JPH0785581B2 - Camera control device for speaker shooting of video conference system - Google Patents

Description

DETAILED DESCRIPTION [Table of Contents] Outline Industrial field of use Conventional technology Problems to be solved by the invention Means for solving the problems Action Example Effect of the invention [Overview] The present invention is a video conference. It relates to a control device that automatically detects the attendee of the current speaker and drives the camera for shooting the attendee so that the attendee can be filmed on TV. The purpose of this is to provide a device capable of enabling the input time of the microphone provided for each attendee over the past approximately 4 seconds, and during that time, the total input time of more than the time set to approximately 2 seconds The feature is that detection is made with the attendee of the confirmed microphone as the current speaker.

INDUSTRIAL APPLICABILITY The present invention automatically identifies an attendee who is the current speaker in a video conference and drives a camera for shooting the attendee so that the speaker is filmed on television. The present invention relates to a control device.

In the video conference, another camera is provided in addition to the camera for taking a panoramic view of the conference room.

The speaker who is currently speaking is selected from among the attendees as an object to be photographed by this camera, and the selection is made by a specific attendee (for example, the moderator).

However, doing this places a heavy burden on the attendees and is prone to error.

Therefore, this type of device is used in a video conference system.

[Prior Art] In this type of device, a microphone is provided for each attendee,
The microphone attendee whose input is confirmed is identified as the speaker who is currently speaking.

Then, one attendee photographing camera is driven so that the speaker is photographed on the television, and as a result, they are sequentially photographed every time the speaker changes.

Here, conventionally, when a microphone input of 2 seconds or more is confirmed as shown in FIG. 7 (A), the attendant of the microphone is identified as a speaker, and the microphone input of 2 seconds or less is noisy. Is treated as.

Further, when two or more attendees are speaking as shown in FIGS. 8B and 8C, it is considered that there is no microphone input for them.

[Problems to be Solved by the Invention] However, in the past, the attendee who made an intermittent microphone input as shown in FIG. There was a problem in that when the speech was interrupted for the purpose of watching the situation, the attendees were not specified as speakers even though they were actually speaking.

Further, when the noise shown in FIG. 6F occurs during the speech shown in FIG. 8E, they are simultaneously input in FIGS. 7B and 7C, so that the speech at this time is ignored. ,
There was also the problem that the attendees were not identified as speakers.

Further, conventionally, as shown in Japanese Patent Laid-Open No. 61-66487, the amplitude change of each audio signal is smoothed prior to the level comparison, and the level comparison is performed every constant period T1. For each audio signal channel, the number of times that the maximum level becomes a certain value or more is measured for each constant period T2, and the audio signal whose measurement result is the maximum and a certain value or more is detected. A video signal switching method is also generally disclosed, which detects a channel and selects and switches a video signal corresponding thereto.

However, in such a switching method, for example, when A and B start a conversation at the same time, one of them usually takes the leadership of the conversation, and even if the other party is merely depressed, the conversation is Since the speaker is selected according to the level of the sound pressure level at the start, there is a problem that the other party who simply strikes is selected.

The present invention has been made in view of the above conventional problems,
It is an object of the present invention to provide a high-performance device which has a high speaker identification accuracy and is not affected by noise.

[Means for Solving Problems] In order to solve the above-mentioned object, the device according to the present invention is
It is configured as shown.

In the figure, the microphone input judging means 10 judges whether or not the microphones prepared for attendees of the video conference are input.

Then, the judgment result holding means 12 holds the judgment result of each microphone input for a period from the present to about 4 seconds or less.

Further, the time totaling means 14 obtains the total time of each microphone input in the above period from the judgment result held by the means 12.

Further, the speaker identifying means 16 identifies the attendee of the microphone whose total time calculated by the means 14 exceeds the set time of about 2 seconds as the current speaker.

Then, the camera driving means 20 drives the camera so that the attendee identified as the current speaker by the means 16 falls within the imaging range.

[Operation] In the present invention, each microphone input is held for the past about 4 seconds, and the microphone attendee whose total microphone input for about 2 seconds or more is confirmed is identified as the speaker.

[Embodiment] A preferred embodiment of an apparatus according to the present invention will be described below with reference to the drawings.

FIG. 2 shows an example of a system to which the present invention is applied. The participants 21-1, 21-2, 21-3, 21-4 of the video conference are shown.
On the other hand, microphones 22-1, 22-2, 22-3, 22-4 are prepared respectively.

And these microphones 22-1, 22-2, 22-3, 22-4
The audio signal of is given to the microphone mixer 23,
The audio signal is given to the audio transmission device 24.

The voice signal from the microphone mixer 23 is the speaker detection circuit 25.
The speaker detection circuit 25 identifies the speaker by the input voice signal.

