JP2016071050A - Voice interactive device, voice interactive system, terminal, voice interactive method, program for letting computer function as voice interactive device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a voice interactive device that authenticates a user without using a radio tag nor a camera, and replies a topic matching the user according to utterance.SOLUTION: A voice interactive system 10 comprises: a voice recognition module 110 for recognizing voice data based upon utterance; a voice authentication module 120 for determining a speaker based upon utterance; a topic estimation module 130 for estimating a topic corresponding to the determined speaker; and an interaction generation module 140 for generating details of utterance to the user based upon a result (for example, topics of sports that the user's children are interested in, topics of fashion that females are interested in, etc.) of the topic estimation module 130.SELECTED DRAWING: Figure 1

Description

  The present disclosure relates to voice recognition and voice authentication, and more specifically, to a technique for performing voice recognition and voice authentication at the same timing.

  Devices using voice recognition technology are known. For example, Japanese Patent Laying-Open No. 2011-000681 (Patent Document 1) discloses a “communication robot that can execute various communication behaviors according to intimacy with a communication target” (see [Summary]). .

JP 2011-000681 A

According to the technology disclosed in Patent Document 1, user authentication of a communication robot is performed using a wireless tag possessed by the user. Therefore, when the same wireless tag is used by another person, user authentication may not be performed correctly, and the communication robot may operate inappropriately. Therefore, there is a need for a technology that allows appropriate communication.

  The present disclosure has been made in order to solve the above-described problems, and an object in one aspect is to provide a voice interactive apparatus in which user authentication is accurately performed and communication is realized. .

  An object in another aspect is to provide a voice interaction system for accurately performing user authentication and realizing communication according to a user.

  An object in another aspect is to provide a terminal for accurately performing user authentication and realizing communication according to a user.

  An object in another aspect is to provide a voice interaction method for accurately performing user authentication and realizing communication according to a user.

  Still another object of the present invention is to provide a program for causing a computer to function as a voice interactive apparatus in which user authentication is accurately performed and communication is realized.

  A speech dialogue apparatus according to an embodiment is configured to identify a speaker based on a speech recognition unit configured to recognize an utterance, a recognized utterance, and user information registered in advance. A speech estimation unit, a recognized speech, and a topic estimation unit configured to generate a topic that the speaker is interested in based on the identified speaker, and outputs the topic in speech And an audio output unit configured as described above.

  A voice interaction device according to another embodiment includes a voice signal input unit configured to accept an input of a voice signal based on an utterance, and a voice recognition configured to recognize the utterance based on the input voice signal Based on the voice authentication unit configured to identify the speaker, the recognized utterance, and the identified speaker based on the input unit, the input voice signal and the user information registered in advance Thus, a topic estimation unit configured to generate a topic in which the speaker is interested and an output unit configured to output a topic signal for outputting the topic by voice.

  According to another embodiment, a voice interaction system is provided. This voice interaction system includes a terminal and a server capable of communicating with the terminal. The terminal is configured to identify a speaker based on a speech recognition unit configured to accept an utterance and recognize the utterance, and the recognized utterance and user information registered in advance. An authentication unit; and a transmission unit configured to transmit an audio signal based on the utterance and an identification signal of the identified speaker to the server. The server transmits a topic estimation unit configured to generate a topic in which the speaker is interested based on the voice signal and the identification signal, and a topic signal for outputting the topic as a voice to the terminal. A topic transmission unit configured as described above. The terminal further includes an output unit configured to output the topic by voice based on the topic signal received from the server.

  According to another embodiment, a terminal used for the above-described voice interaction system is provided. The terminal includes a voice recognition unit configured to recognize an utterance, a voice authentication unit configured to identify a speaker based on the recognized utterance and pre-registered user information, A transmission unit configured to transmit an audio signal based on the utterance and an identification signal of the identified speaker to the server, and a topic signal for outputting a topic that the speaker is interested in by voice from the server And an output unit configured to receive and output the topic by voice.

  Preferably, the voice interaction device further includes a storage unit configured to store a history of interaction with each user of the voice interaction device. The topic estimation unit is configured to generate a topic based on a history of dialogues with the user.

  Preferably, the voice interaction device further includes a closeness calculation unit configured to calculate a closeness between the user and the voice interaction device based on a history of interaction with the user of the voice interaction device. The topic estimation unit is configured to adjust the tone of the topic according to the familiarity.

  According to another embodiment, a voice interaction method is provided. This method is based on a step of recognizing an utterance, a step of identifying a speaker based on the recognized utterance and pre-registered user information, a recognized utterance, and an identified speaker. And generating a topic that the speaker is interested in, and outputting the topic as a voice.

  A voice interaction method according to another embodiment includes a step of receiving an input of an audio signal based on an utterance, a step of recognizing an utterance based on the input audio signal, and a user who is registered in advance with the input audio signal A step of identifying a speaker based on the information; a step of generating a topic of interest to the speaker based on the recognized utterance and the identified speaker; and outputting the topic in speech Outputting a topic signal for the purpose.

  A speech interaction method according to another embodiment includes a step of recognizing an utterance, a step of identifying a speaker based on the recognized utterance and pre-registered user information, an audio signal based on the utterance, A step of transmitting an identification signal of the identified speaker to the server, and a topic signal for outputting a topic of interest of the speaker, which is estimated based on the voice signal and the identification signal. And a step of outputting the topic by voice based on the topic signal.

  According to another embodiment, a program for causing a computer to function as a voice interaction device is provided. The program includes, in one or more processors, a step of recognizing an utterance, a step of identifying a speaker based on the recognized utterance and pre-registered user information, a recognized utterance, A step of generating a topic signal for outputting a topic in which the speaker is interested in a voice based on the speaker, and a step of outputting the topic in a voice based on the topic signal are executed.

  According to another embodiment, a program for causing a computer to function as a voice interaction apparatus includes a step of accepting an input of an audio signal based on an utterance to one or more processors, and an utterance is recognized based on the input audio signal. And the step of identifying a speaker based on the input voice signal and pre-registered user information, the recognized utterance, and the identified speaker. A step of generating a topic having interest and a step of outputting a topic signal for outputting the topic as a voice are executed.

