JPH0329555A - Storage system for agenda - Google Patents

Abstract

PURPOSE:To facilitate the generation of an agenda after a conference by storing the content of agenda during the conference while adding each caller name. CONSTITUTION:Voice information from terminal equipments 1-3 is sent to a selection section 30 of a conference system 10 and a storage switch 20. The selection section 30 sends selection information selecting a voice signal having the highest voice level to the storage switch 20. The voice is sent to a store control section 40, from which the voice is sent to a store section 50 and a talker recognition section 60, the store section 50 informs the storage location of an input voice to the store control section 40, which informs the store location to a caller name provision section 70. The caller name provision section 70 records the voice store location information to a caller name location store section 80 and also records the caller name obtained from the caller recognition section 60 to the caller name location store section 80 corresponding to the caller name obtained from the caller recognition section 60. Thus, from whom each utterance is made is identified.

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial Application Field) The present invention relates to a method for storing the contents of a conference system, and particularly relates to a method for storing the contents of a conference system that is connected to a plurality of terminals via a communication network. ．． <Conventional Technology> An example of a method for storing agenda contents in a conventional conference system is configured as shown in Figure 7. In Figure 7, 3
The audio signals from the two terminals 1, 2.3 are sent to the storage switch 20 of the conference system 10 and the selection unit 3 via the communication network 100.
Sent to 0. The selection unit 30 selects the audio signal with the highest level among these input audio signals, and transmits the selection information to the storage switch 20 as a control signal. The storage switch 20 selects only one sound for storage according to the control signal received from the selection section 30. The audio signals selected in this way are transmitted to the storage section 50 and stored therein. When reproducing audio during a conference, audio signals from the terminal and the storage unit 50 are sent to the terminal via the exchange, and the terminal obtains audio information. (Problems to be Solved by the Invention) As described above, in the conventional conference system content storage method, only the maximum level audio signal is stored in the storage unit 50, and therefore, when playing back the stored content, , 'ff?' regarding the name of the speaker of the playback audio. The problem is that there is no information and it is unclear who made the statement. (Means for Solving the Problems) In order to solve the above-mentioned problems, the conference system according to the present invention has a content storage method that connects to a plurality of terminals via a communication network and converts audio signals or the audio signals. In a conference content storage method of a conference system in which text obtained by a speaker recognition unit that holds a feature of a conference participant's audio signal as a reference feature, compares the input audio signal with the reference feature, identifies a speaker of the input audio signal, and outputs a speaker name; The storage section adds the name of the speaker of the audio signal output from the speaker recognition section and selected by the selection section to the content of the proceedings and stores it. Another method for storing the contents of a conference according to the present invention is the contents of a conference system that is connected to a plurality of terminals via a communication network and stores an audio signal or a text obtained by converting the audio signal in a storage section. In the storage method, a selection section that specifies one audio signal to be stored in the storage section among the audio signals from the plurality of terminals, a correspondence table that holds user names for the uncorresponding terminals, and a selection section that selects one of the audio signals from the plurality of terminals to be stored in the storage section; a conversion unit that converts a control signal indicating a transmission source terminal of the audio signal into a user name, and the storage unit adds the user name of the transmission source terminal obtained by the conversion unit to the content of the proceedings. and accumulate it. (Operation) In the present invention, according to the instructions of the selection unit in the conference system, only one voice generated during the conference is selected for storage at a time, and the speaker is recognized by speaker recognition. Information is added to the statement and recorded in the storage section. As a result, a terminal that refers to the content of WA events during the meeting after the end of the meeting can identify who made each comment in the content of the proceedings stored during the meeting from the storage unit. Further, in another method of storing minutes according to the present invention, only one audio generated during the conference is selected at a time according to instructions from a selection unit in the conference system, and the selection unit indicates the source terminal of the audio. The conversion unit that receives the control signal converts the voice transmission source terminal into a user name, adds this as the speaker name to the utterance content, and records it in the storage unit. This results in
A terminal that refers to the content of the meeting after the meeting ends can identify who made each comment in the content of the content stored during the meeting from the storage unit. (Example) Next, the present invention will be explained with reference to the drawings. FIGS. 1(a) and 1(b) are block diagrams showing one embodiment of the present invention and an example of a method for storing agenda contents in a multipoint conference.The operation of this embodiment will be explained below. First, to explain the basic components, Figures 1(a) and (b)
In this case, the storage control section 40 stores the audio signal from the storage switch 20 in the storage section 8o, and notifies the speaker name assigning section 7o of the storage position. The audio signal is also transmitted to the speaker recognition section 60. The speaker recognition unit 60 stores the voice characteristics of each conference participant, recognizes the speaker based on the voice signal characteristics, and outputs the speaker's name. The speaker name assigning unit 70 receives the audio signal from the storage controller 40.
The I product position information is received, the speaker name is received from the speaker recognition unit 60, and the accumulated position information of each utterance and the speaker name are stored in the speaker name position storage unit 80 in association with each other. The speaker name position storage unit 80 records the accumulated position information of each utterance and its speaker. The specific operation of this embodiment will be explained below. Terminals 1, 2.
The audio information from 3 is transmitted to the selection section 30 and the storage switch 20 of the conference system 10, as shown in FIG. 1(a). The selection unit 30 compares the levels of all voices,
Selection information for selecting the one person with the highest audio level is sent to the storage switch 20. Here, the storage switch 20
selects one audio for storage according to the control signal received from the selection section 30. This voice is transmitted to the storage control section 40, and from there to the storage section 50 and the speaker recognition section 60. The storage section 50 notifies the storage control section 40 of the storage position of the input voice, and the @product control section 40 notifies the speaker name adding section 70 of the storage position. The speaker name assigning unit 70 records the voice storage position information in the speaker name position storage unit 80 and at the same time assigns the speaker name obtained from the speaker recognition unit 60 to the speaker name in association with the storage position information. It is recorded in the position storage section 80. Next, the procedure for replaying the agenda after a meeting will be explained using Figure 1(b). The terminal 2 that requested the playback of the contents of the meeting is connected to the communication network 1.
It is connected to the playback processing unit 90 in the conference system 10 via 00. The reproduction processing unit 90 reads out the speaker names in the order stored in the speaker name position storage unit 80 and displays them on the display of the terminal, and at the same time reproduces the audio starting from the stored position information for the speaker from the storage unit 50. , send to terminal 2. Repeat this until the stored audio is finished. Second
The figure shows a memory structure that is stored in the speaker name location storage unit 80 and indicates the speaker name corresponding to the storage location of the voice. In the figure, storage position 1 of the storage section 50
The voice information stored in 000 indicates that the speaker is "Yoshida."Here, the storage location is the location information such as the track or sector of the disk if the storage device is a disk device, or the location information of the disk track or sector if the storage device is a disk device. It shall indicate address information, etc.

