JP7223469B1 - Utterance information documentation device - Google Patents

Abstract

A speech information documentation device is provided that allows easy grasp of important matters in speech.
SOLUTION: An utterance information documentation device 1 records voice of a video conference held between a user terminal 4 and a customer terminal 5, analyzes the video, and extracts important information from a plurality of utterances. , to create a summary sentence. Extraction of the summary utterance 32 is performed by the analysis means 12. If there is a classification of budget, authority, etc., or a switch from the explanation stage to the question-and-answer stage, etc., the analysis means 12 gives points to the utterance. . A predetermined number of utterances with high accumulated points are extracted and used as summary utterances 32 . The summarizing means 13 determines whether or not there is a relationship between the utterances before and after the utterance for summarization 32, if the same speaker is present (A). If the speakers are different, the sentence similarity is used for determination (B). In addition to summary utterances 32, related utterances 33 are combined and converted into short sentences to create a summary sentence 35. - 特許庁
[Selection drawing] Fig. 4

Description

Application of Article 30, Paragraph 2 of the Patent Law (1) July 7, 2020 software (Ver.1.6.2) release (2) July 20, 2020 software (Ver.1.6. 3) Release (3) July 8, 2020 Software (Ver.1.6.2) article published on the web (4) July 8, 2020 Software (Ver.1.6. 2) Article published on the web (5) August 24, 2020 Software (Ver.1.6.3) content released as video (6) August 25, 2020 Software (Ver. .1, 6, 3) will be published on the web (7) Exhibition at DXPO (Tokyo Big Sight) from August 24 to 25, 2022

TECHNICAL FIELD The present invention relates to a device for documenting the content of speech in a video conference, lecture, or the like.

In recent years, with the spread of telework and the like, opportunities to use video conferencing are increasing. Also, in sales activities, there are increasing opportunities for sales activities to be conducted not only by direct visits to customers by sales personnel, but also by using video conferences. Furthermore, online lectures and the like are being held frequently.

In Patent Document 1, when conducting business activities using a telephone, a conversation between a callee and a user is recorded, an utterance period in which voice exists is detected and extracted from the recorded data, and the extracted voice A sales support device is disclosed that performs element-based analysis based on data.

The apparatus of Patent Document 1 discloses, as elements, distinguishing between the voice of the user and the voice of the called party, the strength of intonation in conversation, the speed of speech, and the like. Conversation is digitized.

The apparatus of Patent Literature 1 makes it possible to capitalize the data thus collected as big data and utilize the big data for processing using AI (artificial intelligence). For example, it is possible to improve the quality of sales by providing real-time AI feedback during a call. Alternatively, it is disclosed that the analysis data is viewed by the manager to evaluate the salesperson, or that the analysis data is used for training other salespersons.

Japanese Patent No. 7053069

In the technology described in Patent Document 1, in telemarketing, it is possible to verify who made the meeting unsuccessful and for what reason, confirm the call when a problem occurs, or examine the speaking skills of an excellent salesperson with good sales performance. It is said that there is an effect such as being able to make the sales representatives learn.

However, although the technology disclosed in Japanese Patent Laid-Open No. 2002-200311 can verify the conference, it is difficult to grasp the key points of the conference, such as which conversation was the most important conversation in the conference. Also, even if the material is used as a training material for inexperienced sales personnel, it is difficult to learn which conversation leads to a contract because the main point of the conversation is unclear.

SUMMARY OF THE INVENTION In view of the above problems, it is an object of the present invention to provide an utterance information documentation apparatus that enables users to easily grasp the important matters of utterances in conferences and lectures. Further, the present invention provides an utterance information documentation device capable of providing training materials with high learning effects by clarifying important matters.

