KR20220122100A - Conference assistance system according to the type of multiple languages - Google Patents

Abstract

The present invention relates to a conference assistance system according to the type of multilanguage and, more specifically, to a conference assistance system capable of storing conference schedules from multiple speakers and storing and classifying minutes as an automatically created text. In accordance with one embodiment of the present invention, the conference assistance system includes: a voice signal input unit into which voice signals generated from a plurality of speakers are inputted; a text conversion unit converting the voice signals inputted from the voice signal input unit, into a text; a word extraction unit extracting a specific word from the conversation text converted through the text conversion unit; a storage unit inputting and storing scheduling information from the specific word extracted by the word extraction unit; and an output unit outputting the scheduling information stored in the storage unit.

Description

Conference assistance system according to the type of multiple languages

The present invention relates to a conference assistance system according to the types of multiple languages, and more particularly, to a conference assistance system for storing meeting schedules from a plurality of speakers and automatically generating meeting minutes as generated text and distinguishing them.

Unless otherwise indicated herein, the material described in this section is not prior art to the claims of this application, and inclusion in this section is not an admission that it is prior art.

A prerequisite for many companies to promote mutual collaboration is to schedule a meeting. Not only meeting schedules between domestic companies, but also meeting schedules with multinational companies has become one of the current tasks. Scheduling has been made by adjusting the schedule and delivering it to the meeting attendees.

However, with the recent development of the automation system, the demand for an automation system that can schedule or adjust a meeting schedule and automatically store it in the scheduler is also increasing.

As an example of this, Korean Patent Laid-Open No. 10-2019-0004526 discloses "Meeting Scheduling Method, Electronic Device and System", and in more detail, "Past related to conference reservation information received from the first terminal" extracting a participant list from the meeting history and transmitting it to the first terminal; a schedule generated based on the schedule for at least one participant according to a selection signal for at least one participant included in the participant list received from the first terminal Transmitting the list to the first terminal, at least one schedule to at least one second terminal corresponding to the at least one participant according to a selection signal for at least one schedule included in the schedule list received from the first terminal It includes the steps of requesting a vote for the second terminal and performing meeting scheduling according to the voting result received from the at least one second terminal."

1. Korean Patent Publication No. 10-2019-0004526 (published on Jan. 14, 2019)

An object of the present invention is to provide a meeting assistance system capable of automatically generating a meeting schedule from a plurality of speakers and automatically creating meeting minutes.

In addition, it is not limited to the technical problems as described above, and it is obvious that another technical problem may be derived from the following description.

A conference assistance system according to an embodiment of the present invention includes a voice signal input unit to which voice signals generated from a plurality of speakers are input, a text converter for converting the voice signals input from the voice signal input unit into text, and the text A word extraction unit for extracting a specific word from the conversation text converted through the conversion unit, a storage unit for inputting and storing scheduling information from the specific word extracted from the word extraction unit, and an output unit for outputting the scheduling information stored in the storage unit include

According to a preferred feature of the present invention, the specific word extracted by the word extraction unit is characterized in that it consists of any one or more words selected from among date, time, place, subject, number of people, and meeting method.

According to a preferred feature of the present invention, the scheduling information output from the output unit is automatically stored in a scheduler installed in the user's mobile terminal, and the scheduler is provided with detailed information on place, subject, number of people, and meeting method. characterized.

According to a preferred feature of the present invention, after the meeting is made, it is characterized in that it comprises a meeting minutes generator for generating the minutes by converting the voice signals generated from the plurality of speakers into text through the text conversion unit.

According to a preferred feature of the present invention, the minutes generated by the minutes generator are automatically stored in the scheduler.

According to a preferred feature of the present invention, the meeting minutes generating unit includes a speaker classification unit that extracts voice frequencies of a plurality of speakers input from the voice signal input unit, respectively, and analyzes frequency bands to classify the plurality of speakers; It is characterized in that it comprises an STT conversion unit that converts the dialogue between the speakers separated from the part into text by using STT and a meeting contents listing unit that lists the dialogue texts between the speakers separated through the STT conversion unit in time series.

According to a preferred feature of the present invention, the dialogue text output through the conference content listing unit is stored after being included in the voice dialogue after the conference is over, and the dialogue text provided by a plurality of speakers is arranged in time series through the output unit. It is characterized in that it is provided to users' mobile terminals.

According to a preferred feature of the present invention, the speaker classification unit pre-classifies speakers according to the type of language when multiple languages are input from a plurality of speakers through the voice signal input unit, and extracts and It is characterized in that the speaker is post-classified by analyzing the frequency band.

According to a preferred feature of the present invention, when a language other than a preset language is recognized from the conversation text between the classified speakers, the language is translated into a preset language and provided together with the dialog text.

According to a preferred feature of the present invention, it is characterized in that repeatedly used words are extracted from the dialogue text between the speakers converted through the STT conversion unit, and the extracted words are categorized and stored.