FIG. 3 is a block diagram showing a second embodiment of the present invention. In this embodiment, the voice from the voice processing section 40 is converted into text by the voice-to-text conversion section 51, and is stored in the text storage section 52 as text information. The procedures for speaker recognition, assigning a speaker name, and reproducing stored information are described in Section 1 above.
This is similar to the example in . FIG. 4 is a block diagram showing a third embodiment of the present invention. In this embodiment, audio from all terminals is temporarily stored in buffers that are not supported by each terminal. Audio transmission from each buffer to the storage control section 40 is under the control of the order control section 35. For example, when audio is generated from another terminal, such as terminal 2, while audio is being transmitted from terminal 1, the audio from terminal 2 generated later is accumulated in the audio buffer 22, and the sequence controller 35 stores the audio in the storage controller 40. It waits in the audio buffer 22 until it is instructed to send to. The sequence control unit 20 performs voice/silence detection and recognizes the end of the voice from the terminal 1.
Instructs to start transmitting the next voice waiting in the voice buffer 22. The voice from the voice buffer 22 is transmitted to the storage control section 40, the speaker is recognized in the same manner as in the first embodiment, and the speaker name assigning section 70 stores each utterance in the speaker name position storage section 80. The stored location information and the speaker's name of the utterance are memorized. In this practical example as well, a method of converting the output from the storage control unit 40 into text and storing it as in the second embodiment can also be considered. FIG. 5 is a block diagram showing an embodiment of the method for accumulating minutes in a multipoint conference according to the second invention. In the present invention, unlike the first invention, the speaker's name for each utterance is not recognized by speaker recognition, but the converter 65 recognizes the speaker's name! ，! 'g, ),,. Conversion section 6
5 has a correspondence table in which the user name of each terminal is stored in advance. When a control signal indicating the selected terminal is obtained from the selection section 30, the conversion section 65 converts the terminal into a user name, and transmits this as a speaker name to the speaker name assignment section 70.
FIG. 6 shows a memory structure showing terminals and corresponding user names stored in the correspondence table of the conversion unit 65. The figure shows that the users of terminals 1, 2, and 3 are Tanaka, Yoshida, and Igeda, respectively. The operation of the embodiment will be explained below. The audio information from terminal 1.2.3 is
The data is sent to the selection unit 30 and storage switch 20 of the conference system 10. The selection unit 30 compares all the audio levels and transmits selection information for selecting the one with the highest audio level to the storage switch 20. Here, storage switch 2
0 selects one audio for storage according to the control signal received from the selection section 30. This audio is transmitted to the storage control section 40, and from there to the storage section 80. At the same time, the selection unit 30 transmits the selected terminal information to the conversion unit 65, and selects a different terminal. Identify the user. Furthermore, the voice processing unit 40 notifies the speaker name assigning unit 70 of the storage position of the input voice in the storage unit 50. The speaker name assigning unit 70 records the voice storage location information in the speaker name location storage unit 80, and at the same time assigns the user name obtained from the conversion unit 65 as the speaker name in association with the storage location information. It is stored in the position storage section 80.

