JP2000259389A - Interaction recording system and interaction recording synthesizer - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an interaction recording system and an interaction recording synthesizer where a participant of an interaction can control whether or not an interaction recording is handed down after interaction and which can hand down the interaction recording at relatively lower cost. SOLUTION: In this interaction recording system, each interaction participant carries a user unit 12 with himself, inputs each speech contents from its voice input part 42, and stores them as narration data on his own. Then, after the interaction finishes, the narration data stored in the user unit 12 of all the interaction participants are received from a beam reception/emission part 18, an interaction data generation part 36 generates interaction data on the basis of the former data, and they are stored in an interaction data storage part 38.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a dialog recorder and a dialog recorder / synthesizer, and more particularly to a technique for recording a dialog by a plurality of dialog participants.

[0002]

2. Description of the Related Art In many cases, new knowledge can be obtained by recording conversations (conversations, conferences) by a plurality of people and analyzing the records. For example, a dialogue with a customer can later be scrutinized (especially by showing the record to a manager or top sales representative and asking for opinions), You can come up with better solutions. Alternatively, it is widely known that a protocol analysis technique of analyzing a human utterance is effective for checking the usability of the system.

[0003] However, the dialogue does not always take place only in an environment equipped with recording facilities. For example, a conversation that happened accidentally in an open space, such as standing in a corridor,
They may give you important hints and ideas for doing your job. In addition, even when visiting a customer, a place where records can be easily taken, such as an interview space, cannot always be used.

[0004] As a conventional technique for recording a dialogue in such a place, for example, as shown in Japanese Patent Application Laid-Open No. Hei 10-134407, a portable device such as a PHS sends voice information to a voice recognition server through a network, and There is a technique in which a recognition result by a server is sent back to the portable device.

[0005]

However, the prior art has the following two problems.

First, feeling that a conversation is being recorded affects human communication. In the prior art, the person recording the conversation (either the organizer or the participant of the conversation) prepares the recording before the conversation begins,
Record all utterances of the participant in the conversation. The participant (remaining participant) may be asked to agree to record in advance. Once the consent has been reached, the person receiving the dialogue has to be aware during the dialogue that his / her remark has been recorded by another person and that the recording will be used in some form later. This suppresses the participants' remarks, and conversely, the participants are likely to take an excessively aggressive or compliant attitude, which hinders free and vigorous communication. Also, without good consent, even if good ideas and opinions were formed in the dialogue, they were not recorded,
The records will have to be reconstructed, relying on the participants' memories.

The second concerns the accuracy and cost of dialog recording. In the related art disclosed in Japanese Patent Application Laid-Open No. H10-134407, the mobile device and the network must always be connected in a good state. However, the dialogue location is not always a place that can be connected to the network, and even if it can be connected, the connection to the network may be unstable or lost due to access congestion or radio wave conditions. . Also, even in a connected situation, one voice recognition server may convert voice information of a plurality of persons sent from individual terminals between terminals and temporally the level of the volume and sound quality may vary greatly. It is very difficult to recognize speech correctly. When a packet communication network is used, the correct utterance order may not be transmitted to the server due to the delay of the network. Further, the communication cost due to continuing to use the network becomes large.

On the other hand, as an application of the conventional technique disclosed in Japanese Patent Application Laid-Open No. Hei 10-134407, it is conceivable to construct and use a small-scale wireless network or a wired network at a place of conversation. However, in the case of wireless communication, the same problems as in the above-described example remain regarding connection stability and speech recognition accuracy. Also, in the case of a wired connection, it cannot be guaranteed whether or not a line can be laid at a conversation place. In any case, it goes without saying that extra costs are required for network construction and withdrawal.

The present invention has been made in view of the above problems, and has as its object to enable a participant of a dialogue to control whether or not to keep a dialogue record after the dialogue, and to provide a dialogue of desired quality at a relatively low cost. It is an object of the present invention to provide a dialog recording system and a dialog recording synthesizing device capable of recording.