According to a preferred feature of the present invention, it is characterized in that a word repeatedly used is extracted by machine learning the dialogue text, and the extracted word is categorized and stored.

According to the embodiment of the present invention, there is an advantage in that the meeting schedule and contents can be more easily managed by automatically generating the meeting schedule and automatically creating and storing the meeting minutes at the end of the meeting in the scheduler.

In addition, according to an embodiment of the present invention, conversations generated from a plurality of speakers are divided by speaker and stored as text, and the divided conversation contents are provided in real time through the speakers' mobile terminals, etc., so that a desired word or sentence can be easily found It has the advantage of being able to see.

The effects of the present invention are not limited to the above effects, and it should be understood to include all effects that can be inferred from the configuration of the invention described in the detailed description or claims of the present invention.

1 is a conceptual diagram of a conference assistance system according to an embodiment of the present invention.
Figure 2 is a block diagram of a conference assistance system according to an embodiment of the present invention.
3 and 4 are detailed block diagrams of a conference assistance system according to an embodiment of the present invention.

Hereinafter, with reference to the accompanying drawings, the configuration, operation and effect of the conference assistance system according to a preferred embodiment will be described. For reference, in the following drawings, each component is omitted or schematically illustrated for convenience and clarity, and the size of each component does not reflect the actual size. In addition, the same reference numerals refer to the same components throughout the specification, and reference numerals for the same components in individual drawings will be omitted.

The conference assistance system 100 according to an embodiment of the present invention includes a voice signal input unit 110 to which voice signals generated from a plurality of speakers are input, and a voice signal input from the voice signal input unit 110 to text. A text conversion unit 120 for converting, a word extraction unit 130 for extracting a specific word from the conversation text converted through the text conversion unit 120, and a specific word extracted from the word extraction unit 130 and a storage unit 140 for inputting and storing scheduling information, and an output unit 150 for outputting scheduling information stored in the storage unit 140 .

First, it may include a voice signal input unit 110 to which voice signals generated from a plurality of speakers are input. In the voice signal input unit 110 , when a plurality of speakers communicate with each other through the user's portable terminal, a voice signal generated from the conversation is input through the voice signal input unit 110 . Here, the voice signal input unit 110 is configured to be able to input a voice signal including various frequencies, and when the voice signal is input, the voice signal is converted into a data value.

Next, the voice signal input from the voice signal input unit 110 is converted into text by the text conversion unit 120 . The text conversion unit 120 is configured to convert speech into text through a speech-to-text method, and analyzes information such as pronunciation, intonation, and length of speech data and converts it into text information.

In addition, it includes a word extraction unit 130 for extracting a specific word from the conversation text converted through the text conversion unit 120. The word extraction unit 130 extracts a preset specific word from the text data converted from the dialogue between a plurality of speakers and voice information for the dialogue. Here, the specific word is characterized in that it consists of any one or more words selected from the date, time, place, subject, number, and meeting method. For example, when two speakers schedule a meeting while talking to each other, when referring to the date, time, and meeting place, the word is extracted from the text converted by the text conversion unit 120 to obtain meeting information. do. In the corresponding meeting information, not only the date, time and place, but also information on the topic of the meeting, the number of people attending the meeting, whether the meeting method is online or offline, etc. can be extracted together. If you schedule a meeting by mentioning the word, the scheduling is done in an online meeting method.

Next, it includes a storage unit 140 for inputting and storing scheduling information from a specific word extracted from the word extraction unit 130 , and an output unit 150 for outputting scheduling information stored in the storage unit 140 . . When a specific word related to the above-mentioned meeting information is extracted, scheduling information is input and stored in the storage unit 140 from the information. It can be simply and clearly marked and stored, and the stored scheduling information is output through the output unit 150 . The output unit 150 may be any device including a display device, such as a user's portable terminal, and the scheduling information output through the output unit 150 may be output through a calendar or scheduler installed in the user's portable terminal. can

According to a preferred feature of the present invention, the scheduling information output from the output unit 150 is automatically stored in a scheduler installed in the user's mobile terminal, and the scheduler includes detailed information about place, subject, number of people, and meeting method together. characterized in that it is provided.

The scheduling information may be information that includes all information related to the meeting, such as meeting date, time, place, and topic, and the detailed information is automatically stored in the scheduler installed in the user's mobile terminal and provided so that the meeting schedule can be mutually negotiated over the wire. At the same time, there is an advantage that the negotiated schedule can be checked immediately through the scheduler of the user's portable terminal.