The procedure for reproducing stored information after a meeting is the same as in the first embodiment of the first invention. Examples of the present invention include the second to third embodiments described in the first invention.
Possible methods include converting to text and storing the contents of the proceedings as text information, as in Example 3, and providing audio buffers for each end. (Effects of the Invention) As described above, according to the present invention, the contents of the proceedings during a meeting are stored with the names of each speaker added, making it easy to create minutes after the meeting. It has the effect of becoming

[Brief explanation of drawings]

Figure 1(a). (b) is a structural block diagram showing a first embodiment of the agenda content storage method of the first invention in the present invention;
FIG. 2 is a memory configuration diagram in the speaker name position storage section of the first invention, FIG. 3 is a configuration block diagram showing the embodiment of FIG. 2, and FIG. 4 is a configuration block diagram showing the third embodiment. , FIG. 5 is a configuration block diagram showing an embodiment of the agenda content storage method of the second invention in the present invention, FIG. 6 is a memory configuration diagram in the correspondence table of the conversion section of the invention of FIG. 2, and FIG. FIG. 1 is a block diagram illustrating a conventional agenda content accumulation method. 1,2.3...Terminal, 10...Conference system, 20
... Accumulation switch, 21, 22. 23... Audio buffer, 30... Selection section, 35... Sequence control section, 4
0...Speech processing unit, 50...Storage unit, 51...Speech-to-text conversion unit, 52...Text storage unit, 60
...Speaker recognition section, 65... Conversion section, 70... Speaker name assigning section, 80... Speaker name position storage section, 90... Playback processing section, 100... Communication network . (a) Figure 1 (b) Figure 1 Figure 3 Figure 2 Figure 4 J Figure 5 Figure 6

Claims (2)

[Claims] (1) In a conference content storage method for a conference system that is connected to a plurality of terminals via a communication network and stores audio signals or text obtained by converting the audio signals in a storage section, a selection unit that specifies one audio signal to be stored in the storage unit among the audio signals; and a selection unit that holds a predetermined feature of the audio signal of a conference participant as a reference feature, and compares the input audio signal with the reference feature. and a speaker recognition unit that identifies the speaker of the input audio signal and outputs a speaker name, and the storage unit includes a voice output from the speaker recognition unit and selected by the selection unit. A method for storing the contents of an agenda, which is characterized in that the name of the speaker of the signal is added to the contents of the agenda and stored. (2) In a conference content storage method for a conference system that is connected to a plurality of terminals via a communication network and stores audio signals or text obtained by converting the audio signals in a storage unit, a selection section for specifying one audio signal to be stored in the storage section among the audio signals of the above, a correspondence table for holding user names for the unsupported terminals, and a source terminal of the audio signal from the selection section. a conversion unit that converts a control signal into a user name, and the storage unit adds and stores the user name of the transmission source terminal obtained by the conversion unit to the content of the discussion. Content storage method.