JPS62133853A - Conference telephone system - Google Patents

Abstract

PURPOSE:To prevent howling by controlling maximum number of speakers that can speak simultaneously out of subscribers who are taking part in a conference to a preset number. CONSTITUTION:Sound signals made by converting analog sound signals from a subscriber to digital signals are inputted from sound input signal terminals 1-1-1-n, and the sound signals are delayed by delay circuits 2-1-2-n to prevent cutting off of starting and ending of words. Further, sound signals are inputted to voice detecting circuits 3-1-3-n. Then, output signals 4-1-4-n of circuits 3-1-3-n are inputted to a gate circuit 8 and an addition selecting circuit 5. The circuit 5 selects (p) of aural signals of any subscriber, and only selected sound signals are passed at unchanged sound signal level and signal level of other subscribers is checked. Signals for which number of speakers is controlled to a maximum of (p) are added in an adder circuit 10 and outputted. Thus, a conference can be held preventing howling.

Description

DETAILED DESCRIPTION OF THE INVENTION (Technical field to which the invention pertains) The present invention relates to an addition control method for conference calls.

(Conventional technology) The conference call was made to eight subscribers in remote locations.

An audio sum signal obtained by adding and synthesizing the audio signals of other channels excluding the audio signal of the own channel.

This is achieved by distributing the channels to the receiving sections of each subscriber via a transmission line.・For a concrete explanation, an example will be explained in which a group call is made using the current telephone network.

For example, the Institute of Electronics and Communication Engineers communication methods research group material (C585
-67) ``Required insertion loss for voice addition in long point-to-point communication'' (August 1985), if long point-to-point communication connecting a large number of subscribers is to be realized in a telephone network, it will be necessary to Because of the addition, the loop gain within the 4-wire section becomes 1 or more due to the echo caused by the 2-wire/4-wire conversion within the network, resulting in a 4-wire transmission state or howling.

For this reason, it is pointed out that a loss must be inserted for each subscriber to cancel the echo.

Figure 6 shows the specific data contained in the same document.

FIG. 6 is a diagram showing the required insertion loss with respect to the number of ground planes in the conventional system.

From this, it can be seen that as the number of ground planes increases, it is necessary to increase the required insertion loss.

In a general telephone network, the echo return loss at a 2-wire 4-wire converter is the communication power input to the 2-wire 4-wire converter (hybrid) when looking at the 2-wire 4-wire converter (hybrid) from the 4-wire transmission line. Pi is the signal power P flowing to the 2nd line side towards the subscriber
s and P that returns to the 4-wire transmission line via this 2-wire 4-wire converter.
o and is defined as po/Pi. This echo return loss has an average value of 11 dB at most and 6 dB at worst, as shown in Facility Vol. 35-7 (July 1988 issue) "Transmission Loss in Digital Communication Networks".
It is reported that.

Therefore, when connecting multiple destinations, from Figure 6, the limit is 3 to 5 destinations, and if multi-terminal communication is to be performed beyond this, the value of echo return loss is a stochastic event, so at the time of connection It is expected that the number of ground stations that can be connected will vary considerably depending on the state.

In a conference telephone service that realizes multi-destination communication, it is impossible to rely on the value of this echo return loss, and a technique for inserting insertion loss for each subscriber is inevitably required.

In order to solve this problem, conventional techniques have been to use echo suppressors and echo cancellers to increase the insertion loss of each subscriber in an isometric manner.

For example, IEICE Communication Method Study Group Material C383-8
Examples are described in ``Design Conditions for Relay Devices in Audio Conference Communication Systems'' and ``Method for Designing Relay Devices for Conference Telephones'', document C384-3r of the same study group.

However, the echo suppressor may cut off at the beginning or end of a conversation depending on the volume, or may cut off in the middle of a conversation due to the level difference between the two when talking simultaneously. To eliminate this drawback, echo cancellers have been used.

However, since this echo canceller is controlled by adaptive operation at the beginning of connecting each subscriber, it requires operating time to obtain the required amount of echo cancellation.

Therefore, it is not possible to obtain the required amount of echo cancellation at the start of a connection, and as the number of stations increases, a howling phenomenon still occurs.

For this reason, there is a method to obtain the amount of echo cancellation by the echo canceller by sending out white noise at the same time as the connection is started, but since white noise is not a sound that is pleasant to human senses, we try to send out this sound as much as possible. It is preferable to avoid.

Furthermore, if the number of stations for conference telephone service is to be increased in the future, according to the above-mentioned document "Required insertion loss for voice addition in long point-to-point communications," when the number of stations is n, the required insertion loss (dB) will be 201og. Since it is proportional to (n-1), for example, if communication with more than 100 points is to be realized, it will be about 40 dB, which is a value that cannot be achieved with current echo cancellers.