  A program for causing a computer to function as a voice interaction device according to another embodiment is based on the step of recognizing an utterance by one or more processors, the recognized utterance, and user information registered in advance. A step of identifying a speaker, a step of transmitting an audio signal based on the utterance and an identification signal of the identified speaker to the server, and the speaker estimated based on the audio signal and the identification signal A step of receiving a topic signal for outputting a topic of interest by voice from the server and a step of outputting the topic by voice based on the topic signal are executed.

  A voice interaction system according to another aspect includes a terminal and a server. The terminal is configured to transmit a speech signal to the server, a speech recognition unit configured to recognize the speech, a speech signal conversion unit configured to convert the recognized speech into a speech signal, and Including a transmitter. The server is configured to identify the speaker based on the speech recognition unit configured to recognize the speech based on the speech signal received from the terminal, and the speech signal and the user information registered in advance. A speech estimation unit, a recognized speech, and a topic estimation unit configured to generate a topic that the speaker is interested in based on the identified speaker, and outputs the topic in speech And a transmitter configured to transmit a topic signal for transmission to the terminal. The terminal further includes a receiving unit configured to receive a topic signal from the server, and an output unit configured to output the topic as a voice based on the topic signal.

  According to another embodiment, a terminal for use in the above system is provided. The terminal is configured to transmit a speech signal to a server, a speech recognition unit configured to recognize a speech, a speech signal conversion unit configured to convert the recognized speech into a speech signal, and A transmission unit, a reception unit configured to receive a topic signal generated based on the utterance signal from the server, and a topic corresponding to the utterance based on the topic signal. And an output unit.

  The above and other objects, features, aspects and advantages of the present invention will become apparent from the following detailed description of the present invention taken in conjunction with the accompanying drawings.

1 is a diagram conceptually showing the configuration of a voice interaction system 10. FIG. 1 is a block diagram illustrating an example of a configuration of a voice interaction system 10. FIG. 2 is a block diagram showing a hardware configuration of communication terminal 200. FIG. It is a block diagram showing the hardware constitutions of the computer 400 which implement | achieves the server which concerns on embodiment. It is a figure showing the outline | summary of a structure of the voice dialogue system with respect to the utterance using a wireless tag. 3 is a sequence chart showing processing executed when a user is registered in the voice interaction system 10. It is a sequence chart showing the process by which the reply according to the user who uttered is produced | generated. It is a figure showing an example of the interest estimation in a certain situation. 2 is a block diagram illustrating an example of a configuration of a voice interaction system 90. FIG. 3 is a sequence chart showing processing performed in the voice interaction system 10. It is a figure which represents notionally the method of estimating each user's interest. It is a figure showing the table 1200 hold | maintained at the dialog log | history memory | storage part 260. FIG. It is a figure showing the table 1300 extracted about the specific user. It is a figure showing the data structure of dialog DB section 270. It is a figure showing the outline | summary of the topic estimation by the speech dialogue system 80 in a certain situation. It is a figure which represents notionally the structure of the voice dialogue system 1600. It is a figure showing the one aspect | mode which estimates a topic based on each user's utterance. It is a figure which represents the data structure of the dialog DB part 1670 conceptually. It is a figure which represents notionally the one aspect | mode of the data storage in the dialog log | history memory | storage part 1660. FIG. FIG. 18 is a diagram conceptually illustrating one aspect of data storage in a table 2000 provided in the server 1620. It is a figure showing the data structure of dialog DB part 1670. It is a figure showing an example of a structure of the voice interactive system 2200. FIG. FIG. 10 is a diagram conceptually illustrating calculation of intimacy by an intimacy calculating module 2251. It is a figure explaining the aspect from which a response changes according to familiarity. 2 is a block diagram illustrating an outline of a configuration of a voice interaction apparatus 2500. FIG. FIG. 11 is a block diagram illustrating a schematic configuration of a voice interaction system 2600.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following description, the same parts are denoted by the same reference numerals. Their names and functions are also the same. Therefore, detailed description thereof will not be repeated.

<Technology>
First, an outline of the present disclosure will be described. The disclosed technical idea is composed of three elements as follows.

  (1) Voice recognition and voice authentication are performed in parallel. Therefore, recognition of the user's utterance content and authentication of the user are performed simultaneously.

  (2) For each user, a topic of interest of each user is estimated based on the log of the conversation content, and a dialog based on the estimated topic is generated.

  (3) The utterance content of the robot (speech dialogue apparatus or voice dialogue system) changes based on the number of dialogues and their frequency.

  As a result of these factors, the user can become familiar with the robot (voice interaction system).

  For example, the voice interaction system to which the technical idea is applied by the element (1) specifies a user without using information from a device such as a camera or a wireless tag (voice authentication), and the user's Acquisition of speech contents (voice recognition) becomes possible.

  Then, by element (2), the user's daily conversation is stored in the spoken dialogue system and analyzed as needed. Based on the analysis result, the voice dialogue system acquires a topic (sports, entertainment news, etc.) that each user is interested in from another information providing device, and provides the user with a topic according to the user who is interacting. Can do.

  Furthermore, due to element (3), the dialogue between the spoken dialogue system and the user is carried out over a long period and regularly, so that the expression of speech from the spoken dialogue system (wording, tone, etc.) changes according to the dialogue content Can do. As a result, the user can be familiar with the voice interaction system (or a voice input / output terminal such as a robot included in the voice interaction system).

<Configuration of Spoken Dialog System 10>
With reference to FIG. 1, the technical idea of the spoken dialogue system 10 according to the embodiment of the present disclosure will be specifically described. FIG. 1 is a diagram conceptually showing the configuration of the voice interaction system 10. The voice dialogue system 10 includes a voice recognition module 110, a voice authentication module 120, a topic estimation module 130, and a dialogue generation module 140.

  The user's utterance made to the voice interactive system 10 is converted into voice data. The voice recognition module 110 executes a process for recognizing voice data. The process is not particularly limited, and various known voice recognition techniques can be applied. The recognition result (for example, the recognized topic) by the speech recognition module 110 is input to the topic estimation module 130.

  The user's speech to the voice interaction system 10 is input to the voice recognition module 110 and simultaneously to the voice authentication module 120. The voice authentication module 120 authenticates the voice of the user's utterance and identifies the speaker (user of the voice interaction system 10). Therefore, voice recognition and speaker identification are performed at substantially the same timing. The identified user is input to the topic estimation module 130.

