JP2003177776A - Minutes recording system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a minutes recording system which identifies each of conference participants and automatically generates minutes in accordance with voice data in their speeches. <P>SOLUTION: The minutes recording system is constituted of computer terminal devices of which the number is equal to the number of conference participants, a data processing server, and a communication network system, and a network server, and each computer terminal device is provided with an A/D converter for converting the analog waveform of a voice signal inputted from a connected microphone to a digital signal, and digitized voice data of each conference participant is converted to sentence data by voice recognition techniques, and the sentence data is stored in a storage area of the network server while adding data indicating speaker's name and data indicating the time of his or her speech to the sentence data, thus generating minutes. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a minutes recording system for recording the remarks of a plurality of speakers in a conference using electronic technical means and automatically generating minutes.

[0002]

2. Description of the Related Art Conventionally, for the purpose of dictation and the like, there have been provided voice recognition devices or programs for converting the speech of a single specific speaker into characters and sentences. It is possible to identify the speaker from the audio of the conference in which the speakers of the above participate and identify the speaker who made the statement, and record the minutes of the statement of multiple speakers as one sentence. could not. For example, the technique disclosed in Japanese Patent Laid-Open No. 2000-276195 is intended to record the utterances of a plurality of speakers, but the utterances of individual speakers are not associated with each other but individually stored in a plurality of storage devices. Since it is recorded in, it is only possible to generate sentences that differ from the normal minutes format.

[0003]

[Problems to be Solved by the Invention] Usually, a meeting record such as a meeting, a meeting, a dialogue, an interview, etc. created by a human being and a minutes record include all the statements of a plurality of speakers as a series of sentences in chronological order. ing. Therefore, even in the minutes automatically generated by the electronic device, it is desired to record and save the name of the speaker or a symbol indicating the speaker in a correlated form.

[0004]

[Means for Solving the Problems] The same number of computer terminals equipped with a microphone and a communication network connection function as the participants of the conference are prepared, and the voice data obtained by digitizing the speech during the conference and the natural language of the voice data by the voice recognition function. After adding character and text data converted into sentences and data that encodes the name of the speaker corresponding to that data or a symbol that identifies the speaker and data that indicates the time of speech,
It is configured to record and store and perform rearrangement processing in a time series. With this configuration, a minutes document having a normally desired format can be obtained.

In addition, information necessary for identifying each individual computer terminal used for collecting voice data is stored in advance in the computer information of the user, the network address of the computer terminal or the computer terminal. It is configured to be obtained from a unique number.

[0006]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below.

(First Embodiment) FIG. 1 shows a schematic configuration of a minutes recording system of the present invention. The present invention relates to a plurality of microphones 101 and a computer terminal device 1.
02, a data processing server 103, a network server 104, and a communication network system 105 for connecting them. The data processing server 103 and the network server 104 can be shared by one computer.

A plurality of computer terminal devices 102 are prepared according to the number of participants in the conference. The computer terminal device 102 has a built-in or connectable microphone 101 capable of recording voice.
It is equipped with an A / D conversion device for converting an analog signal of voice input from 01 into a digital signal. Even if the A / D conversion device is built in the computer terminal device 102,
The present invention can be configured even if it is built in the microphone 101 and is connected to the computer terminal device 102 by a digital signal. Many of the recent computer devices have both an analog audio input terminal and a USB standard or IEEE 1394 standard digital signal input terminal. Further, the microphone 101 and the computer terminal device 10
The connection with 2 may be a wired system using a coaxial cable or the like, an analog system wireless such as FM radio waves, or a digital system wireless such as Bluetooth system. When the microphone 101 is provided outside the computer terminal device 102, as shown in FIG. 2, a headset-type microphone 202 to be worn on the heads of the conference participants 201, a pin microphone to be worn near the chest of the clothes. The voice can be clearly recorded by adopting a form such as a microphone 203 or a wireless pin microphone 205, or by installing a microphone 204 having a narrow directivity toward the head of each conference participant 201.

Further, the computer terminal device 102 has an interface device connectable to a communication network system 105 based on the TCP / IP protocol, for example. Even if the interface device is a wired system such as Ethernet (registered trademark), the IEEE80
A so-called wireless LAN system such as a 2.11 series or Bluetooth system may be used. Such network interface devices have unique address numbers and do not overlap each other.

In FIG. 1, a computer terminal device 1
Reference numeral 02 is a configuration for connecting to a communication network via the wireless access point 106. Further, the form of this computer terminal device 102 is a desktop (stationary).
Type, notebook type (portable) type, PDA (electronic notebook) type,
It may be a wristwatch (wrist terminal) type, etc.,
If a smaller computer terminal could be used,
It is clear that the convenience of meeting management is high. Recent advances in semiconductor processing devices (microprocessors) have provided devices that have a sufficient processing speed even if they are small.