[0010]

In order to solve the above-mentioned problems, a dialog recording system according to the present invention comprises a statement information recording means for recording statement information for each dialog participant, and a statement information recorded on the statement information recording means. Utterance time information generating means for generating utterance time information for each utterance information, and dialogue for generating dialogue information between dialog participants from the utterance information recorded by the utterance information recording means based on the utterance time information It is characterized by including information generating means, and dialog information recording means for recording the dialog information.

On the other hand, a dialogue recording / synthesizing apparatus according to the present invention comprises:
It is characterized in that, based on the utterance information recorded for each dialogue participant, dialogue information performed between the dialogue participants is generated based on the utterance time information for each utterance information, and the dialogue information is recorded.

Further, the computer-readable medium according to the present invention generates dialog information between dialog participants based on utterance time information for each utterance information from utterance information recorded for each dialog participant. And a program for operating a computer to record the dialog information.

In the present invention, utterance information is recorded for each dialog participant, and dialog information is generated therefrom based on occurrence time information generated for each utterance information. In this way, the participants of the dialog can control whether or not to keep a record of the dialog after the dialog. Further, since the statement information is recorded for each dialog participant, a record of desired quality can be left at a lower cost than recording the statements of all dialog participants via a network or the like.

[0014] In one aspect of the present invention, the dialog information generation means includes a first keyword storage means for storing a predetermined first keyword, and a second keyword storage for storing a second keyword corresponding to the first keyword. Means for generating the dialogue information such that the utterance time of the first keyword of one dialogue participant is close to the utterance time of the second keyword of another dialogue participant.
In this case, the first keyword may indicate any one of a greeting, a call, a question, a question, and a confirmation, and the second keyword may indicate a response.

If the [0015] This, greetings and called for a response such as "Hello", "~ Is" questions or doubts and its response, such as, the respective "would be - or" confirmation of such and its response, and the first By storing the one keyword and the second keyword in the first and second keyword storage means, it is possible to easily generate more accurate conversation information based on the fact that they are exchanged in temporal proximity. Become like

Further, in one aspect of the present invention, the dialogue information generating means includes third keyword storage means for storing a predetermined third keyword, and the other dialogue participant operates during a utterance time period of one dialogue participant. The dialog information is generated in consideration of the fact that a third keyword can be uttered. In this case, the third keyword may be one of a combination, an emotional expression, and an interrupt expression.

[0017] In this way, by storing in the third keyword storage means the impressive expression or the interrupt expression such as "yes", "oh", "slightly", etc. It is possible to easily generate more accurate conversation information in consideration of the fact that it can be repeated during a comment.

In one aspect of the present invention, the dialog information generating means converts the utterance information into character string information for each dialog participant, and generates the dialog information from the character string information. . This makes it possible to relatively easily judge the content of remarks of each dialog participant, so that more accurate dialog information can be easily generated. In particular, by storing the first to third keywords in each storage unit as a character string, it is possible to easily generate dialog information using the first to third keywords.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a diagram showing a configuration of a dialog recording system according to an embodiment of the present invention. As shown in the figure,
The dialog recording system 10 includes a plurality of user units 1.
2a to 12c. In the present embodiment,
One user unit 12 is uniquely given to one user (dialogue participant). Each of the user units 12a to 12c uses a beam communication function described later,
Data can be transmitted / received to / from another user unit 12. The storage 16 having a beam communication function
Or the data can be retrieved from the storage 16. Further, data can be sent to the printer 14 having the beam communication function, and the data can be printed.

Hereinafter, the respective functional units constituting the user unit 12 will be described.

(1) Date / Time Data Generation Unit The date / time data generation unit 44 has a function of passing date / time data at that time to another function unit. The date and time data is 199
It is composed of year, month, day, hour, minute, second, and millisecond as in 8.12.6.17.25.13.335. The function unit has a built-in clock, and the time is adjusted using a command input unit described later.

(2) Identification Number Storage Unit The identification number storage unit 22 has a function of storing an identification number uniquely given to the user unit 12. The identification number stored in the identification number storage unit 22 is used when the communication control unit 20 described later communicates with another user unit 12.