In order to achieve the above object, an utterance information documentation apparatus of the present invention is an apparatus that compiles one or more utterances into materials, comprising acquisition means for obtaining utterance information, analysis means for analyzing the utterance information, Summarizing means for creating a summary sentence of the utterance information based on the analysis result of the analysis means; and display means for displaying at least the summary sentence; Identifying which category it belongs to among categories, and identifying which stage the utterance belongs to among a plurality of stages of progress, and using information of at least one of the classification or the stage, a plurality of A summary utterance is extracted from the utterance, the summary means creates a summary sentence including the summary utterance, and the display means can display the summary sentence.

The speech information documentation device of the present invention extracts one or more speeches according to a predetermined standard from speech information including voices of conferences, lectures, etc., extracts summary speeches from the speeches, are summarized and documented. Therefore, even if the utterance is long, it is possible to easily grasp what is important in the utterance by looking at the summary using the summary utterance. Here, the classification of utterances means classifying the contents of utterances according to types. For example, in the case of a meeting, the budget and authority in business negotiations correspond, and in the case of technical meetings, product performance, production technology, and the like. In the case of a lecture, the classification is related to the content of the lecture. Further, in the present invention, the analysis means specifying the classification or the stage includes the case where both the classification and the stage are specified.

In the present invention, a summary utterance is an utterance that includes matters related to the purpose or outcome of the utterance, and includes utterances that play an important role in understanding the main points of the utterance, and is a classification of utterances. or guided using step information. In the present invention, an utterance includes a conference where multiple people have a conversation online (video conference) or offline (face-to-face). In addition, it includes not only the name "conference" but also what is called "meeting" or "meeting". Further, the conference includes a conversation using a telephone as well as a case of having a conversation through a personal computer (including a tablet terminal, etc.). Furthermore, in the present invention, utterances include lectures, presentations, etc. in which one or more speakers (lecturers, etc.) speak online or offline.

In the speech information documentation device of the present invention, the classification includes budget, authority, necessity, time of introduction, competition, and next action. These classifications are related to business negotiations, each of which is a budget related to expenses, authority related to the approver of business negotiations, needs related to customer needs, etc. in business negotiations, and products, etc. in business negotiations. It is the time to introduce the , the competitors in doing business, and the next action of what to do next. If an utterance contains wording related to such classification, the utterance is recognized as an utterance containing important matters. Moreover, by classifying the utterances in this way, it is possible to clarify in which classification the utterances are important.

In the speech information documentation apparatus of the present invention, the steps include an introduction step, an explanation step, a question-and-answer step, and a summary step . In the present invention, the above stages are respectively divided into an introductory stage (also called ice-breaking) relating to the introduction of utterances, an explanation stage mainly carried out by sales users, a question-and-answer stage in which users and customers proceed to talk, and a summary of utterances. It can be divided into stages (also called closing). By dividing the stages in this way, it is expected that there is a high possibility that the utterance that triggered the transition from one stage to another contains important matters. In addition, it is possible to clarify at what stage the utterance is important.

In the utterance information documentation apparatus of the present invention, the summarizing means may translate the utterance for summary into English, translate the English-translated utterance into Japanese, and create the summary sentence using the Japanese-translated utterance. Speech is spoken language, and Japanese tends to have many ambiguous expressions. Therefore, by translating the utterance into English once to make sentences with less ambiguity, and then translating into Japanese again, ambiguous expressions in the utterance can be reduced and sentences can be shortened.

In the utterance information documentation apparatus of the present invention, the analysis means assigns points according to the classification or the stage of the utterance, totals the points for each utterance, and collects the points for each utterance for summary. One or a plurality of utterances may be extracted.

Further, in the utterance information documentation apparatus of the present invention, the analysis means recognizes linguistic information including information about numerical values or dates and speaker information, which is information about a speaker, in the utterance, and recognizes the language in the utterance. Points may be given according to the information and the speaker information, the points may be totaled for each of the utterances, and one or more of the summarization utterances with the highest total points may be extracted.

According to the utterance information documentation device of the present invention, points are given to each utterance according to the classification or stage of the utterance, or the linguistic information or speaker information, and the utterances with high points are extracted, and from the extracted utterances Since a summary sentence is formed, the summary sentence is formed by utterances with high importance for the utterance. Therefore, by reading the summary sentence, the user can easily grasp important matters in the utterance.

