section 4a-4m separation part 5a-5m Multiplexing part 6 Audio processing unit 7 Video processing unit 8 General-purpose data processing unit 9a-9m speech decoding unit 10a to 10m speech coding unit 11 Voice detector 12 Voice adder 13 Noise control unit 110 noise level calculator 120 threshold setting unit 130 First comparison unit 140 First determination unit 150 Second determination unit 160 Second comparison unit 170 Third determination unit 180 Fourth determination unit 300 voice detection signal 700 first detector 800 Second detection unit 910 Noise detection threshold for noise control 920 AGC voice detection threshold 930 Noise removal unit 940 Hangover addition part

─────────────────────────────────────────────────── ─── Continuation of the front page (56) Reference JP-A-4-851 (JP, A) JP-A-3-157044 (JP, A) JP-A2-148099 (JP, A) JP-A-4- 151953 (JP, A) JP-A-57-171400 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) H04M 3/56 H04N 7/15 H04M 11/00

Claims (7)

(57) [Claims] 1. A multipoint communication control apparatus for realizing a multipoint videoconference between videoconference terminals at a plurality of points, and a separation unit for separating at least voice data from multiplexed data received from the videoconference terminal; Storing a noise control voice detection threshold value for determining whether the voice data separated by the unit is noise or not, and the noise control voice detection threshold value and the level of the voice data separated by the separation unit. And a noise control unit that reduces the level of voice data below the noise control voice detection threshold and outputs the voice data, and outputs voice data that exceeds the noise control voice detection threshold as is. Automatic gain control (Autom) for determining whether the audio data separated by the separation unit is voiced or silent.
(Atic Gain Control, AGC) voice detection threshold value for AGC, which is a value higher than the noise control voice detection threshold value, is stored and stored as the AGC voice detection threshold value. By comparing with the level of the audio data separated by the separation unit , the separation unit
Whether the voice data separated by is voiced or silent
The AGC voice detection threshold value is set at the time of the determination.
The value must be higher than the noise detection threshold for noise control.
And the sound below the noise detection threshold for noise control
The data is judged to be silent, and the judgment result indicates whether or not there is sound.
A voice detection unit that generates voice detection information indicating whether it is sound, and voice data output by the noise control unit is received from another video conference terminal and added to the voice data output by the noise control unit. , A voice addition unit that controls the gain so that the added voice data has an appropriate voice level based on the voice detection information generated by the voice detection unit, and the voice data added by the voice addition unit is converted into another data. And a multiplexing unit that multiplexes and transmits to a video conference terminal. 2. The multipoint communication control apparatus according to claim 1, wherein the noise control unit reduces the noise level according to the number of terminals to which voice is added. 3. The multipoint communication control apparatus according to claim 2, wherein the noise control unit reduces the noise level to a level (1 / n) divided by the number (n) of terminals to be voice-added. . 4. The noise control unit, the noise control voice
The multipoint communication control device according to claim 1, wherein the detection threshold is adaptively changed. 5. The multipoint communication control device according to claim 1, wherein the noise control unit processes the audio data from the plurality of video conference terminals in a time-divisional order and outputs the processed audio data to the audio addition unit. . 6. The noise control unit determines the comparison result for each block, with a comparison unit for comparing the level of the voice data and the noise detection voice detection threshold value and a predetermined number of samples as one block. The multipoint according to claim 1, further comprising: a first determination unit and a second determination unit that determines whether or not the voice data is noise over a plurality of blocks based on the determination result of the first determination unit. Inter-communication control device. 7. The multipoint-to-point communication control device includes a decoding unit that decodes the voice data separated by the separation unit into low-frequency data and high-frequency data, and the noise control unit converts the low-frequency data. The multipoint communication control device according to claim 1, wherein it is determined whether it is noise or not using noise.