The storage area of the computer terminal device 102 is used as the speaker tag 304 shown in the schematic diagram of the utterance data in FIG. 3, and therefore, the name of each user or a unique name for identifying the user is used. Remember in advance. As a method of storing this, input is performed from the keyboard 108 or voice from the microphone 101. Alternatively, if the conference participants each have a PDA (electronic notebook, portable electronic device) or wrist terminal (wristwatch electronic device) separately from the minutes recording system, the data consisting of character codes is converted into a scale. By using the function emitted from the speaker, the scale converted from the user's name stored in those portable devices is input from the microphone 101 connected to the minutes recording system, whereby the speaker's name and each microphone are input. The voice data input from 101 can be associated. The computer terminal device 102 can add a speaker tag and a time tag 305 indicating the time of the speech to the digitized voice data and transmit the data to the communication network system 105. In order for the time represented by the time tag 305 to be accurate, each computer terminal device 102 acquires time data from the server through the network system and corrects the internal clock of each computer terminal device 102 in advance.

The data processing server 103 has the waveform data 303 transmitted from the computer terminal device 102.
Processing to convert text data into text data using voice recognition technology,
Speaker tag 304 and time tag 305 added to the data
The data is rearranged (sorting) based on the above information so that the data is in an appropriate order. In particular, when each computer terminal device 102 has a limited processing capacity of a processor such as a wrist terminal or an electronic notebook, the text conversion processing by the voice recognition technology that requires a large processing capacity requires the high-performance data processing server 103. To use.

The network server 104 includes a mass storage device such as the magnetic disk device 108, and stores and accumulates the above waveform data 303 and text data. When the magnetic disk device 109 provided in the data processing server 103 has a sufficiently large capacity, the function of the network server 104 can also be used.

Next, the operation of the embodiment of the present invention will be described.

First, the name information of the terminal user stored in the computer terminal device 102 or the voice of the conference participant's own name, the code and network address by which the name of the speaker or the speaker can be specified by the voice recognition technology. A speaker address correspondence table for associating (IP address) is created. The speaker address correspondence table is created by the data processing server 103, and the data processing server 1
03 or the network server 104.
When using a means in which the speaker tag is not added to the data transmitted from the computer terminal device 102 to the communication network system 105, the speaker address 304 corresponding to the network address of the waveform data 303 and the sentence data 312 is converted to the waveform data 303 and the sentence data. Three
It is added to 12 and a speaker address correspondence table is used for identification.

Next, a conference, a dialogue or an interview is started. Voice is input from the microphone 101, and A / D conversion is started when the signal level exceeds a certain value. The voice of the speaker is input to the microphone 101, converted into an analog electric signal, and further converted into a digital signal by A / D conversion as described above. The voltage amplitude of the analog signal waveform is converted from 12 bits to digital data represented by a digital value of about 24 bits, and the header information 302 such as the source address and the destination address is converted into this.
And a time tag 305 indicating the time when the voice is recorded is added and transmitted to the data processing server 103 via the communication network system 105. Each computer terminal device 10 is provided with a switch or push button that gives an instruction to transmit data only when pressed or vice versa.
By preparing for No. 2, the speaker can select whether or not to record his / her own message.

When a plurality of switches or push buttons are assigned a function of expressing their intentions such as "agree / agree / agree", "refute / disagree", "question / question", etc., when any of those buttons is pushed The statement of intention is added to the data block with the intention expressing tag 306 indicating the intention and transmitted to the network system. With this function, the intention of the speaker can be clearly recorded in the minutes. These push buttons may be assigned the above-mentioned functions to the push buttons included in the keyboard 107, the mouse, or the PDA included in an ordinary computer device.

The data processing server 103 uses the waveform data 3
When 03 is received, a speaker tag is generated from the source address and speaker address correspondence table of the header information, added to the waveform data 303, and stored in a storage area such as a disk device. Further, the waveform data 303 is converted into sentence data in natural language by the voice recognition technique, and the speaker tag 304, the time tag 305, and the intention expression tag 306 are added and stored. Further, the converted text data is immediately transferred from the network system to each computer terminal device 102 and displayed on the screen.
By reading this, the speaker can make a statement to correct it if his / her statement is erroneously converted. The data processing server 103 accumulates the utterance during the conference in a storage area such as a disk device by repeating the above operation each time the utterance of the conference participant is made. The storage area accumulated at this time is the data processing server 10
A disk device 109 or the like mounted on the network server 104 other than 3 may be used.