In the utterance information documentation apparatus of the present invention, the summarizing means determines whether or not at least one or both of an utterance immediately before the utterance for summary and an utterance immediately after the utterance for summary are related to the utterance for summary. The utterance determined to be the additional utterance may be used as the additional utterance. By this processing, an utterance related to the utterance for summary can be added to the summary sentence, so that the content of the summary sentence can be further enriched.

At this time, the analysis means stores a plurality of utterances made by a plurality of speakers from the utterance information in association with the utterances and the speakers, and stores the summary utterance, the immediately preceding utterance, or the When the immediately following utterances are utterances by the same person, whether or not there is a relationship may be determined by judging the relevance of the adjacent utterances from conjunctions included in the adjacent utterances.

Further, the analysis means stores a plurality of utterances made by a plurality of speakers from the utterance information in association with the utterances and the speakers, and stores the summary utterance and the immediately preceding utterance or the immediately following utterance. When the utterances are utterances by different persons, the presence or absence of the relationship may determine the relevance of the adjacent utterances from the similarity of the wording of the adjacent utterances.

Explanatory drawing which shows the outline|summary of the speech information documentation apparatus of this embodiment. (A) and (B) are explanatory diagrams showing preparation screens in the speech information documentation device. Explanatory drawing which shows the conference screen in an utterance information documentation apparatus. FIG. 4 is an explanatory diagram showing the operation of the summarizing means when creating a summary sentence; FIG. 4 is an explanatory diagram showing a list management screen, which is a screen for using conferences documented.

Next, with reference to FIGS. 1 to 5, a speech information documentation apparatus according to an embodiment of the present invention will be described by taking a conference related to business negotiations as an example. As shown in FIG. 1, the utterance information documentation device 1 of the present embodiment is a device realized by each functional unit configured by a program stored in the server 2 and the hardware of the server 2. is connected to the user terminal 4 via the . A customer terminal 5 used by a customer of the user is also connected to the network 3 .

As an example, the speech information documentation device 1 of the present embodiment records the video and audio of a video conference held between the user terminal 4 and the customer terminal 5, analyzes the recorded audio, It extracts important information from utterances and creates a summary sentence.

The utterance information documentation device 1 of the present embodiment has, as a functional configuration, an acquisition means 11 for acquiring utterance information including speech in a meeting, an analysis means 12 for analyzing the acquired utterance information, and an analysis means 12 for analyzing the acquired utterance information. It comprises a summarizing means 13 for creating a summary 35 (see FIG. 4C) of utterance information based on the analysis result of the analyzing means 12, and a display means 14 for displaying various screens and the summary 35. FIG. The utterance information may include information such as video, date and time, attendees, and location information of the user terminal 4 and the customer terminal 5, in addition to the audio in the conference.

The server 2 includes a computer system (not shown) including a CPU (Central Processing Unit), hard disk, storage means such as memory, and means for connecting to various networks. is stored as the speech information documentation device 1 of the present embodiment. This computer program is executed by the CPU, and various processes described below are performed. The computer system includes so-called cloud computing. A computer program is a concept that includes not only programs stored in the server 2 but also programs using API (Application Programming Interface) linkage.

The analysis means 12 is a functional unit that analyzes speech information. Specifically, a plurality of utterances made by a plurality of speakers are extracted from the utterance information, and the utterances and the speakers are stored in association with each other. Speakers include the salesperson who is the user and the customer of the user. There may be one user and one customer each, or there may be multiple users.

In addition, the analysis means 12 classifies the speech to which class it belongs in the content of the conference. Specifically, when multiple utterances include words related to budget, authority, necessity, time of introduction, competition, and next action, the utterances are classified by these budgets and the like.