  The topic estimation module 130 estimates a topic appropriate for the user who made the utterance based on the topic recognized by the speech recognition module 110 and the user specified by the speech authentication module 120. For example, when the user is a child, a topic in which the child is interested can be extracted from a database holding news, or a topic recently exchanged with the child can be extracted from the conversation history. In another aspect, when the user is an adult woman, topics in which the woman is interested can be extracted from the database, or topics that have been exchanged with the woman in the past can be extracted from the conversation history. The result of estimation by the topic estimation module 130 is input to the dialog generation module 140.

  The dialogue generation module 140 generates the content of the utterance for the user based on the result of the topic estimation module 130 (for example, the topic of sports in which the child user is interested, the topic of fashion in which the woman is interested). In yet another aspect, the dialog generation module 140 generates the content of the utterance for the user in consideration of the closeness between the user and the voice interaction system 10 in addition to the result of the topic estimation module 130. When the dialog generation module 140 generates the content of the utterance, the voice dialog system 10 generates a signal for outputting the content as a voice, and replies to the user as a device utterance based on the signal.

[First Embodiment]
<Configuration of Spoken Dialog System 10>
With reference to FIG. 2, the configuration of the voice interaction system 10 according to the first exemplary embodiment will be described. FIG. 2 is a block diagram illustrating an example of the configuration of the voice interaction system 10. The voice interaction system 10 includes a communication terminal 200 and a server 220. Communication terminal 200 includes an audio input unit 210 and an audio output unit 211.

  The server 220 includes a control unit 230, a speech recognition unit 240, a dialog analysis unit 250, a dialog history storage unit 260, a dialog DB (Database) unit 270, a dialog generation unit 280, and a voice synthesis unit 290. . The voice recognition unit 240 includes a voice recognition module 241 and a speaker identification module 242.

  The communication terminal 200 is implemented as an electronic device having a stuffed appearance in a certain aspect, for example. In another aspect, communication terminal 200 is also realized by a liquid crystal television or other display device that can display an image of a person prepared in advance. In this case, the human image may be displayed three-dimensionally as a three-dimensional image.

  In the communication terminal 200, the voice input unit 210 receives an utterance input to the communication terminal 200 and transmits a signal corresponding to the utterance to the server 220.

  The voice output unit 211 outputs a voice as an utterance of the communication terminal 200 based on a signal sent from the server 220.

In the server 220, the control unit 230 controls the operation of the server 220.
In one aspect, control unit 230 processes a signal sent from communication terminal 200 and sends the processed signal to voice recognition unit 240 for voice recognition in server 220.

  The voice recognition unit 240 executes a voice recognition process using a known technique using a voice signal sent from the control unit 230 and a process of specifying a user (speaker) who has given the voice. More specifically, in the speech recognition unit 240, the speech recognition module 241 executes speech recognition processing accepted by the communication terminal 200. The speaker specifying module 242 specifies a speaker (user of the communication terminal 200) who has spoken the voice received by the communication terminal 200. For example, the speaker identifying module 242 may extract the voice information (for example, user identification information and voiceprint data already stored) of the user registered in advance in the server 220 and the voice signal (extracted) transmitted by the communication terminal 200. And the voiceprint data) are identified.

  The dialog analysis unit 250 analyzes the user's utterance to the communication terminal 200 based on the recognition result by the voice recognition unit 240. More specifically, the dialogue analysis unit 250 cuts out morphemes from the result of speech recognition (utterance content) and executes a specific expression extraction process.

  The dialogue history storage unit 260 holds the result of the analysis by the dialogue analysis unit 250. More specifically, the dialogue history storage unit 260 holds past conversations between the user and the voice dialogue system 10, morphemes based on the results of those conversations, appearance history of specific expressions, and the like.

  The dialogue DB unit 270 holds a pair of dialogue input phrase and response phrase for generating a prepared dialogue sentence. When a condition for generating a dialog is given to the dialog DB unit 270, a phrase corresponding to the situation specified by the condition is searched from a plurality of response phrases.

  The dialog generation unit 280 generates a dialog using the data held in the dialog history storage unit 260 and the database held by the dialog DB unit 270. More specifically, the dialog generation unit 280 uses the response phrase searched based on the user's utterance content given to the dialog DB unit 270 and the dialog history stored in the dialog history storage unit 260 to use the voice dialog system 10. Generate a dialogue with the user who is speaking. The dialogue sentence is generated as character string information, for example.

  The voice synthesis unit 290 performs voice synthesis using the dialogue sentence generated by the dialogue generation unit 280 and generates data for the speech of the voice dialogue system 10.

  When control unit 230 receives the data synthesized by speech synthesis unit 290, control unit 230 transmits the data to communication terminal 200.

  In the communication terminal 200, the voice output unit 211 outputs the dialogue sentence as a voice.

  As an example, in a certain situation, when the child user 201 utters “just now”, the communication terminal 200 receives an input of the content of the utterance and transmits an audio signal to the server 220. The server 220 performs a speech recognition process on the utterance content of the user 201, recognizes that the uttered content is “now”, and specifies that the utterer is the user 201 (child). . Based on the result of such speech recognition, the server 220 executes speech synthesis processing for speaking “Return. How was the school?” As a dialogue for “I ’m right now”, and the result of the processing is the communication terminal 200. Send to. As a result, the communication terminal 200 utters “Return. How was the school?” To the user 201.

  In another aspect, when the adult female user 202 utters the same word “I'm right now”, the communication terminal 200 accepts the voice input of the utterance and sends a voice signal to the server 220. The server 220 recognizes the content of the utterance by voice and identifies the speaker. More specifically, the server 220 recognizes the content of the utterance “immediately” and, at the same time, specifies that the content of the utterance is from the user 202 (adult female). The server 220 generates a dialogue sentence for “Now” based on the result of such speech recognition. More specifically, the server 220 generates an utterance content “Return. Thank you for your work” as a dialog for responding to the utterance content by the user 202. When the dialogue is synthesized with the dialogue, the server 220 transmits the synthesized signal to the communication terminal 200. The communication terminal 200 outputs, to the user 202, “Return.