Now, in the voice recognition technology for converting voice waveform data into character / sentence data, a dictionary file for each speaker, which stores words with high frequency of appearance and words in the speaker's specialized field, is used. By referring to it, the conversion efficiency can be further improved. To this end, in advance, a pattern file containing pronunciation patterns of conference participants and a user dictionary file accumulating specialized fields, frequently used words, and idioms are stored in each computer terminal device 10.
2 in the storage area, and the data processing server 103 stores these files in the communication network system 10.
5 or transfer or copy to the storage area of the data processing server 103 before the conference starts, and then reference.

The series of utterance data accumulated as described above is rearranged in the order of utterances according to the time information of the time tag 305, and the name of the speaker or a symbol for identifying the speaker and the intention or intention expressed by the speaker are displayed. The minutes are generated by adding the indicated symbol to the sentence of the remark and outputting it as one document file.

In addition, the minutes are obtained by adding a code or a name for identifying the speaker to the documented data sorted in the order of occurrence time, and using a large data such as a plasma display or an image data projector installed in the conference room. The text may be displayed on the screen of the screen image display device immediately.

(Second Embodiment) The speaker address correspondence table in the first embodiment is not used, and the computer terminal device 102 speaks from the data indicating the user stored in its internal storage area. Every time a speaker tag is generated and a packet of data is sent to the communication network system 105, a speaker tag is added to the data.

(Third Embodiment) In most computer products in recent years, a high-performance microprocessor is mounted even in a portable type or notebook type personal computer, and an electronic notebook or PD is used.
Even in a portable electronic device such as A, the processing capability can be improved by mounting a digital signal processor specialized for signal processing in addition to the main processor. Text conversion processing using voice recognition technology is possible. An electronic notebook or PDA may have a small storage area and may not be able to store a dictionary file having a sufficient amount of information necessary for voice recognition technology.

In such a case, by referring to the dictionary file stored in the storage area of the data processing server 103 or the network server 104 through the communication network system 105, a dictionary file of a sufficient size can be used. The waveform data 303 of the utterance is A / D converted by each computer terminal device 102, converted into sentence data by the voice recognition technology, and then the speaker tag 304,
A time tag 305 and the like are added. The utterance data obtained by adding the speaker tag 304, the time tag 305, etc. to the text data is transmitted to the data processing server 103 via the communication network system 105. Data processing server 10
3 rearranges the utterance data based on the information of the time tag 305, and stores and saves the data. In this embodiment, the data processing server 103 uses the waveform data 30
It is not necessary to convert 3 into text data.

(Fourth Embodiment) If each computer terminal device 102 can be connected to the Internet 112 or a wide area network via the previous communication network system 105 and a network connection device such as a router 111, each computer terminal device can be connected. The data processing server 113 and the network server 114 are located at a location remote from the conference room where the device 102 is used and the conference is held.
Can be installed. According to this embodiment, for example, in a company or the like, a meeting is held at a business office or a business office remote from the head office where the data processing server 113 is installed, and business offices in which participants of the meeting are different from each other. It is also possible to hold a meeting while at the sales office.

A computer having a particularly large processing speed and storage area is used as the data processing server 113.
By installing the computer, each computer terminal device 102
Even if the processing capability of is weak, it is possible to increase the speed of sentence conversion processing by voice recognition. Furthermore, the data processing server 113 as in this embodiment
Can be used as a business such as renting the processing service of (1) or using the waveform data 303 of the customer's meeting as the minutes.

When carrying out such services and businesses,
Similar to the method described in the first embodiment, the pronunciation pattern file of the speaker and the user dictionary are stored in the data processing server 11 via the Internet or a wide area network.
3 is referred to or copied to the storage area of the data processing server 113, the conversion efficiency of the voice recognition technology can be improved.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram of a minutes recording system of the present invention.

FIG. 2 is an explanatory diagram showing an arrangement of microphones according to the embodiment of the present invention.

FIG. 3A is a schematic diagram showing a structure of audio data.
FIG. B is a schematic diagram showing the structure of utterance data.