A budget corresponds to an utterance having a wording related to an amount such as an expense or an investment amount. The authority corresponds to an utterance having wording related to the decision-making power to decide budget execution, introduction timing, etc. at the meeting. A need corresponds to an utterance having wording centered on the customer's needs at the meeting. The time of introduction corresponds to an utterance having a phrase including the time of introduction of a product or the like in a meeting. A competition corresponds to an utterance including the name of a company other than the company to which the user or customer belongs, or words related to competition such as a competitor. The next action corresponds to an utterance containing words related to action to be taken after the meeting, such as treatment or execution.

In addition, the analysis unit 12 identifies which of the stages of progress in the conference the utterance corresponds to. The stages include an introduction stage, an explanation stage, a Q&A stage, and a summary stage.

The introductory stage is the stage in which greetings and self-introductions are performed in the first half of the meeting. The explanation stage is a stage in which the user mainly explains the company's products or services, and the user's utterances are mainly performed continuously. This explanation step is often done after the introduction step. The question-and-answer stage is a stage in which utterances by the user and utterances by the customer are alternately performed, and is often performed after the explanation stage. In addition, the Q&A stage is characterized in that both the user and the customer often ask questions. The closing stage is a stage related to the closing of the meeting, such as deciding on business negotiations, and is often held in the second half of the meeting.

The summarizing means 13 is a functional unit that forms a summary sentence 35 using the summary utterances 32 extracted by the analyzing means 12 and enables the display means 14 to display the summary sentence 35 on the user terminal 4 . The display unit 14 can display various screens such as the preparation screen 21 shown in FIG.

The network 3 is a communication network represented by the Internet, and is a concept including telephone lines and the like. The user terminal 4 and the customer terminal 5 may be a personal computer or the like as long as it can display a screen, and may be a terminal such as a desktop computer, a notebook computer, a tablet terminal, or a smartphone. .

Next, a method of documenting utterances in a conference by means of the utterance information documenting device 1 of the present embodiment will be described. The conference in the present embodiment will be described as an example where the user is a sales representative of a company that provides products or services to customers and conducts business negotiations with the customers via video conference.

First, the user performs preparation work using the preparation screen 21 shown in FIG. 2 before the meeting with the customer. As an example of the preparation work, on the preparation screen 21 shown in FIG. Enter your information.

Also, on the second preparation screen 22 shown in FIG. 2B, the purpose of the meeting and items to be confirmed with the customer can be input. The purpose of the meeting is entered in consideration of the content to be kept in mind at the meeting and the possibility of searching by purpose when documenting. This input may be directly entered in the input field, or may be selected from a pop-up menu displayed by clicking the input field.

Next, screens displayed on the user terminal 4 in an actual conference will be described with reference to FIG. On the conference screen 23 displayed on the user terminal 4, the title of the conference, the date and time, customer information, the user name, the name of the person in charge of the customer, etc. are displayed in the information display column 24 at the upper left.

In FIG. 3, as an example, a face display screen 25 when there are four participants, a conversation sentence display screen 26 for displaying conversation sentences in which voices spoken in the conference are transcribed into character information, and a conference purpose are shown. A purpose screen 27 to be displayed, a confirmation item screen 28 describing items to be confirmed in the conference, and a screen on which a highlight display 29 displaying utterances considered to be important in the conference are displayed.

On the face display screen 25, facial expressions of each person in charge in the conference can be displayed in real time, and voice recording and video recording are also performed at the same time. On the dialogue display screen 26, every time a user or a customer makes a statement, the content of the statement is displayed as character information as an utterance 31. FIG. The displayed character information displays the time when the speech was made, and is stored in the server 2 in association with the information of the speaker who made the speech 31 .

The purpose screen 27 displays the items such as the purpose of the meeting input on the preparation screen 21 . In this way, by displaying the purpose of the meeting on the screen during the meeting, the user can proceed without deviating from the main topic of the meeting.

For example, as shown in FIG. 3, the confirmation item screen 28 displays the contents of a need 28a, a competition 28b, and an authorizer 28c. The needs 28a include questions about the background and issues leading up to the current meeting, questions about the operation of negotiations to be discussed in the current meeting, and questions about the role of the person in charge of the customer.