<Configuration of communication terminal 200>
With reference to FIG. 3, the configuration of communication terminal 200 according to the present embodiment will be described. FIG. 3 is a block diagram showing a hardware configuration of communication terminal 200.

  The communication terminal 200 includes a CPU (Central Processing Unit) 20, an antenna 23, a communication device 24, operation buttons 25, a camera 26, a flash memory 27, a RAM (Random Access Memory) 28, and a ROM (Read Only). Memory) 29, memory card drive device 30, microphone 32, speaker 33, audio signal processing circuit 34, monitor 35, LED (Light Emitting Diode) 36, data communication interface 37, vibrator 38, An acceleration sensor 39 and an actuator 40 are provided. A memory card 31 can be attached to the memory card drive device 30.

  The antenna 23 receives a signal transmitted by the server 220 or transmits a signal for communicating with another communication device via the server 220. The signal received by the antenna 23 is subjected to front-end processing by the communication device 24, and the processed signal is sent to the CPU 20.

  The operation button 25 receives an operation on the communication terminal 200. The operation button 25 is realized as a hard key or a soft key, for example. When the operation button 25 receives an operation by the user, the operation button 25 sends a signal to the CPU 20 according to the operation mode of the communication terminal 200 at that time.

  The CPU 20 executes a process for controlling the operation of the communication terminal 200 based on a command given to the communication terminal 200. When the communication terminal 200 receives the signal, the CPU 20 executes a predetermined process based on the signal sent from the communication device 24 and sends the processed signal to the audio signal processing circuit 34. The audio signal processing circuit 34 performs predetermined signal processing on the signal, and sends the processed signal to the speaker 33. The speaker 33 outputs sound based on the signal.

  The microphone 32 receives an utterance to the communication terminal 200 and sends a signal corresponding to the uttered voice to the voice signal processing circuit 34. The audio signal processing circuit 34 performs a predetermined process on the signal, and sends the processed signal to the CPU 20. The CPU 20 converts the signal into data for transmission, and sends the converted data to the communication device 24. The communication device 24 generates a signal for transmission using the data, and transmits the signal to the antenna 23. A signal transmitted from the antenna 23 is received by the server 220. In another aspect, instead of antenna 23, server 220 and communication terminal 200 may be connected by wire.

  The flash memory 27 stores data sent from the CPU 20. In addition, the CPU 20 reads data stored in the flash memory 27 and executes a predetermined process using the data.

  The RAM 28 temporarily holds data generated by the CPU 20 based on the operation performed on the operation button 25. The ROM 29 stores a program or data for causing the communication terminal 200 to execute a predetermined operation. The CPU 20 reads the program or data from the ROM 29 and controls the operation of the communication terminal 200.

  The memory card driving device 30 reads data stored in the memory card 31 and sends it to the CPU 20. The memory card drive device 30 writes the data output by the CPU 20 in the empty area of the memory card 31.

  The audio signal processing circuit 34 performs signal processing for a call as described above. In the example shown in FIG. 3, the CPU 20 and the audio signal processing circuit 34 are shown as separate configurations. However, in other aspects, the CPU 20 and the audio signal processing circuit 34 may be configured as an integral unit. Good.

  The monitor 35 displays an image based on data acquired from the CPU 20. The monitor 35 has, for example, still images (for example, conference materials, contracts and other electronic documents), moving images, and music file attributes (name of the file, performer, performance time, etc.) stored in the flash memory 27. Is displayed. The still image may include a drawn image and an image prepared in advance by a manufacturer of the communication terminal 200 by default.

  The LED 36 realizes a predetermined light emission operation based on a signal from the CPU 20. The data communication interface 37 accepts attachment of a data communication cable.

  The data communication interface 37 sends a signal output from the CPU 20 to the cable. Alternatively, the data communication interface 37 sends data received via the cable to the CPU 20.

  Vibrator 38 performs an oscillating operation at a predetermined frequency based on a signal output from CPU 20.

  The acceleration sensor 39 detects the direction of acceleration acting on the communication terminal 200. The detection result is input to the CPU 20.

  The actuator 40 drives some members (not shown) of the communication terminal 20 based on a signal from the CPU 20. For example, when the communication terminal 20 is realized as an electronic device having the appearance of a stuffed toy, the actuator 40 can drive the hand, the leg, the neck, and other parts of the stuffed toy. Thereby, the communication terminal 40 can perform an operation (nodding, swinging, etc.) according to the user's utterance.

  Note that the communication terminal 200 according to the present embodiment need not include all the above-described components, and may be any information processing terminal having at least a voice input / output function and a communication function.

<Server configuration>
With reference to FIG. 4, a configuration of server 220 according to the present embodiment will be described. FIG. 4 is a block diagram illustrating a hardware configuration of a computer 400 that implements the server 220 according to the embodiment.

  The computer 400 includes, as main components, a CPU 1 that executes a program, a mouse 2 and a keyboard 3 that receive input of instructions from a user of the computer 400, data generated by execution of a program by the CPU 1, or a mouse 2 or a keyboard 3 A RAM 4 that stores data input via the volatile memory, a hard disk 5 that stores data in a nonvolatile manner, an optical disk drive device 6, a communication IF (Interface) 7, and a monitor 8. Each component is connected to each other by a bus. A CD-ROM 9 and other optical disks can be mounted on the optical disk drive 6. The communication IF 7 includes, but is not limited to, a USB (Universal Serial Bus) interface, a wired LAN (Local Area Network), a wireless LAN, and a Bluetooth (registered trademark) interface.

  Processing in the computer 400 is realized by each hardware and software executed by the CPU 1. Such software may be stored in the hard disk 5 in advance. The software may be stored in a CD-ROM 9 or other non-volatile computer-readable data recording medium and distributed as a program product. Alternatively, the software may be provided as a program product that can be downloaded by an information provider connected to the Internet or other networks. Such software is read from the data recording medium by the optical disk drive 6 or other data reading device, or downloaded via the communication IF 7 and then temporarily stored in the hard disk 5. The software is read from the hard disk 5 by the CPU 1 and stored in the RAM 4 in the form of an executable program. The CPU 1 executes the program.