(3) Beam Communication Control Unit The beam communication control unit 20 cooperates with the beam communication control unit 20 of the other user unit 12 to operate the user unit 1
It has a function to realize beam communication between the two. The start of beam communication follows an instruction using a command input unit described later.

(4) Beam receiving / emitting unit The beam receiving / emitting unit 18 has a function of receiving or emitting a beam according to the control of the beam communication control unit. In this embodiment, in order to prevent communication contents from reaching another person, a range where a beam reaches (distance and relative angle between light receiving and emitting units)
Is smaller.

(5) UI Control Unit The UI control unit 26 has a function of interacting with a user using a command input unit 24 and an information display unit 28 described later in cooperation with other function units described later. .

(6) Information Display Unit The information display unit 28 has a function for displaying information to the user.

(7) Command Input Unit The command input unit 24 has a function for a user to input a command. Specifically, the following five buttons are prepared.

"Menu" button: Used by the user to switch between displaying and hiding the operation menu. The operation menu items include a) registration of personal information, b) registration of voice characteristics, c) data transmission, d) data reception, e) data deletion, f) comment on data, g) microphone volume adjustment, h)
There are seven top-level menus for setting the time, and a submenu for determining details for each item.

"+" Button: Used to change the value in the positive direction, such as selecting the next option in the list or increasing the numerical value.

"-" Button: Used to change the value in the negative direction, such as selecting the previous option in the list or decreasing the numerical value.

"O" button: Used for various decisions, such as selecting an option or answering a confirmation message with "Yes".

"X" button: Used for various cancellations, such as returning from a submenu to a higher-level menu or answering a confirmation message with "No".

When transmitting and receiving data between the user units 12, first, the user unit 12 on the receiving side selects "data reception" and waits for reception. Select "Send" to send the data. When transmitting data, in order to identify the sender, the identification number of the user unit 12 on the transmitting side and the name of the transmitting user stored in the user personal information management unit 32 described below are transmitted. Sent together with the barrel data.

(8) Voice Input Unit The voice input unit 42 includes a microphone and has a function for inputting a user's voice. The microphone for voice input has strong directivity so as not to pick up sounds other than the voice of the user. The user can use the microphone volume adjustment in the operation menu to adjust the microphone mounting position so that only the user's voice can be recognized.

(9) Voice Recognition Unit The voice recognition unit 40 has a function of converting the user's voice input by the voice input unit 42 into text data.
In order to improve the accuracy of the recognition, the voice recognition unit 40 can have the user utter some examples in advance and register the characteristics of the voice unique to the user, such as frequency distribution and volume change. This registration is performed by selecting “Registration of voice feature” from the operation menu.

(10) User Personal Information Management Unit The user personal information management unit 32 has a function of managing user personal information such as the user's name, affiliation, and contact information. Data managed by the user personal information management unit 32 is used for confirmation during communication, generation of conversation data, and the like.

The data managed here is input using the voice input unit 42 and the voice recognition unit 40. Specifically, first, "Register personal information" is selected from the operation menu, and individual items such as "Name" are selected from a submenu displayed thereafter. Next, the user speaks data to be input to the voice input unit 42. The voice input from the voice input unit 42 is converted into text data by the voice recognition unit 40, and is converted into text data.
Displays the text data. If the user is correct, press the “○” button to confirm, if incorrect, “X”
I press the button. When the “○” button is pressed, the text data is stored in the user personal information management unit 32 as the value of the selected item. When the "x" button is pressed, a submenu for correcting the recognition result is called.

(11) Narration Data Generation Unit The narration data generation unit 34 has a function of generating narration data. The narration data is an example of utterance information and utterance time information according to the present invention, and is permutation data including a set of utterance order, time, and content.

FIG. 2 shows an example of the narration data.
FIG. 4 shows a narration generated by each user unit 12 when the dialog shown in FIG. 3 is performed between two users (here, users A and B) of the dialog recording system 10. Shows the data. FIG. 2A shows the narration data of the user A, and FIG. 2B shows the narration data of the user B.