Further, the output signal of the speaker detection circuit 25 having this content is given to the swivel controller 26, and the swivel controller 26 controls the electric swivel 27.

As a result, the camera 28 is directed to the attendees 21-1, 21-2, 21-3 or 21-4 of the speaker identified by the speaker detection circuit 25,
A video signal in which a speaker fits within the screen is given from the camera 28 to the video transmission device 29.

A speaker detection circuit 25 is shown in FIG. 3, and the voice signals of the microphones 22-1, 22-2, 22-3, 22-4 are input to the sampling circuit 30 via the microphone mixer 23. .

In this sampling circuit 30, a digital signal that is turned on when the audio signal is above a predetermined level and turned off when it is below that level is obtained for each microphone 22-1, 22-2, 22-3, 22-4, The values of those digital signals are sampled and taken into the input buffer 31.

In the input buffer 31, each microphone 22-1, 22-2, 22-3,
Flags are set and reset according to the digital signal value for 22-4, and the contents of each flag in the input buffer 31 are prepared for the microphones 22-1, 22-2, 22-3, 22-4. It is distributed to the storage buffers 32-1, 32-2, 32-3, 32-4.

The number of bits of these accumulation buffers 32-1, 32-2, 32-3, 32-4 is the same as the number of samplings for 4 seconds, and the contents of the oldest bit are updated with the latest.

As a result, the bit when the voice signal exceeds a predetermined level is set to the oldest one, and each accumulation buffer 32-
Total time MIC1on, MIC2on, MIC3on, MIC4on that the voice input of the microphone 22-1, 22-2, 22-3, 22-4 was confirmed by the set bit number in 1,32-2, 32-3, 32-4. Is shown.

In the processing circuit 44, as shown in FIG. 4, first, they are sequentially read (step 40), and the preset detection time (2 seconds)
(Step 41).

In addition, the total time for checking the microphone input that is currently loaded is MIC.
(N) on exceeds the detection time (YE in step 41)
S) shows the corresponding one of the numbers assigned to the microphones 22-1, 22-2, 22-3, 22-4 and the total time MIC.
(N) on is stored (step 42).

When it is confirmed that the comparison of all the total time MIC (n) on and the detection time (step 41) is completed (YE in step 43).
S), it is determined whether the total number of microphone numbers stored when the total time MIC (n) on exceeds the detection time is 0, singular or plural (step 44).

When the number of microphone numbers stored at that time is 0, it is judged that all the attendees 21-1, 21-2, 21-3, 21-4 are not speaking and there is no speaker. (Step 45).

When the number of stored microphone numbers is singular, the microphones 22-1, 22-2, 22-
3 or 22-4 attendees 21-1, 21-2, 21-3 or 21-
4 is identified as the speaker (step 46), and a signal containing this microphone number is output to the swivel controller 26 (step 47).

As a result, the attendees 21-1, 21- of the microphone 22-1, 22-2, 22-3 or 22-4 indicated by the microphone number at this time
The camera 28 is swiveled by the electric swivel base 27 so that 2, 21-3 or 21-4 fits on the screen.

Further, when there are a plurality of microphone numbers whose total time MIC (n) on exceeds the detection time, the total time MIC (n) o
Attendees of microphones 22-1, 22-2, 22-3 or 22-4 whose longest n is searched (step 48) and whose total time MIC (n) on is indicated by the microphone number.
21-1, 21-2, 21-3 or 21-4 is specified as a speaker (step 49).

Therefore, the microphone number of the speaker who is forcibly prioritized in this way is output (step 47), and the camera 28 is directed to the speaker.

FIG. 5 explains the speaker specifying operation in this embodiment. When the microphone input shown in FIG. 5A is performed, the total time MIC (n) on from the present to 4 seconds before is shown in FIG. ) Changes.

Here, as shown in FIG. 7A, noise is first given as a microphone input, but that noise has not occurred for more than 2 seconds in total within the time frame of 4 seconds. 28 cannot be misdirected.

Further, as shown in FIG. 3A, when the pronunciation is started following the noise, when the speaking time becomes 2 seconds in the time frame of 4 seconds, the speaking source is recognized as the speaker, which is different from the conventional case. Immediate camera to the speaker without having to say for 2 seconds
28 directed.

Further, even if the speaker interrupts his / her utterance in order to obtain other consents or to check their appearance, the same person is recognized as the speaker because the speaker is specified in the total time.

Here, it was confirmed that the noise did not continue for 2 seconds or more, the speech was not interrupted for 2 seconds or more during one speech, and the noise did not occur continuously for 2 seconds or more. ing.

Therefore, false speaker identification due to noise can be completely eliminated.