Therefore, the realization of multi-destination communication using an echo canceller has the above-mentioned drawbacks when starting a connection or in response to requests for increasing the number of destinations.

Furthermore, it is clear that a multi-destination communication system in which a subscriber is connected using a four-wire system also has similar drawbacks due to echoes from the received signal on the terminal side being mixed into the transmitted signal.

(Object of the Invention) In order to eliminate these drawbacks, the present invention provides a conference call addition control method that can ensure call quality comparable to that of a normal telephone call.

(Structure and operation of the invention) Before a detailed explanation of the present invention, the basic idea of the present invention will be explained.

As mentioned above, in order to prevent howling as the number of ground planes increases, the amount of insertion loss must also increase.

On the other hand, in general, there are many cases in which one person speaks in a meeting, and the second most common case is a case in which two people speak. Furthermore, if each person speaks for more than 3 Å, the content of the meeting becomes the same as the voices of the crowd,
There's no point in having a meeting anymore.

When operating a multi-site audio communication conference, a moderator or organizer is always required, and in fact, in general meetings with a large number of people, the moderator is the central person in proceeding with the agenda.

Even if there is a moderator, if a sound of 3 Å or more is generated, it is the sound of laughter, greetings, or exclamations, and it is necessary to convey this information as well as the atmosphere of the meeting. However, even if a large number of voices are conveyed, in the case of a voice communication conference, it is difficult to know who is the person, and the number information is still meaningless.

Therefore, at the beginning of the meeting, the moderator or organizer who runs the meeting always explains the main purpose and agenda of the meeting, so placing a limit on the number of people who can add audio will not have any effect. In fact, it is more important in terms of call quality to prevent ringing.

The present invention was devised based on this idea.

FIG. 1 is an overall block diagram of an embodiment showing the configuration of the present invention, in which 1-1 to 1-n are audio input signal terminals, 2-1 to 1-n are audio input signal terminals;
2-n is a delay circuit, 3-1 to 3-n are audio detection circuits, 4
-1 to 4-n are output signals of voice detection circuits 3-1 to 3-n, 5 is an addition selection control circuit, 6-1 to 6-n are output signals of addition selection control circuit 5, 7-1 to 7 -n is delay circuit 2-1
~2-n output signal, 8 is gate circuit, 9-1 ~ 9-n
is the output signal of the gate circuit 8, 10 is the adder circuit + 11-
1 to 11-n are audio output signal terminals.

An audio signal obtained by converting an analog audio signal from a subscriber into a digital signal is input from audio input signal terminals 1-1 to 1-11, and is used for audio detection and addition selection control in order to prevent cutting at the beginning of one line and cutting at the end of one line. The audio signal is delayed by the required time by delay circuits 2-1 to 2-n. The audio signals from the audio human input signal terminals 1-1 to 1-n are processed by audio detection circuits 3-1 to 3 for detecting the presence or absence of the audio signals.
- input to n.

Next, the output signals 4-1 to 3-n of the voice detection circuits 3-1 to 3-n are
4-n is the on-off control of the audio signals 7-1 to 7-n output from the delay circuit 2, and the contents of the call presence/absence status of all subscribers among the subscribers participating in the conference. The addition selection control circuit 5 selects a specific audio signal from among a plurality of subscriber audio signals.

The output signals 9-1 to 9-n of the goo 1-circuit 8 are input to the audio adder circuit 10 and output to the audio output signal terminal 1l-1-11-n.
is output to.

FIG. 2 is a diagram showing the configuration of one embodiment of the addition selection control circuit, and shows an example in which the priority is given in descending order of subscriber numbers, 12 is a selector, 13 is a setting circuit, and 14 is a downlink. 15 is a NOR circuit, 16 is a D flip-flop circuit, 17 is an AND circuit, and 18 is a divider.

Generally, when convening a meeting, subscribers are connected in the order of reservation registration, and naturally the moderator or organizer is connected first, and then the person in charge is registered. Naturally, except for the moderator and organizer, the people with the highest ranking rank in descending order of rank. Naturally, the speaking rights are also in this order, so it is thought that the number of times the younger players speak will increase.

To operate this, the output signals 4-1 to 4-n of the voice detection circuits 3-1 to 3-n are first input to the selector 12.

This selector 12 sweeps the output signals 4-1 to 4-n of the audio detection circuits of the respective channels constituting the conference within the sampling time of the digitized audio signals.

On the other hand, the output of the setting circuit 13 (Efl, in fact, ROM or a group of simple switches can also be used) for setting the maximum number of speakers p set in advance is used as the initial value of the down counter 14, and the output signal of the selector 12 mentioned above is used as the initial value of the down counter 14. Audio detection output signal of each channel 4. -1 to 4-n sweep signals are used as a clock, and if the logic of the binary parallel output of the down counter 14 is all "0", the number of speakers will be the same as the set value p at that sweep channel. from.