According to a preferred feature of the present invention, after the meeting is made, the voice signal generated from a plurality of speakers is converted into text through the text conversion unit 120, and a meeting minutes generating unit 200 for generating the meeting minutes is characterized. do it with

After the contents of the meeting schedule are stored in the scheduler through the above-described meeting assistance system 100 , it is necessary to store the contents discussed in the meeting as minutes after the actual meeting is made. However, it is not easy for everyone who attended the meeting to record and organize all the conversations that occurred during the meeting, so there was a need for a system that automatically records and stores them. Accordingly, the meeting minutes generating unit 200 converts the voice signals generated from a plurality of speakers during the meeting into text through the text converting unit 120 to generate the meeting minutes.

In addition, according to a preferred feature of the present invention, the minutes generated by the minutes generation unit 200 is characterized in that automatically stored in the scheduler. The meeting minutes generation unit 200 allows the minutes generated through the conversation text converted through the text conversion unit 120 to be automatically stored in the scheduler for the contents mentioned by a plurality of speakers, so that the schedule-related application installed in the mobile terminal Alternatively, you can check the schedule for a meeting on a specific date through the scheduler, and check the contents of the meeting through the recorded meeting minutes.

According to a preferred feature of the present invention, the meeting minutes generating unit 200 extracts voice frequencies of a plurality of speakers input from the voice signal input unit 110, respectively, and analyzes frequency bands to distinguish the plurality of speakers. The division unit 210, the STT conversion unit 220 that converts the conversation between the speakers separated from the speaker division unit 210 into text using STT, and the STT conversion unit 220 between the divided speakers It is characterized in that it includes a meeting contents listing unit 230 that lists the dialogue text in time series.

First, the voice signal input through the voice signal input unit 110 is converted into a data value and stored. Based on this, the speaker classification unit 210 extracts the voice frequencies of a plurality of speakers, respectively, and analyzes the frequency band. Distinguish between multiple speakers. The speaker classification unit 210 extracts the voice frequency of the voice signal input through the voice signal input unit 110 and analyzes the frequency band, which is by analyzing the frequency band from factors such as sound length, pitch, and power. speaker can be distinguished. Through the above method, a plurality of speakers are distinguished from each other, and the conversation between the divided speakers is converted into text through the STT conversion unit 220 .

The STT conversion unit 220 is configured to convert speech into text through Speech-to-Text, and analyzes information such as pronunciation, intonation, and length of speech data and converts it into text information. The dialogue in which the speaker is separated from the speaker classification unit 210 converts the dialogue into text through the STT conversion unit 220, and is provided by arranging it as a sentence. Finally, the STT conversion unit 220 includes a conference content listing unit 230 that lists the dialogue texts between the speakers in time series. The meeting contents list unit 230 converts the contents mentioned by each speaker into dialogue text in a state where several speakers are separated from each other through a display device such as a user's portable terminal, so that the meeting contents listed in time series can be provided. have.

According to a preferred feature of the present invention, the dialogue text output through the conference content listing unit 230 is included in the voice dialogue after the conference is ended and stored, and the dialogue text provided by a plurality of speakers is arranged in time series. Thus, it is characterized in that it is provided to users' portable terminals through the output unit 150 .

Conversation text provided to the portable terminal through the output unit 140 may be included and stored in the voice recorded file, and when the corresponding voice file is played back after the meeting is over, the recorded conversation content is played back and at the same time , a dialogue text matching the contents of the conversation is provided together through the screen of the mobile terminal. The dialogue text provided here is divided into a plurality of speakers, the dialogue text corresponding to the dialogue contents mentioned by each speaker is divided, and the dialogue text is arranged in time series to output.

According to a preferred feature of the present invention, the speaker classification unit 210 pre-classifies the speakers according to the type of language when multiple languages are input from a plurality of speakers through the voice signal input unit 110, and the classified speakers It is characterized in that the speaker is post-classified by extracting the voice frequency between them and analyzing the frequency band.

A voice signal may be input from a plurality of speakers through the voice signal input unit 110. When a meeting is held through the present invention between people of different nationalities residing in various countries, such as in an international conference, Korean from a plurality of speakers and multiple languages such as English, Japanese, and Chinese can be input. When multiple languages are input in this way, the task of pre-classifying the speakers according to the type of language can be done first. In the case of conducting this group meeting, the speaker classification unit 210, which recognizes that there are multiple languages from the input voice signal, first pre-classifies the speakers according to the type of language (by Korean, English, and Japanese), After extracting the voice frequencies between the speakers and analyzing the frequency bands, the pre-classified speakers are again classified for each speaker, followed by a post-classification operation. When the speakers are divided in this way, in the case of multinational conversations, the division between speakers is first made according to the type of language, and then, according to the needs of the user, only the conversation contents for a specific language can be extracted and searched and checked. There is an advantage that

According to a preferred feature of the present invention, when a language other than a preset language is recognized from the conversation text between the classified speakers, the language is translated into a preset language and provided together with the dialog text.