That is, utterances that are temporally continuous are converted by the voice recognition unit 40, and the converted text data or a null value indicating silence is stored in the right end column of each row in FIG. In the center column of each line, the time required for the utterance (or silence) of the same line is stored (in the present embodiment, the unit is seconds and the effective digit is 1/100 second). The time is measured using the voice input by the voice input unit 42 as a measurement target, and is measured using time data generated by the date and time data generating unit 44. Also, the order of the utterance (or unvoiced) stored in the right end column of the same line is stored in the left end column of each line.

The determination of whether or not utterances are temporally continuous (determination of whether or not data should be stored in the next line as another utterance), which is performed when generating the narration data shown in FIG. This is performed depending on whether or not the time during which the voice is interrupted exceeds a certain time (for example, 0.2 seconds). In addition, a certain period of time from a certain utterance (150 in this embodiment)
(Second) If the user is unvoiced, the previous utterance is stored as one narration data in a narration data storage unit 30 described later.

As will be described later, "time" recorded in the center column of the narration data is used when generating dialogue data by combining the statements of each user, and various modifications can be made. . For example, the speech start time or the speech end time can be stored for each speech.

(12) Narration Data Storage Unit The narration data storage unit 30 has a function of storing the narration data generated by the narration data generation unit 34 together with the date and time data and the comment at the time of storage. The date and time data and the comment are displayed on the information display unit 28 when operating on the narration data stored in the narration data storage unit 30,
Helps the user to recall the contents of the narration data. The comment is input as text data via the voice recognition unit 40 when the user selects “comment to data” in the operation menu and speaks the comment.

The data stored in the narration data storage unit 30 is used for generating interactive data described later.
Further, in response to the “data transmission” instruction of the operation menu, the data stored in the narration data storage unit 30 is stored in another user unit 12 or storage 16.
And the printer 14. Further, the data stored in the narration data storage unit 30 is deleted in response to the “data deletion” instruction in the operation menu.

(13) Dialogue Data Generator The dialogue data generator 36 has a function of generating dialogue data. As shown in an example in FIG. 4, the conversation data is obtained by combining the contents of narration data of different users (all conversation participants) according to their utterance order.

The principle algorithm for generating interactive data is:
In a conversation, when one person is talking, the other person is often silent, so we find and combine places where one person's speech part fits in the other person's silent part in time. . For example, the utterance time zone of the user A is shown in FIG.
As shown in (a), it is a1, a2, a3, and the user B
In the case where the utterance time period of b is b1 and b2 as shown in FIG. 3B and the utterance time period of user C is c1, c2 and c3 as shown in FIG. As shown in (d), the relative position of each user's speaking time zone is determined. In other words, the conversation data generation unit 36 relatively moves the uttering time period of each user forward and backward so that a state in which another user is uttering while one user is uttering is not generated as much as possible. . Specifically, in the narration data of different users, the time value of a set whose content is text (character string information) is compared with the time value of a set whose content is a null value to find out a combination method. .
After that, all the utterances of the user are extracted in chronological order, and conversation data is generated based on the extracted utterances. FIG. 6 is a diagram illustrating the dialog data generated by the dialog data generating unit 36 when each user speaks during the speaking time period illustrated in FIG. 5 and a dialog is performed.

In a dialogue, there is a case where one person speaks and another person speaks. In this case, it may be a combination that encourages the other party's talk (such as “Eh”) or an emotional expression (“Oh”
There is an interrupt expression ("uh,""slightly,""oh," etc.) that indicates what you want to talk about. In finding the combination, ignoring such utterances makes it easier to find the combination. This is an exception to the principle algorithm described above. In the present embodiment, it is assumed that the content of such an utterance is a null value, and a combination method is found using data obtained by adding time to the null value before and after the utterance (this is called “degenerate data”). FIG. 7 shows, as an example, degenerated data of the narration data of FIG.