  Each component constituting the computer 400 shown in FIG. 4 is general. Therefore, it can be said that an essential part according to the present embodiment is a program stored in the computer 400. Since the operation of the hardware of computer 400 is well known, detailed description will not be repeated.

  The data recording medium is not limited to a CD-ROM, FD (Flexible Disk), and hard disk, but is a magnetic tape, cassette tape, optical disk (MO (Magnetic Optical Disc) / MD (Mini Disc) / DVD (Digital Versatile Disc)). ), IC (Integrated Circuit) card (including memory card), optical card, mask ROM, EPROM (Electronically Programmable Read-Only Memory), EEPROM (Electronically Erasable Programmable Read-Only Memory), flash ROM, etc. It may be a non-volatile data recording medium that carries a fixed program.

  The program here may include not only a program directly executable by the CPU but also a program in a source program format, a compressed program, an encrypted program, and the like.

<Voice interaction system using wireless tags>
With reference to FIG. 5, a voice interaction system according to another aspect will be described. FIG. 5 is a diagram showing an outline of the configuration of a voice interaction system for utterance using a wireless tag.

  The voice interaction system 50 includes a communication terminal 500 and a server 520. Communication terminal 500 includes a voice input unit 210, a voice output unit 211, and a wireless tag information transmission unit 510. The server 520 includes a control unit 530, a voice recognition unit 540, a user identifier unit 541, a dialog analysis unit 550, a dialog history storage unit 560, a dialog DB unit 570, a dialog generation unit 580, and a voice synthesis unit 590. Including.

  In one aspect, the user 201 has a mobile phone 501. The cellular phone 501 has, for example, a wireless tag A as its identification information. In a certain situation, when the user 201 speaks “I'm right now”, it is input to the communication terminal 500 together with the identification information (wireless tag A) of the mobile phone 501. The communication terminal 500 recognizes the content of the user's utterance. In the communication terminal 500, the wireless tag information transmission unit 510 extracts the wireless tag A accompanying the utterance from the user 201, and transmits the extracted content to the server 520. The server 520 recognizes the user 201's utterance content “now” and the wireless tag A, and a dialog corresponding to the user 201 is generated by the dialog generation unit 580.

  When the server 520 generates a dialogue sentence for the user 201, the server 520 transmits the signal to the communication terminal 500. The communication terminal 500 utters “Return. How was the school?” To the user 201.

  In another aspect, when an adult user 202 uses the same mobile phone 501 and utters “immediately”, the communication terminal 500 receives the utterance content “immediately” and communicates with the mobile phone 501. By doing so, the wireless tag A associated with the mobile phone 501 is acquired. The communication terminal 500 transmits the utterance content “Now” and the wireless tag A to the server 520. That is, the communication terminal 500 transmits the wireless tag A associated with the mobile phone 501 regardless of whether the subject of the utterance is the user 201 or the user 202. When the server 520 receives the wireless tag A and the content of the utterance “immediately”, the server 520 determines using the wireless tag A that the utterance is made by the child user 201 even though the speaker is the adult user 202. To do. The server 520 synthesizes the same dialogue sentence as the dialogue sentence for the user 201. The server 520 transmits a dialogue sentence “Return. How was the school?” To the communication terminal 500. The communication terminal 500 speaks to the user 202 “Return. How was the school?” As described above, when a user is recognized using a wireless tag like a wireless communication terminal (for example, a mobile phone 501) that can be owned by the user, the speaker can easily impersonate another person. Since the server 520 cannot identify which user has spoken, the server 520 generates a dialogue sentence unique to the mobile phone 501 instead of a dialogue sentence according to the user.

<Control structure>
With reference to FIG. 6, the control structure of the spoken dialogue system 10 according to the present embodiment will be described. FIG. 6 is a sequence chart showing processing executed when a user is registered in the voice interaction system 10.

  In step S <b> 610, the user who requests registration in the voice interaction system 10 transmits a request for voice authentication learning to the communication terminal 200. When receiving the request, the communication terminal 200 establishes communication with the server 220 and transmits the request to the server 220. The control unit 230 of the server 220 receives the request.

  In step S620, the control unit 230 of the server 220 notifies the user of a voice authentication learning message in response to receiving the voice authentication learning request. More specifically, the control unit 230 transmits the message to the communication terminal 200. When the communication terminal 200 receives the message from the server 220, the voice output unit 211 outputs the message by voice. The user can know the message that must be spoken for voice authentication. Thereafter, control is transferred to step S625.

  In step S625, control unit 230 transmits a command to instruct voice acquisition to communication terminal 200.

  In step S630, upon receiving the command from server 220, communication terminal 200 outputs a message prompting the user to speak. More specifically, for example, the communication terminal 200 outputs a message for prompting utterance “Please speak after this message is finished” by voice. In another aspect, communication terminal 200 may display a message on a monitor. In still another aspect, when communication terminal 200 is realized as a stuffed toy having a voice input / output function, a communication function, and a drive function, communication terminal 200 performs an operation for prompting utterance like a gesture of placing a hand on an ear. You may do it.

  In step S640, when the user recognizes the message prompting the utterance, the user utters the voice authentication learning message toward the communication terminal 200.

  In step S650, voice input unit 210 of communication terminal 200 receives an input of a message utterance by a user, and transmits voice data corresponding to the utterance to server 220.

  In step S660, when the control unit 230 of the server 220 detects reception of the voice data, the control unit 230 transmits a learning request for the message to the voice recognition unit 240. In response to receiving the learning request, the voice recognition unit 240 performs voice recognition processing and speaker identification processing. More specifically, the voice recognition unit 240 performs voice recognition processing on the content of an utterance made by the user as the voice recognition module 241. In addition, the speech recognition unit 240 extracts, as the speaker identification module 242, morphemes from the content of the utterance and acquires information for identifying the speaker who performed the utterance.

  In step S670, the speech recognition unit 240 transmits a learning completion response indicating that learning is completed to the control unit 230.

  In step S680, when receiving the learning completion response, control unit 230 notifies communication terminal 200 that learning of the user has been completed.

  In step S690, communication terminal 200 utters a message notifying completion of learning to the user. In this way, registration processing for user identification is executed. In another aspect, the request in step S610 may be made directly by the user to the server. A terminal used for registering a user in the voice interaction system 10 is not limited to the communication terminal 200. Any information processing communication terminal having at least a voice recognition function and a communication function with the server 220 may be used.