The competition 28b includes a question as to whether or not the customer is considering introducing another company's tool (other company's product), a question inquiring about points of comparison between the other company's tool and the company's own tool, and the like. The approver 28c includes a question or the like for inquiring about the approver who has the authority to decide the business negotiations of this conference. In addition, on the confirmation item screen 28, questions regarding the budget, introduction timing, or next action can be displayed as questions regarding classification in the meeting.

Highlighting 29 may display utterances 31a that are determined to include important matters in the meeting, for example, the customer's response to need 28a.

When actually holding a meeting, the user and the customer proceed to talk while the above displays are made on the meeting screen 23 . When the conference ends, the server 2 obtains the speech information including the voice of the conference obtained by the user terminal 4 by the obtaining means 11 and stores it in the storage means within the server 2 .

Next, the analysis means 12 extracts a plurality of utterances of the user or the customer from the voice data of the utterance information. The utterance extraction process is performed using a known method such as a method of separating which speaker made which utterance based on the frequency of the voice of the speaker, or a method of separating with silent intervals as delimiters.

In addition, the analysis means 12 classifies which category the plurality of utterances belong to in the content of the conference. Classifications include budget, authority, necessity, time of introduction, competition, and next action as described above. This classification is performed by judging from the voice of each utterance whether or not words (keywords) peculiar to each classification are included. For example, if the utterance includes the expression "○○ yen", it is determined that the utterance including that phrase is the utterance about the budget.

In addition, the analysis unit 12 identifies at what stage of the progress of the conference the plurality of utterances were uttered. The stages include the introduction stage, explanation stage, question-and-answer stage, and summary stage described above. Identification at this stage is also done using speaker information, such as whose voice each utterance is.

For example, when specific expressions such as numerical values are used repeatedly in a meeting, it is assumed that the discussion is about specific products, so it is identified as being in the explanation stage or question-and-answer stage. Also, at that time, when the user continues to speak, it is identified as being in the explanation stage. Also, when the user and the customer are having conversations in the middle of the meeting, it is identified as the question-and-answer stage.

Next, the analysis means 12 extracts a summary utterance 32 from among the multiple utterances. The summary utterances 32 are utterances that include items related to the purpose or results of the meeting, and play an important role in understanding the main points of the meeting.

In the utterance information documentation device 1 of the present embodiment, the abstract utterance 32 is extracted as follows. First, when wordings included in a plurality of utterances are classified into any one of budget, authority, etc., classification points are given to each wording. Also, for a plurality of utterances, stage points are given to each of the stage switching, for example, the last utterance in the explanation stage when switching from the explanation stage to the question and answer stage and the first utterance in the question and answer stage.

The analysis means 12 assigns points to each utterance as described above, and extracts a predetermined number of utterances with the highest total points from among the aggregated utterances. The number of extractions can be appropriately selected according to the purpose of the abstract, and one or more can be extracted. In general, in a meeting of about one hour, the number of extractions can be set to about 10 to 30 in order to know the important matters of the meeting. On the other hand, it is also possible to narrow down the number of extractions to one in order to know the important matters of the meeting.

In addition, in the utterance information documentation device 1 of the present embodiment, when giving points to each utterance, language points may be given to linguistic information including numerical values and dates in each utterance. If a specific numerical value is included in an utterance, it is highly likely that the utterance is an important utterance. The same applies to dates. This linguistic information includes not only numerical values and dates, but also important matters in the meeting, such as product names and company names.

Further, in the utterance information documentation device 1 of the present embodiment, when giving points to each utterance, in each utterance, speaker points may be provided for giving points according to the speaker. Users and customers are the main speakers, but there are cases where the user's boss, the customer's boss, or the like participates in the conference.

The classification points, grade points, language points, and speaker points can each be arbitrary points, and for example, the classification points and grade points are 1 point. Speaker points are weighted to each point, such as 0 points for user's utterance, 2 points for customer's utterance, 1 point for user's boss, and 3 points for customer's boss. may be given. Similarly, language points may be given such that one point is given to numerical values and dates, and two points are given to proper nouns such as product names.