Also, when the microphone input for a total of 2 seconds or more is confirmed within the time frame, the utterance is always made, so that the speaker who made the utterance can be surely specified.

Furthermore, since the speaker identification is performed by the total time of microphone input, even if the speech is interrupted for the first time,
The speaker can be identified quickly.

As described above, according to the present embodiment, it is possible to quickly perform accurate speaker recognition while completely eliminating the influence of noise.

FIG. 6 is a diagram for explaining the operation when there is another microphone input while a speaker is speaking, as shown in FIG. 6B while a speaker is speaking as shown in FIG. Then, noise is input to the microphone by another, and another speech is started following this before the speech ends. Here, since the speaker is specified by the total of microphone inputs and the total time of noise input does not exceed the detection time, as will be understood from FIGS. Even if noise is superimposed on a person's speech, erroneous speaker recognition as shown in FIG. 7 is not performed.

Further, even if another speech is made during a certain speech, the speaker having the longer speech duration is preferentially recognized as the speaker, so that switching the speaker to be photographed makes the viewer feel uncomfortable. It is done smoothly without giving.

Also, if there are multiple microphone inputs at a certain time,
Recognize the integration of the input time during the cycle of about 4 seconds,
By identifying the current speaker, it is possible to automatically select who has the leadership in the conversation, rather than simply selecting the louder one.

For example, when A and B start a conversation at the same time, one of them usually takes the leadership of the conversation, and the other party has a sound pressure level at the beginning of the conversation even when the other side is just about to hit each other. Without making a decision, you will be able to observe and select which one will take the leadership for about 4 seconds.

As described above, according to the present embodiment, even when the microphone inputs overlap, the camera 28 can be smoothly pointed to the speaker who is speaking without error.

[Effects of the Invention] As described above, according to the present invention, the attendee whose microphone input is confirmed for 2 seconds in total within the time frame from the present to about 4 seconds before is specified as the speaker. , It becomes possible to perform the identification extremely accurately, and it is also possible to completely eliminate the influence of noise.

[Brief description of drawings]

FIG. 1 is an explanatory view of the principle of the invention, FIG. 2 is an explanatory view of the overall configuration of the embodiment, FIG. 3 is an explanatory view of the configuration of a speaker detection circuit, and FIG. 4 is a flowchart explaining the operation of a processing circuit. FIG. 6 and FIG. 6 are explanatory views of the operation of the embodiment, and FIG. 7 is an explanatory view of the conventional example. 10: Microphone input judging means 12: Judgment result holding means 14: Time totaling means 16: Speaker identifying means 20: Camera driving means 21-1, 21-2, 21-3, 21-4. Attendees 22-1,22-2,22-3,22-4 …… Microphone 23 …… Microphone mixer 24 …… Speech transmission device 25 …… Speaker detection circuit 26 …… Swivel base control device 27 …… Electric swiveling Platform 28 …… Camera 29 …… Video transmission device 30 …… Sampling circuit 31 …… Input buffer 32-1,32−2,32−3,32-4 …… Storage buffer 33 …… Processing circuit

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Toshihiro Honma 1015 Kamiodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture, Fujitsu Limited (72) Inventor Hitoshi Sato, 1015, Kamedotachu, Nakahara-ku, Kawasaki City, Kanagawa Prefecture, Fujitsu Limited ( 72) Inventor Hitoshi Ishiguro, 1015 Kamiodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa, Fujitsu Limited (72) Inventor, Ashita Yoshishi, 1015, Kamedotachu, Nakahara-ku, Kawasaki, Kanagawa, Fujitsu, Ltd. (72) Inventor Masakazu Yamaguchi 1015 Kamiodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture, Fujitsu Limited (56) References JP-A-61-87489 (JP, A) JP-A-61-66487 (JP, A)

Claims (2)

[Claims] 1. A microphone input judging means (10) for judging presence / absence of input of a microphone prepared for attendees of a video conference, and a result of judging presence / absence of each microphone input from the present to about 4 seconds or less. A determination result holding means (12) for holding for a period of time, a time totaling means (14) for obtaining a total time of each microphone input in the period based on the determination result being held by the determination result holding means (12), and a time totaling means The speaker identification means (16) for identifying the microphone speaker who has exceeded the set time of about 2 seconds obtained in (14) as the current speaker, and the speaker identification means (16) A speaker photographing camera control device for a video conference system, comprising: a camera driving unit (20) for driving a camera in which an attendee identified as a person is within an imaging range. 2. A means for totalizing the number of microphones whose total time calculated by the time totaling means (14) is equal to or longer than the set time, and the longest total time when there are a plurality of calculated microphones. A means for forcibly selecting an attendee of a microphone as a current speaker, and a camera control device for photographing a speaker of a video conference system according to claim (1). .