  The above processing is realized when a processing program for registering a user in the communication terminal 200 and the server 220 is executed in advance. The trigger for starting the process for registering the user may be a specific utterance (such as user registration request) by the user, an input switch of the communication terminal 200 or the server 220, or other input operation.

  With reference to FIG. 7, an utterance sequence based on user identification according to the present embodiment will be described. FIG. 7 is a sequence chart showing a process of generating a response according to the user who spoke.

In step S710, the user speaks to communication terminal 200.
In step S <b> 720, when receiving an input of the user's utterance, communication terminal 200 transmits voice data corresponding to the utterance to server 220.

  In step S730, when the control unit 230 of the server 220 detects reception of voice data from the communication terminal 200, the control unit 230 transmits a request for recognizing the voice data to the voice recognition unit 240.

  In step S740, in response to the recognition request, the speech recognition unit 240 executes speech recognition processing for recognizing the content of the utterance and speaker identification processing for identifying the speaker. Further, the voice recognition unit 240 transmits the result of voice recognition and the result of speaker identification to the control unit 230 as a recognition response.

  In step S750, in response to receiving the recognition response, control unit 230 transmits an analysis generation request to dialog analysis unit 250 and dialog generation unit 280, respectively. When the dialog analysis unit 250 receives the request, the dialog analysis unit 250 refers to the dialog history storage unit 260 and extracts the past dialog history of the user. The dialogue generation unit 280 uses the dialogue history held in the dialogue history storage unit 260 and the dialogue database held in the dialogue DB unit 270 in response to receiving the generation request, and the user who made the utterance Generate a dialog sentence specific to.

  In step S760, dialog analysis unit 250 and dialog generation unit 280 transmit the result of the utterance analysis and the generated dialog to control unit 230. The control unit 230 causes the speech synthesis unit 290 to perform speech synthesis of the dialogue sentence based on the reception of the analysis result and the generated dialogue. The speech synthesizer 290 generates a response to the utterance by speech synthesis processing. In step S770, control unit 230 transmits the response phrase generated by speech synthesis unit 290 to communication terminal 200.

  In step S780, when receiving a response phrase from server 220, communication terminal 200 utters the response phrase to the user. Thereby, the dialog peculiar to the user who performed the utterance to the voice dialog system 10 can be realized.

<Interest estimation>
With reference to FIG. 8, an outline of a case where a reply tailored to a user based on user identification (interest estimation) is described as an example of the voice interaction system 10 according to the present embodiment. FIG. 8 is a diagram illustrating an example of interest estimation in a certain situation. A voice interaction system 80 according to an aspect includes a communication terminal 500 and a server 820. Communication terminal 500 includes an audio input unit 210, an audio output unit 211, and an RFID tag information transmission unit 510. Server 820 includes a control unit 530, a speech recognition unit 540, a dialog analysis unit 850, a dialog generation unit 880, a dialog DB unit 570, and a speech synthesis unit 590.

  The example shown in FIG. 8 is a case where the voice conversation system 80 cannot identify the user who made the utterance and cannot know the user's interest. In this case, when the user 201 utters “tell news” to the communication terminal 500, the communication terminal 500 communicates with the server 820, and as an appropriate response to “tell news”, for example, the latest news “Today, The Japanese national team drew with Greece. When the server 820 transmits the result of the dialogue to the communication terminal 500, the communication terminal 500 speaks to the user 201 “Today, the Japanese representative has drawn with Greece”.

  When another user (for example, an adult female user 202) issues the same question “tell news” to such a voice interaction system 80, the communication terminal 500 communicates with the server 820. At this time, if the latest news has not been updated, the server 820 outputs the result of the dialogue to the user 201 as “Tell me the news” “Today, the Japanese representative has drawn with Greece” Identify utterances. As a result, the communication terminal 500 speaks to the adult female user 202 as “Today, the Japanese representative has drawn with Greece”. In other words, regardless of the type and interest of the user, the voice interaction system 80 makes the same response to the same utterance (an utterance having the same keyword).

<Configuration of Spoken Dialog System 90>
With reference to FIG. 9, a spoken dialogue system 90 according to the present embodiment will be described. FIG. 9 is a block diagram illustrating an example of the configuration of the voice interaction system 90. The voice interaction system 90 can make a response tailored to the user based on the user identification. The voice interaction system 90 includes a communication terminal 200 and a server 920. Communication terminal 200 includes an audio input unit 210 and an audio output unit 211. The server 920 includes a dialog generation unit 980 instead of the dialog generation unit 280 with respect to the configuration of the server 220. The dialog generation unit 980 includes an interest estimation module 990.

  In a certain situation, when the user 201 utters “tell news” to the communication terminal 200, the communication terminal 200 communicates with the server 920 in accordance with the interest of the user 201 and the content of the utterance (tell news). Create a conversation. More specifically, in the dialog generation unit 980, the interest estimation module 990 estimates the interest of the user 201 based on the interest unique to the user 201 and the content of the utterance (tell the news). For example, the interest estimation module 990 detects from the dialogue history storage unit 260 that sports are included as the interest of the user 201. The interest estimation module 990 generates a dialog corresponding to the user 201 based on such a detection result. For example, the interest estimation module 990 refers to the data held in the dialogue DB unit 270 (sports-specific news), and generates a dialogue “Today the Japanese representative has drawn with Greece”. When the server 920 generates a dialogue unique to the interest of the user 201 based on the result of the interest estimation and performs speech synthesis of the dialogue, the communication terminal 200 gives the user 201 “Today, the Japanese representative is Greek. "I was drawn."

  In another aspect, when the adult female user 202 utters “Tell me the news”, the communication terminal 200 transmits the content of the utterance to the server 920. In the server 920, the interest estimation module 990 generates a dialog corresponding to the speaker. More specifically, first, the interest estimation module 990 specifies that the user 202 who made the utterance “tell me the news” is an adult woman, and the interest (for example, entertainment related) of the user 202 is specified. Identify. The interest estimation module 990 accesses the dialogue DB unit 270 and identifies the latest entertainment-related news. When the dialogue-related unit 980 identifies entertainment-related news as a dialogue according to the user 202, the dialogue generation unit 980 generates a dialogue “Tell me the news” “Yushima Oshima graduated just 9 days after a surprise return”. When the server 920 transmits the generated dialog to the communication terminal 200, the communication terminal 200 utters to the user 202 “Yuko Oshima graduated. In this way, the voice interaction system 90 performs an utterance according to the speaker.