In this way, when a predetermined number of summary utterances 32 of high importance are extracted in the meeting, the summary section 13 creates a summary sentence 35 . First, it is determined whether or not the summary utterance 32 with a high score is related to the utterances before and after it.

For example, as shown in FIG. 4A, if the speaker of the extracted summary utterance 32 and the immediately preceding utterance 33 are the same person, the conjunctions included in the sentences are analyzed, and the adjacent utterances are extracted from the conjunctions . If the utterances are determined to be consecutive utterances, the utterances may be determined to be related. The same applies to the summary utterance 32 and the utterance 34 immediately following it .

When the speaker is the same person, the judgment is made using the conjunction . This is because, if the utterance is "but" or the like, it is expected that the utterance has a different content from the preceding and following utterances.

On the other hand, as shown in FIG. 4B, if the summary utterance 32 and the utterance 33 immediately preceding it are different people, the similarity between these utterances is determined, and if the similarity is equal to or greater than a predetermined threshold, Those utterances can be relevant if there is The same applies to the summary utterance 32 and the utterance 34 immediately following it. For this similarity determination, for example, when each utterance includes a phrase with a high degree of common vocabulary linguistically, it can be determined that the degree of similarity is high.

In this way, when the speakers are different persons, it is unlikely that a conjunction will intervene, so it is determined whether or not the content is similar. In this case, it is sufficient to determine whether or not there is similarity with an arbitrary threshold of the degree of commonality, using known judgment criteria such as the degree of commonality of vocabulary in linguistics.

In this embodiment, the summarizing means 13 extracts an utterance containing important matters as an utterance for summarization, extracts utterances related to the utterances before and after the utterance for summarization as utterances for addition, and extracts these utterances for summary and the utterance for addition. A summary sentence is created using the utterance.

As shown in FIG. 4C, the summary sentence 35 is created by converting the extracted utterances 32 and 33 including the utterances before and after the summary utterance 32 into short sentences. At this time, the sentences in the utterance may be converted from spoken words to written words, and redundant sentences (dotted lines in FIG. 4(C)) in the spoken words may be deleted. Alternatively, this conversion may be performed using a deep-learned model.

Moreover, when converting the extracted utterances 32 and 33 into short sentences, so-called double translation may be performed in which the utterances 32 and 33 are translated into English and the English-translated utterances are further translated into Japanese. The utterances 32 and 33 are composed of spoken words, and Japanese tends to have many ambiguous expressions. When translating a Japanese utterance into English, the translation engine converts the utterance into simple English in consideration of ambiguous expressions in Japanese, so the utterance after English translation is often short. Therefore, by further translating this shortened English sentence into Japanese, it is possible to convert it into a simple short sentence in which ambiguous expressions are deleted.

The summary sentence 35 created in this way is stored in the server 2 as a sentence containing important matters of the meeting in association with the meeting, and can be viewed on the list management screen 41 shown in FIG. In FIG. 5, when the video icon 43 is enabled in the status column 42, the video of the meeting objection can be browsed. In addition, in the status column 42, when the document icon 44 is valid, the abstract 35 can be browsed.

By enabling this list management screen 41 to be browsed within the company to which the user belongs, it is possible to share important information regarding business negotiations within the company. For example, if the person in charge of the meeting is an excellent salesperson, the analysis means 12 extracts important utterances from the excellent salesperson's utterances, and the summary means 13 extracts the important utterances as easy-to-read sentences. Therefore, an inexperienced salesperson can obtain a high learning effect by browsing the summary sentence 35 .

In the above-described embodiment, the analysis means 12 classifies the utterances and specifies the stage. Extraction may be performed.

In addition, in the above-described embodiment, classification was explained by taking budget, authority, necessity, time of introduction, competition, and next action as examples, but not limited to these, other items such as product performance, service form, etc. can be classified. Similarly, although the stages are described as an introduction stage, an explanation stage, a question-and-answer stage, and a summary stage, other stages such as a customer request listening stage may be provided.