At this point, it is sufficient to assemble a circuit configuration that closes the gate circuit 8 until the sweep period ends.

Furthermore, since it is reset after sweeping the conference constituent channels, the down counter 14 starts operating from the initial value.

Therefore, this circuit configuration can be realized by performing the operations described below.

The binary parallel logic output of down counter 14 is all 1
The logic output signal II from the NOR circuit 15 becomes 1011.
It is determined that the set value p has been reached from I11, the D flip-flop circuit 16 is turned on, and the logic 11 is determined from the output Q.
1" to logic 11011. An AND circuit 17 is used to close the channels from which audio has been detected, and a divider 18, which is synchronized with the selector 12, is used to invert the channel from 1" to logic 11011. Logic ``0'' is distributed to subsequent channels.

FIG. 3 is a diagram showing the configuration of the gate circuit, and 19 shows a group of three-manufactured AND circuits.

The audio signals from each subscriber are the output signals 7-1 to 7-1 of the delay circuit.
7-n, the output signals 4-1 to 4-n of the audio detection circuit, and the output signals 6-1 to 6-rl of the addition selection control circuit 5 constitute a three-man power AND circuit group 19, respectively. Only the subscriber channels detected as signals are summed by the output signals 4-1 to 4-n, and the output signal 6
-1 to 6-n, audio signals up to the set value p are added.

FIG. 4 is a diagram showing an example of a general audio addition circuit, with 20
is an adder circuit, 21-1 to 21-n are delay circuits, 22-1
~22-n is a subtraction circuit.

The configuration of this circuit is the same as that of the Japanese Patent Publication No. 57-29896R Digital Conference Telephone System.

This circuit operation is performed by adding all the audio signals of each subscriber channel in an adder circuit 20, delaying each subscriber's audio signal from the added signal by one sample time in a delay circuit 21, and then By subtracting signals 22-1 to 22-n, the signals of the sums of voices other than voice A and voice 1 are obtained.

This circuit requires only a small amount of hardware, especially when human signals and output signals are digitally multiplexed signals.

FIG. 5 is a diagram showing another embodiment of the addition selection control circuit, and the addition selection control circuit stores the numbers detected as audio signals from the start of the conference and selects p pieces in order of the stored number. This is an example of 5.

Here, 23 is an adder circuit, 24 is a temporary memory, 25 is a sweep address clock oscillation circuit, 26 is a selector, 27 is a gate circuit, 28 is a RAM, 29 is a comparator, :3
0 is RAM, 31 is AND circuit, 32.33 is RAM,
34 is an 11-sentence circuit, 35 is a selector, and 36 is a 1-sentence circuit.

This is the output signal 4 of the detection circuit 3-1 to 3-n.
In order to count the cumulative number of voice detections for each of -1 to 4-n, it is composed of an adding circuit 23 and a temporary memory 24.

Next, the contents of the temporary memory 24 are repeatedly swept 2Xp times within one sample period via the selector 26 by the clock of the sweep address clock oscillation circuit 25 which sweeps the address (corresponding to the same number as each subscriber).

In the first sweep of the selector 26, the gate circuit 27 is ON, and the RAM 2 outputted via the selector 35
The content of 8 is O.

Therefore, the comparator 29 compares the number of times the voice has been detected with the content B of the internal RAM 24 and the content A of the RAM 30 via the selector 26 and the gate circuit 27.

If A<B, the RAM 30 via the AND circuit 31
The contents of RAM24 are taken in, and after one sweep, the contents of RAM24 are taken in.
The maximum number of voices detected during the meeting is taken into the contents of 30, and at the same time, the contents of the address having the maximum number of voices detected are input into the RAM 32 by the output of the comparator 29.

Then, at the start of the second sweep, the contents of the RAM 32 are shifted to the RAM 33, and in the second sweep cycle, the contents of the RAM 33 that coincide with the sweep clock are set to logic "1" by the matching circuit 34. Then, the logic II I ++ is written into the RAM 28 at this address number.

next, ! (The contents of AM28 are transferred to the gate circuit 2 by the selector 35 synchronized with the address clock of the selector 26.
7, the gate circuit 27 inputs it to RAM2.
Since the contents of 8 are in the form where the address number is closed at the logic IZ I II, RAM 30 and AND circuit 31 . The comparator 29 selects the address with the second largest number of voice detections in the third sweep, and repeats p times every odd number of times. l I 11 is written, and in one sample period, contents having nine logic 111 ++ are stored in the RAM 28 in association with each address number.