  The voice dialogue system 90 analyzes information on the user's past utterance contents (for example, a lot of soccer talks or a lot of entertainment-related talks) from the utterance information so far. Keep as history. As a result, the voice interaction system 90 can make an interesting utterance according to each of a plurality of users.

<Control structure>
With reference to FIG. 10, the control structure of the voice interaction system 90 will be described. FIG. 10 is a sequence chart showing processing performed in the voice interaction system 10. The same steps as those described above are denoted by the same step numbers. Therefore, the description of the same process will not be repeated.

  In step S1010, dialog analysis unit 250 and dialog generation unit 280 transmit an interest acquisition request to dialog history storage unit 260. The dialogue history storage unit 260 extracts an interest unique to the recognized user from the interest acquisition request.

  In step S1020, dialog history storage unit 260 transmits an interest acquisition response to dialog analysis unit 250 and dialog generation unit 980. More specifically, the dialogue history storage unit 260 extracts the interest from the dialogue DB unit 270 via the dialogue generation unit 980 with reference to the interest specific to the user included in the interest acquisition request, and The extraction result is transmitted to the dialog analysis unit 250 and the dialog generation unit 280.

  In step S1030, control unit 230 transmits a dialogue log storage request to dialogue history storage unit 260. More specifically, the control unit 230 transmits a request to save the dialogue log and a dialogue log (or data for identifying the dialogue log) to be saved to the dialogue history storage unit 260.

  In step S1040, dialogue history storage section 260 saves a dialogue log specified by the request based on the reception of the request.

<Method of estimating interest>
With reference to FIG. 11, a user's interest estimation method of the voice interaction system will be described. FIG. 11 is a diagram conceptually showing a method for estimating the interest of each of a plurality of users.

  In one aspect, the user 201 utters “Do you know a new game?” To the voice interaction system 90. When the spoken dialogue system 90 identifies the speaker (user 201) and recognizes the content of the utterance (Do you know a new game?), The voice dialogue system 90 extracts a keyword (for example, a noun “game”) included in the utterance, and stores the dialogue history storage unit. The keyword “game” is stored in 260 in association with the user 201.

  In another aspect, when another user 202 speaks to the voice interaction system 90 that “a new cafe has been created in the neighborhood”, the server 920 extracts the keyword “cafe” and the extracted keyword and the user 202. Are stored in the dialogue history storage unit 260 in association with each other.

  In this manner, the dialogue history storage unit 260 sequentially accumulates keywords (for example, nouns) included in the user's utterance for each user.

  The interest estimation module 990 performs scoring based on the number of appearances of the noun included in the utterance content of each user stored in the dialogue history storage unit 260 and the appearance time. The interest estimation module 990 treats the user as an event of interest from the highest score. For example, the interest estimation module 990 sets the coefficient so that the score of a noun that appears in a more recent fixed period is higher. The coefficient setting method can be set to increase proportionally or stepwise, for example.

<Data structure>
With reference to FIG. 12 and FIG. 13, the data structure of the voice interaction system 90 according to the present embodiment will be described. FIG. 12 is a diagram showing a table 1200 held in the dialogue history storage unit 260. FIG. 13 is a diagram showing a table 1300 extracted for a specific user.

  As shown in FIG. 12, the table 1200 is created and updated as a dialog history interest record table. The table 1200 includes a record ID 1210, a user ID 1220, a speaker 1230, an interest noun 1240, and a time stamp 1250. The record ID 1210 identifies a dialogue performed by the voice dialogue system 90 and each user. The user ID 1220 is registered in the voice interaction system 90. Identifies the user. The speaker 1230 is the name of the user who made the utterance. The interest noun 1240 represents a noun that the user is interested in. The time stamp 1250 specifies the time when the utterance is recognized. The time stamp 1250 can be used to appropriately set a period from which records are to be extracted.

  Referring to FIG. 13, table 1300 includes record ID 1210, user ID 1220, speaker 1230, interest noun 1240, and time stamp 1250. For example, when the user ID 1220 is specified as “12344312” as the user of the voice interaction system 90, each record having the value of the user ID is extracted as shown in the table 1300. It can be seen that this user has, for example, “music”, “kyary pamyu pamyu”, and “summer sonic” as interests.

  The data structure of the voice interaction system 90 will be further described with reference to FIG. FIG. 14 is a diagram showing the data structure of dialog DB unit 270. The dialogue DB unit 270 includes an input phrase 1110, an interest noun 1120, and an output phrase 1130. The input phrase 1110 represents the utterance content input to the voice interactive system 90.

  The interest noun 1120 corresponds to the interest noun 1240 stored in the dialogue history storage unit 260. The output phrase 1130 represents the response content for the user associated with each of the interest nouns 1120.

[Effect of the embodiment]
As described above, according to the present embodiment, it is possible to authenticate a user and provide a topic tailored to the user without requiring the user to possess an RFID or introducing a camera to the voice interaction system. It becomes possible. Since the voice interaction system can provide what the user has spoken or related in the past, a smooth conversation between the user and the voice interaction system is possible.

  When the user uses it regularly and for a long period of time, the utterance content of the robot (communication terminal 200) constituting the voice interaction system changes to something more familiar. When the robot responds based on the content that the user is interested in, it becomes possible for the user to be familiar with and attached to the robot.

[Second Embodiment]
Hereinafter, a second embodiment will be described. In the voice interactive system according to the second embodiment, the topic of a specific user can be estimated.

<Topic estimation>
With reference to FIG. 15, another response (topic estimation) that responds to the user based on the user identification will be described. FIG. 15 is a diagram showing an outline of topic estimation by the voice interaction system 80 in a certain situation.