In the above-described embodiment, as the stage points, points are given based on the switching of stages as a guide, but this is not limiting, and points are set in advance for each stage, and the utterance determined to be at that stage , the points may be given.

Further, in the above-described embodiment, the utterance is explained by taking a meeting as an example, but it is not limited to this, and can be applied to lectures and presentations in which one or more speakers speak to an audience. . In this case, as for the classification, if it is a lecture meeting, the contents of the lecture may be classified and translated. As for the stages, similar to the above embodiment, the introduction stage, the explanation stage, and the summary stage may be used.

1 Speech information documentation device 2 Server 3 Network 4 User terminal 5 Customer terminal 11 Acquisition means 12 Analysis means 13 Summary means 14 Display means 21 Preparation screen 22 Second preparation screen 23 Conference Screen 24... Information display column 25... Face display screen 26... Conversation sentence display screen 27... Purpose screen 28... Confirmation item screen 29... Highlight display 31... Utterance 32... Utterance for summary 35... Summary sentence 41... List management screen

Claims (7)

A device for documenting one or more utterances,
an acquisition means for acquiring utterance information;
analysis means for analyzing the speech information;
summarizing means for creating a summary sentence of the utterance information based on the analysis result of the analyzing means;
Display means for displaying at least the summary,
The analysis means is
Identifying which category the utterance belongs to among a plurality of preset categories, and identifying which stage the utterance belongs to among a plurality of stages of progress,
Extracting a summary utterance from among the plurality of utterances using at least one information of the classification or the stage;
The summarizing means creates a summary sentence including the summary utterance, and enables the display means to display the summary sentence,
The categories include budget, authority, necessity, time of introduction, competition, and next action;
A speech information documentation apparatus , wherein the steps include an introduction step, an explanation step, a question-and-answer step, and a summary step . A device for documenting one or more utterances,
an acquisition means for acquiring utterance information;
analysis means for analyzing the speech information;
summarizing means for creating a summary sentence of the utterance information based on the analysis result of the analyzing means;
Display means for displaying at least the summary,
The analysis means is
Identifying which category the utterance belongs to among a plurality of preset categories, or identifying which stage the utterance belongs to among a plurality of stages of progress,
Extracting a summary utterance from among the plurality of utterances using at least one information of the classification or the stage;
The summarizing means creates a summary sentence including the summary utterance, and enables the display means to display the summary sentence,
An utterance information documenting apparatus, wherein the summarizing means translates the utterance for summary into English, translates the English translated utterance into Japanese, and creates the summary sentence using the Japanese-translated utterance. A device for documenting one or more utterances,
an acquisition means for acquiring utterance information;
analysis means for analyzing the speech information;
summarizing means for creating a summary sentence of the utterance information based on the analysis result of the analyzing means;
Display means for displaying at least the summary,
The analysis means is
Identifying which category the utterance belongs to among a plurality of preset categories, or identifying which stage the utterance belongs to among a plurality of stages of progress,
Extracting a summary utterance from among the plurality of utterances using at least one information of the classification or the stage;
The summarizing means creates a summary sentence including the summary utterance, and enables the display means to display the summary sentence,
The analysis means recognizes, in the utterance, linguistic information including information about numerical values or dates, and speaker information that is information about a speaker,
giving points according to the linguistic information and the speaker information in the utterance, and totaling the points for each utterance;
An utterance information documentation device, characterized by extracting one or more of the utterances for summarization having a high total point. The speech information documentation device according to claim 3 ,
The summarizing means determines whether or not at least one or both of an utterance immediately before the utterance for summary and an utterance immediately after the utterance for summary are related to the utterance for summary, and the utterance determined to be related is taken as an additional utterance. ,
The analysis means stores a plurality of utterances made by a plurality of speakers from the utterance information in association with the utterances and the speakers;
When the summary utterance and the utterance immediately before or the utterance immediately after are utterances by the same person, the presence or absence of the relationship is determined from the conjunctions included in the adjacent utterances, and the relevance of the adjacent utterances. An utterance information documentation device characterized by determining A device for documenting one or more utterances,