After the meeting ends, the meeting minutes are provided from the meeting contents listing unit 230 to the user's mobile terminal through the output unit 150. If not, there is a limit to the search. Accordingly, when a specific language is preset through the user's mobile terminal and a language other than the corresponding language is input through the voice signal input unit 110 and recognized, the translated dialogue text is translated into the preset language may be provided together. The translated dialogue text provides the translated dialogue text along with the dialogue text for the corresponding language while the recorded file is played after the meeting is ended, so that the contents of the conversation can be more easily searched.

According to a preferred feature of the present invention, it further comprises a storage unit for extracting repeatedly used words from the dialogue text between the speakers converted through the STT conversion unit 220, and categorizing and storing the extracted words. do it with

It is necessary to classify and store the files in which the meeting contents are recorded from the conversation text by category. For example, when categorizing and storing frequently mentioned words or topics, words repeatedly used from the conversation text between the speakers converted through the STT conversion unit 220 (eg, a specific place or name) etc.), and categorizes and stores the extracted repeatedly used words. Accordingly, if you want to search the conversation text stored together with the recorded file after the meeting is over, there is an advantage in that it is possible to more easily find the classified recorded file among a plurality of recorded files to search for the desired dialogue text.

According to a preferred feature of the present invention, it is characterized in that a word repeatedly used is extracted by machine learning the dialogue text, and the extracted word is categorized and stored.

When categorizing from words that are repeatedly used from the dialogue text, a machine learning function may be utilized to more accurately and quickly categorize. The machine learning function is defined as an algorithm that analyzes data and applies what it learns to make a decision based on self-learning information. By extracting frequently used words, it is possible to categorize and store the extracted words. It has the advantage that users can find the conversation text they want faster and easier by extracting frequently used words more accurately and quickly through the machine learning function and storing them as categories.

Although the preferred embodiment of the present invention has been described with reference to the accompanying drawings, the embodiments described in the present specification and the configuration shown in the drawings are only the most preferred embodiment of the present invention and represent all of the technical spirit of the present invention. Therefore, it should be understood that there may be various equivalents and modifications that can be substituted for them at the time of filing the present application. Therefore, the embodiments described above are to be understood as illustrative and not restrictive in all respects, and the scope of the present invention is indicated by the following claims rather than the detailed description, and the meaning and scope of the claims and their All changes or modifications derived from the concept of equivalents should be construed as being included in the scope of the present invention.

100: meeting assistance system
110: voice signal input unit
120: text conversion unit
130: word extraction unit
140: storage
150: output unit
200: meeting minutes generating unit
210: speaker division
220: STT conversion unit
230: meeting contents list part

Claims (9)

a voice signal input unit to which voice signals generated from a plurality of speakers are input;
a text converting unit converting the voice signal input from the voice signal input unit into text;
a word extraction unit for extracting a specific word from the conversation text converted through the text conversion unit;
a storage unit for inputting and storing scheduling information from the specific word extracted from the word extraction unit; and
an output unit for outputting the scheduling information stored in the storage unit;
Meeting assistance system that includes. According to claim 1,
The specific word extracted by the word extraction unit is,
Meeting assistance system, characterized in that it consists of any one or more words selected from date, time, place, subject, number of people, and meeting method. According to claim 1,
The scheduling information output from the output unit is,
It is automatically stored in a scheduler installed in the user's mobile terminal, and the scheduler is provided with detailed information on place, subject, number of people, and meeting method. 4. The method of claim 3,
After the meeting is made, the meeting assistant system comprising a meeting minutes generator for generating the minutes by converting the voice signals generated by the plurality of speakers into text through the text conversion unit. 5. The method of claim 4,
The meeting minutes generated by the meeting minutes generator are automatically stored in the scheduler. 5. The method of claim 4,
The meeting minutes generating unit,
a speaker classification unit for extracting voice frequencies of a plurality of speakers input from the voice signal input unit, respectively, and analyzing frequency bands to classify the plurality of speakers;
an STT conversion unit that converts the conversation between the speakers separated from the speaker classification unit into text using STT; and
Conference assistance system, characterized in that it comprises a meeting contents listing unit for chronologically listing the dialogue texts between the speakers divided through the STT conversion unit. 7. The method of claim 6,
Conversation text output through the conference contents listing unit is stored in the voice conversation after the end of the conference,
A conference assistance system, characterized in that the dialogue text provided by a plurality of speakers is arranged in time series and provided to users' portable terminals through an output unit. 7. The method of claim 6,
The speaker classification unit, when multiple languages are input from a plurality of speakers through the voice signal input unit, pre-classifies speakers according to the type of language, extracts voice frequencies between the classified speakers, and analyzes the frequency band to select the speakers. Meeting assistance system, characterized in that the classification. 9. The method of claim 8,
When a language other than a preset language is recognized from the classified conversation text between the speakers, the conference assistance system is translated into a preset language and provided together with the conversation text.

Images