Further, when finding the combination, the combination may not be uniquely determined only by the time of the utterance and the time of the unvoiced portion. For example, in the data shown in FIG. There are two combinations that can be entered. In this case, from the content of the utterance, 1) greeting / call to the other party ("Good morning" or "**" etc.) and the response, and 2) Questions / questions to the other party ("Are you good?" 3) Check with the other party (eg, "Is this OK?""Howabout?") And find the response. Specifically, the conversation data generation unit 36 sets a pair of keywords that are considered to be uttered in close proximity to each other as a first keyword and a second keyword, respectively, and stores them in a memory in advance. I have.

[0050] in a degraded data shown in FIG. 7, in each of the two narration data there is a greeting of "Hello", they should have close temporal proximity. Therefore, a combination in which the unvoiced part of the order 4 of the user A includes the utterance part of the order 1 of the user B is more appropriate as the combination method. As described above, a combination method can be found by using the utterance time and the three content characteristics.

FIG. 8 shows the flow of processing in the interactive data generation unit 36. The process of the interactive data generation unit 36 is started when data reception is selected from the operation menu and narration data of another user is received from another user unit (S101). In S102, a list for selecting narration data is displayed on the information display unit 28, and prompts the user to make a selection. After the selection is made by the user, the received narration data and the degenerated data of the selected narration data are generated (S10).
3) Based on this, the way of combining the narration data is determined (S104). Then, in accordance with the determination, the contents of the narration data are transcribed to generate interactive data (S105). Here, as for the contents uttered at the same time, the contents uttered earlier in time are arranged before in the conversation data, and the contents uttered later are arranged after the utterance. At the time of this generation, a user name is added to the conversation data according to the content of the utterance so that the utterance can be understood. Next, if there is a portion where the same person continuously speaks in the conversation data, these are combined into one (S10).
6, FIG. 9). Finally, the dialog data is stored in a dialog data storage unit 38 described below (S10).
7).

(14) Dialogue Data Storage Unit The dialogue data storage unit 38 has a function of storing the dialogue data generated by the dialogue data generator 36 together with the date and time data and comments at the time of storage. The functions provided by the conversation data storage unit 38 are the same as those of the narration data storage unit 30.

According to the dialog recording system 10 described above, the utterance data is individually recorded in the user unit 12 carried by each user, and the user unit 1 is later recorded.
When the utterance data is exchanged between the two, the at least one user unit 12 can automatically combine the utterance data to generate conversation data.

As a result, the utterance of the individual can be primarily recorded only in the user unit 12 owned by the person, and only when the person agrees, the utterance data is passed to another person and the conversation data is transmitted. Can be generated. Therefore, a record of the dialogue is generated only when the participants agree upon the dialogue after the dialogue, and the record disclosure control that the dialogue record does not exist anywhere otherwise is possible.

In the dialog recording system 10, there is no need for each user unit 12 to transmit / receive data or synchronize during a user's dialog. Therefore, even in places where network connection is not possible, in places with poor acoustic or radio wave conditions, or where it is physically difficult to set microphones, it is possible to record conversations much more easily than in the conventional technology, and to achieve the same quality. The cost of preparing and operating or operating the equipment required to keep a record, and the cost of editing the dialogue can be greatly reduced.

The above-described dialog recording system 10 can be variously modified.

For example, in the dialog recording system 10 described above, communication with the other user units 12 is performed in plain text by using a beam. However, this communication may be performed by encrypting a file and communicating by radio waves. By using radio waves, restrictions on the position and orientation of the user unit 12 are greatly eased as compared with beams. Further, even if another person within the range of the radio wave receives the file due to the encryption, the other person cannot develop the encrypted file, so that the other person can be prevented from viewing the conversation record.

In the above dialogue recording system 10,
Although the content of each user's remark is once converted into text data and converted into the form of narration data, the process may proceed with the original voice information. In the above description, the dialog data is generated in a text format. However, the dialog data may be configured by voice information. In this case, for example, the combination of the conversations may be performed based on the narration data in the text format, and the voice of each user may be edited according to the combination to generate the conversation data.