[Explanation of symbols] 101 ... Microphone 102 ... Computer terminal device 103 ・ ・ ・ Data processing server 104 ・ ・ ・ Network server 105 ... communication network system 106 ... Wireless access point 108 ... Keyboard 109 ... Disk device 201 ・ ・ ・ Participation in the conference 202 ・ ・ ・ Headset type microphone 203 ... pin microphone type microphone 204 ... Microphone with narrow directivity 205 ・ ・ ・ Wireless pin microphone 301 ・ ・ ・ Voice data 302 ・ ・ ・ Header 303 ・ ・ ・ Waveform data 304 ... Speaker tag 305 ・ ・ ・ Time tag 306 ... Indication of intention tag 311 ... utterance data 312 ・ ・ ・ Sentence data

─────────────────────────────────────────────────── ─── Continued Front Page (51) Int.Cl. ⁷ Identification Code FI Theme Coat (Reference) G10L 3/00 521P

Claims (14)

[Claims] 1. A plurality of notebook-type, notebook-type or list-type computer terminal devices, the number of which is equal to the number of participants of the conference, a data processing server, a wired or wireless communication network system, and a network server. The computer terminal device is connected to a wired or wireless communication network connection function and a microphone capable of inputting the voice of each participant's speech during the meeting. And an A / D converter that converts the analog waveform of the input voice signal into a digital signal, which is input from the microphone installed near each conference participant and digitized by the A / D converter. The voice data of each of the conference participants who were Natural performs processing for converting language into text data consisting of character codes representing the adds the data indicating the sign or name to identify a speaker corresponding to each of the data coded with data time of the speech,
A minutes recording system characterized by storing and accumulating the voice data and the text data subjected to the voice recognition in a storage area of a data processing server or a storage area of a network server. 2. The minutes recording system according to claim 1, wherein the speaker is specified from the user name registered in advance in the computer terminal device into which the voice is input, whereby the digitized voice data and the document data are recorded. The minutes recording system is characterized in that the code for identifying the speaker or the coded name is added to and stored in the storage area of the data processing server or the storage area of the network server. 3. The minutes recording system according to claim 1, wherein the speaker is specified from the network address unique to the computer terminal device to which the voice is input, whereby the digitized voice data and the document data are
A minutes recording system characterized in that the code for identifying the speaker or the coded name is added and stored in the storage area of the data processing server or the storage area of the network server. 4. The minutes recording system according to claim 1, wherein a speaker inputs his / her name by voice from the microphone, and data for identifying the speaker is generated from the voice data by voice recognition technology. A minutes recording system characterized by the following. 5. The minutes recording system according to claim 2 or 3, wherein the data processing server is based on the digitized voice data and the data indicating the time of speech added to the documented data. And a document data are rearranged in order of time, and are stored and accumulated in the storage area of the data processing server or the storage area of the network server. 6. The minutes recording system according to claim 4, wherein the coded data or the name for identifying the speaker is added to the documented data sorted in the order of occurrence time, and immediately displayed on the screen of each computer terminal device. A minutes recording system characterized by being displayed as text on the. 7. The minutes recording system according to claim 4, wherein a code or name for identifying the speaker is added to the documented data sorted in order of time of occurrence, and a plasma display installed in the conference room or A minutes recording system characterized by instantly displaying as text on the screen of a large-screen image display device such as image data projector. 8. The minutes recording system according to claim 1, wherein the processor of the data processing server is used to convert the voice data into text data by voice recognition technology. A minutes recording system, wherein a dictionary file stored in each of the computer terminal devices is referred to in the conversion processing into. 9. The minutes recording system according to claim 1, wherein the processor of each computer terminal device assigned to each conference participant is used to convert each voice data into text data by voice recognition technology. It is a minutes recording system that stores and stores in the storage area of the network server after transferring it to the network system, and then storing it in the storage area of the network server. A minutes recording system characterized by referring to a dictionary file stored in. 10. The minutes recording system according to claim 1, wherein each computer terminal device or microphone device assigned to each conference participant is provided with a button or a switch, and a specific button or switch is pressed for a certain period of time. A minutes recording system characterized by not recording utterances. 11. The minutes recording system according to claim 1, wherein each computer terminal device or microphone device assigned to each conference participant is provided with a button or switch, and A minutes recording system characterized by recording utterances. 12. The minutes recording system according to claim 1, wherein each computer terminal device assigned to each conference participant is provided with a plurality of buttons or switches.
It is characterized in that information corresponding to each button is added and recorded to voice data or its text data obtained by digitizing the speech during the time when a specific button or switch selected by the speaker is pressed, or the speech immediately after the switch is pressed. A minutes recording system. 13. The minutes recording system according to claim 1, 2, or 3, wherein the communication network system is connectable to a wide area network system or the Internet via a router device or the like, and is collected by the computer terminal device. The voice data of the speech of the attended conference participant is converted into natural language character / sentence data by the voice recognition function of the data processing server via the wireless or wired communication network system and wide area network system or the Internet. A minutes recording system, which is stored and stored in a storage area of the network server. 14. The minutes recording system according to claim 13, wherein the data processing server stores a dictionary file stored in each computer terminal device and a dictionary file stored in the data processing server via a wide area network system or the Internet. Referring to, a minutes recording system characterized by performing a process of converting voice data into text data by a voice recognition technology.