Therefore, when transferring this content to the next sample period, R
If you shift to AM36, this content will be used to control the gate circuit 8, and the speaker whose voice has been detected with a large number will be selected from p.
Pieces can be selected sequentially.

Here, the preset value of p varies depending on the value of echo return loss (terminal echo return loss) within the network, but
In the case of a general telephone network, a value of 3 to 5 is generally appropriate based on the values mentioned above.

In order to make the logic easier to understand, we have explained the method using parallel processing, but in the actual circuit, since the audio signal is digitally multiplexed, the above-mentioned parallel processing can be done by time-division arithmetic processing. It is obvious that the hardware would be very simple.

(Effects of the Invention) As explained above, the present invention controls the maximum number of speakers who can talk at the same time among the subscribers participating in the conference to a preset number p. Even if the amount of echo cancellation cannot be obtained during initial operation, it is possible to compensate for the lack of echo cancellation due to the increase in the number of ground stations.
This has the advantage of preventing howling.

[Brief explanation of drawings]

FIG. 1 is an overall block diagram of an embodiment showing the configuration of the present invention, FIG. 2 is a diagram showing the configuration of an addition selection control circuit, FIG. 3 is a diagram showing the configuration of a gate circuit, and FIG. The figure shows an example of a general audio addition circuit. FIG. 5 is a diagram showing another embodiment of the addition selection control circuit. FIG. 6 is a diagram showing the required insertion loss with respect to the number of ground planes in the conventional system. 1-1~1-n...Audio input signal terminal, 2-1~2-n
... Delay circuit, 3-1 to 3-n... Voice detection circuit, 4-1 to 4-n... Output signal of voice detection circuit. 5...Addition selection control circuit, 6-1 to 6-n...Output signal of addition selection control circuit. 7-1 to 7-n...Delay circuit output signals. 8...Gate circuit, 9-1 to 9-n...Output signal of gate circuit, 10...
-Addition circuit, 11-1 to 11-n...Audio output signal terminal, 12.2
6.35...Selector, 13...Setting circuit. 14...down counter, 15...NOR circuit, 16
...D flip-flop circuit, 17...AND circuit, 18...divider-119.
... 3-man power AND circuit group, 20...addition circuit. 21-1 to 21-n...delay circuit, 22-1 to 22-n...subtraction circuit, 23...addition circuit, 24...temporary memory, 25...sweep address clock oscillation circuit, 27...
Gate circuit, 2g, 30, 32, 33.36...RA
M. 29... Comparator, 31... AND circuit,
34...-number circuit. Patent Applicant Nippon Telegraph and Telephone Corporation NEC Corporation Fujitsu Corporation Figure 1 Exit I-1 to I-n-t P 2\Kai5-tau4-
l ~4-n -94pR arc@niricp<546-1
~6-n ゛ Aroma z name (control π bow) 1 point
Ko2gu Fukuzo 17-1~7-n...Zun'ei U Yang 45 No. 9
-1 ~9-n... Date l! ]Kinari Muha Anger +
+-+ ~1l-jl-! fpgfia843 Fig. 2 15 ・NOR times + 6-D flip-flop I! ] 17 ° AND square 3 Figure 4 Figure 2o... 〃B 11 circuits 22-1 to 22-n゛・Yukii Saiguchi Takashi 5 Figure 23・Addeda Masaru 27・K-) 1! l well 29 comparator 31- AND invert)

Claims (3)

[Claims] (1) In a conference telephone system that allows simultaneous calls between three or more subscribers, a voice detection means detects the presence or absence of a voice signal from each subscriber, and the voice signal is detected by the voice detection means. Count the number of subscribers that have been created, and if the count does not reach a preset number p of at least one subscriber, only the voice signal of that subscriber is passed through as is at the voice signal level, and the rest are in a silent state. When the subscriber's signal level is blocked and the count exceeds a set value p, p audio signals of arbitrary subscribers are selected and only the selected audio signals are passed through at the same audio signal level. , addition selection control means for blocking signal levels of other subscribers and always controlling the number of callers to a maximum of p; Create a sum signal of voice signals other than subscribers,
A conference telephone system comprising: addition means for distributing to each subscriber. (2) When the number of audio signals detected as audio signals by the audio detection means exceeds the set number p, p or more signals detected as audio are selected in the order of registration of the subscribers constituting the conference, and The conference telephone system according to claim 1, further comprising addition selection control means for selecting an audio signal as an input to the addition means. (3) When the count detected as an audio signal by the audio signal detection means exceeds a set number p, storing the number detected as an audio signal by the audio detection means from the start of the conference;
Claim (1) characterized in that it has an addition selection control means for selecting p pieces in order of the stored number.
Conference telephone method described in section.