Further, in the above dialogue recording system 10, only the contents of remarks among the dialogues actually performed between the users are recorded in the dialogue record. An image (for example, a moving image) such as a facial expression of each user may be associated with the data.

The dialog recording system 10 described above
Can use a portable dedicated device occupied by each user, such as an electronic organizer or a smartphone. In addition, software for realizing the function of the dialog recording system 10 is installed in a portable general-purpose computing device used by each user such as a mini-notebook PC or a wearable PC, or hardware for realizing the function is installed. You may.

Further, a form of a device including an integrated device such as an electronic organizer or a mini-notebook and a plurality of units prepared according to functions may be adopted. Examples of this mode include a headset type unit (having a voice input function), a wristwatch type unit (having a button input, a screen display, and a beam communication function), and a belt-mounted type unit (computation related to the present invention. (Having a memory function) in a wired or wireless manner. As another example, an eyeglass-type unit (having voice input, screen display, and communication function using a beam), a pad-type unit (having input by a button), and a holster-mounted unit (calculation and storage related to the present invention) (Having a function).

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a dialog recording system according to an embodiment of the present invention.

FIG. 2 is a diagram showing an example of narration data.

FIG. 3 is a diagram illustrating an example of a dialogue between a user A and a user B;

FIG. 4 is a diagram illustrating an example of conversation data.

FIG. 5 is a diagram illustrating a procedure for combining a plurality of narration data.

FIG. 6 is a diagram showing dialog data generated by the procedure shown in FIG.

FIG. 7 is a diagram illustrating degenerated data.

FIG. 8 is a flowchart illustrating processing of a conversation data generation unit.

FIG. 9 is a diagram showing dialog data in which remarks regarding the same person are put together in one line.

[Explanation of symbols]

10 dialogue recording system, 12 user unit, 1
4 printer, 16 storage, 18 beam receiving / emitting unit, 20 beam communication control unit, 22 identification number storage unit,
24 command input section, 26 UI control section, 28 information display section, 30 narration data storage section, 32 user personal information management section, 34 narrator generation section, 36
A dialogue data generator, 38 a dialogue data storage, 40 a voice recognizer, 42 a voice input unit, 44 a date and time data generator;

Claims (8)

[Claims] 1. utterance information recording means for recording utterance information for each dialogue participant; utterance time information generation means for generating utterance time information for each utterance information recorded in the utterance information recording means; Dialog information generating means for generating dialog information between dialog participants based on the utterance information recorded by the utterance information recording means, based on the information, and dialog information recording means for recording the dialog information. A featured dialogue recording system. 2. The dialogue recording system according to claim 1, wherein the dialogue information generation unit stores a first keyword storage unit that stores a predetermined first keyword, and stores a second keyword that corresponds to the first keyword. Second keyword storage means, wherein the dialog information is generated such that the utterance time of the first keyword of a certain dialog participant and the utterance time of the second keyword of another dialog participant are close to each other. And dialogue recording system. 3. The dialogue recording system according to claim 2, wherein the first keyword represents one of a greeting, a call, a question, a question, and a confirmation, and the second keyword represents a response. Dialogue recording system. 4. The dialogue recording system according to claim 1, wherein said dialogue information generating means includes third keyword storage means for storing a predetermined third keyword, and a utterance time of a dialogue participant. A dialog recording system, wherein the dialog information is generated considering that the third keyword may be uttered by another dialog participant in a band. 5. The dialogue recording system according to claim 4, wherein said third keyword represents one of a joint expression, an emotional expression, and an interrupt expression. 6. The dialogue recording system according to claim 1, wherein said dialogue information generating means converts utterance information into character string information for each dialogue participant, and converts the dialogue information from the character string information. A dialogue recording system for generating information. 7. A method for generating dialog information between dialog participants based on utterance time information for each utterance information from the utterance information recorded for each dialog participant, and recording the dialog information. Dialogue recording synthesizer. 8. A computer which generates, based on utterance information recorded for each dialogue participant, dialogue information between dialogue participants based on utterance time information for each utterance information, and records the dialogue information. A medium on which a program for operation is recorded.