an acquisition means for acquiring utterance information;
analysis means for analyzing the speech information;
summarizing means for creating a summary sentence of the utterance information based on the analysis result of the analyzing means;
Display means for displaying at least the summary,
The analysis means is
Identifying which category the utterance belongs to among a plurality of preset categories, or identifying which stage the utterance belongs to among a plurality of stages of progress,
Extracting a summary utterance from among the plurality of utterances using at least one information of the classification or the stage;
The summarizing means creates a summary sentence including the summary utterance, and enables the display means to display the summary sentence,
giving points according to the classification or stage of the utterance and totaling the points for each utterance;
extracting one or more of the summarizing utterances with a high total point;
Determining whether or not at least one or both of the utterance immediately before the summary utterance and the utterance immediately after the summary utterance are related to the summary utterance, and the utterance determined to be related is used as an additional utterance,
The analysis means stores a plurality of utterances made by a plurality of speakers from the utterance information in association with the utterances and the speakers;
When the summary utterance and the utterance immediately before or the utterance immediately after are utterances by the same person, the presence or absence of the relationship is determined from the conjunctions included in the adjacent utterances, and the relevance of the adjacent utterances. An utterance information documentation device characterized by determining A device for documenting one or more utterances,
an acquisition means for acquiring utterance information;
analysis means for analyzing the speech information;
summarizing means for creating a summary sentence of the utterance information based on the analysis result of the analyzing means;
Display means for displaying at least the summary,
The analysis means is
Identifying which category the utterance belongs to among a plurality of preset categories, or identifying which stage the utterance belongs to among a plurality of stages of progress,
Extracting a summary utterance from among the plurality of utterances using at least one information of the classification or the stage;
The summarizing means creates a summary sentence including the summary utterance, and enables the display means to display the summary sentence,
The analysis means gives points according to the classification or the stage of the utterance and aggregates the points for each utterance;
extracting one or more of the summarizing utterances with a high total point;
The summarizing means determines whether or not at least one or both of an utterance immediately before the utterance for summary and an utterance immediately after the utterance for summary are related to the utterance for summary, and the utterance determined to be related is taken as an additional utterance. ,
The analysis means stores a plurality of utterances made by a plurality of speakers from the utterance information in association with the utterances and the speakers;
When the summary utterance and the utterance immediately before or the utterance immediately after are utterances by different people, the presence or absence of the relationship is determined by the similarity of the wording of the adjacent utterances. An utterance information documentation device characterized by judging gender. A device for documenting one or more utterances,
an acquisition means for acquiring utterance information;
analysis means for analyzing the speech information;
summarizing means for creating a summary sentence of the utterance information based on the analysis result of the analyzing means;
Display means for displaying at least the summary,
The analysis means is
Identifying which category the utterance belongs to among a plurality of preset categories, or identifying which stage the utterance belongs to among a plurality of stages of progress,
Extracting a summary utterance from among the plurality of utterances using at least one information of the classification or the stage;
The summarizing means creates a summary sentence including the summary utterance, and enables the display means to display the summary sentence,
recognizing linguistic information including information about numerical values or dates and speaker information that is information about a speaker in the utterance;
giving points according to the linguistic information or the speaker information in the utterance and totaling the points for each utterance;
extracting one or more of the summarizing utterances with a high total point;
The summarizing means determines whether or not at least one or both of an utterance immediately before the utterance for summary and an utterance immediately after the utterance for summary are related to the utterance for summary, and the utterance determined to be related is taken as an additional utterance. ,
The analysis means stores a plurality of utterances made by a plurality of speakers from the utterance information in association with the utterances and the speakers;
When the summary utterance and the utterance immediately before or the utterance immediately after are utterances by different people, the presence or absence of the relationship is determined by the similarity of the wording of the adjacent utterances. An utterance information documentation device characterized by judging gender.

Images