  The user 1501 utters “I'm traveling to Kyoto on the weekend” to the voice interaction system 80 (utterance 1510). When the voice conversation system 80 recognizes the utterance 1510, the communication terminal 500 utters “Kyoto is a golden pavilion” (utterance 1520). Thereafter, when the user 1501 utters “Is there a recommended souvenir?” (Utterance 1530), the voice dialogue system 80 recognizes the voice of the utterance 1530, and based on the recognition result, the communication terminal 500 displays the message “ "What souvenir?" Is output by voice (utterance 1540).

  As is clear from the content of the utterance 1540, the voice dialogue system 80 does not know the topic that the user 1501 has spoken until immediately before, so the user needs to express it accurately.

<Configuration of Spoken Dialog System 1600>
Therefore, with reference to FIG. 16, a spoken dialogue system 1600 according to the present embodiment will be described. FIG. 16 is a diagram conceptually showing the configuration of the voice interaction system 1600. The voice interaction system 1600 includes a communication terminal 200 and a server 1620. The server 1620 includes a dialogue history storage unit 1660 instead of the dialogue history storage unit 260 in the configuration of the server 220 shown in FIG. The server 1620 also includes a dialog generation unit 1680 and a dialog DB unit 1670. Dialog generation unit 1680 includes a topic estimation module 1690. The other configuration of the voice interaction system 1600 according to the present embodiment is the same as the configuration of the voice interaction system 90. Therefore, the description of the same configuration will not be repeated.

  In voice dialogue system 1600 according to the present embodiment, when user 1501 utters “communication on weekends in Kyoto” with respect to communication terminal 200 (utterance 1510), communication terminal 200 says “Kyoto is Kinkaku”. Is returned to the user 1501 (utterance 1520). When the user 1501 returns “Is there a recommended souvenir?” To the communication terminal 1200 (utterance 1530), the voice dialogue system 1600 recommends Yatsuhashi based on the past history and topic. ”To the user 1501 (utterance 1640).

  That is, according to the voice dialogue system 1600, the server 1620 can refer to the topic (for example, a story about Kyoto) that has been spoken immediately before between the user 1501 and the voice dialogue system 1600. A more natural dialogue.

  Note that the utterance sequence using topic estimation by the voice interaction system 1600 is the same as the utterance sequence using interest estimation (FIG. 10). Therefore, the description of the utterance sequence of the voice interaction system 1600 will not be repeated.

<Topic estimation method>
With reference to FIG. 17, the topic estimation method in the voice interaction system 1600 will be described. FIG. 17 is a diagram illustrating an aspect in which the voice conversation system 1600 estimates a topic based on the utterances of each of a plurality of users.

  In one aspect, the user 201 utters (goes to Ueno Zoo on an excursion tomorrow) to the communication terminal 200 of the voice interaction system 1600. The spoken dialogue system 1600 extracts “Ueno Zoo” as an interesting noun (keyword) from the utterance, and stores the extracted content in the dialogue history storage unit 1660. In another aspect, when the adult user 202 utters (there is a delicious cafe in Daikanyama), the server 1620 extracts “Daikanyama” as an interest noun, and associates the extracted result with the user 202 for dialogue. Stored in the history 1660. That is, the server 1620 extracts a specific expression, and stores the obtained word and its type in the dialog history storage unit 1660 as a dialog history. The unique expressions include, for example, eight types of names such as an organization name, a person name, a place name, a date expression, a time expression, a monetary expression, a ratio expression, and a unique object name.

  The dialogue generation unit 1680 refers to the dialogue history stored in the dialogue history storage unit 1660 and extracts topics. More specifically, the topic estimation module 1690 extracts, from the data stored in the dialogue history storage unit 1660, a specific expression recorded within a predetermined fixed time as a topic. The topic estimation module 1690 uses the extracted topics as a filter to narrow down candidates from data stored in the dialog DB unit 1670. In the example shown in FIG. 17, a specific expression (Okinawa, Ishigakijima) extracted within a certain period of time that is predetermined in advance from the last utterance by the user 202 is used as a filter.

<Data structure>
The data structure of the server 1620 will be described with reference to FIG. FIG. 18 is a diagram conceptually showing the data structure of dialog DB unit 1670. In one aspect, dialog DB unit 1670 includes a table 1800. The table 1800 has a topic filter structure. More specifically, table 1800 includes user utterance 1810, place name 1820, and reply phrase 1830. User utterance 1810 represents an utterance made by a user of voice interaction system 1600. The place name 1820 represents a place name extracted as a specific expression at the time of the utterance. The response phrase 1830 represents a response output in the dialog with the user.

  With reference to FIG. 19, the data structure of dialog system 1600 according to the present embodiment will be further described. FIG. 19 is a diagram conceptually showing one mode of data storage in dialog history storage unit 1660. The dialogue history storage unit 1660 includes a table 1900. The table 1900 stores a history of dialogue between the user and the voice dialogue system 1600. The table 1900 includes a record ID 1910, a user ID 1920, a speaker 1930, a topic keyword 1940, a topic type 1950, and a time stamp 1960. The record ID 1910 identifies each record included in the table 1900. The user ID 1920 identifies the user who made the utterance of the record. The speaker 1930 specifies the user (speaker) specified by the user ID. The topic keyword 1940 represents a noun extracted as a specific expression from the utterance by the user. The topic type 1950 represents the type of topic specified by the topic keyword 1940. The topic type 1950 is expressed as, for example, an organization (ORGANIZATION), a location (LOCATION), or the like. Time stamp 1960 represents the time when the utterance was added to table 1900.

  The data structure of the server 1620 will be further described with reference to FIG. FIG. 20 is a diagram conceptually illustrating one aspect of data storage in table 2000 provided in server 1620. The table 2000 includes a record ID 1910, a user ID 1920, a speaker 1930, a topic keyword 2040, a topic type 1950, and a time stamp 1960. The table 2000 includes an utterance record only for the user whose user ID 1920 is specified by “12344312”.

  More specifically, as indicated by the topic keyword 2040, the user has Ueno Zoo, Ueno, and Tokyo as topics. Therefore, when a certain user utters the voice dialogue system 1600, the dialogue generation unit 1680 uses the topic keywords 2040 of Ueno Zoo, Ueno, and Tokyo as the latest topic of the user, and responds to the utterance from the user. Can be generated.

  With reference to FIG. 21, the data structure of voice interactive system 1600 according to the present embodiment will be further described. FIG. 21 shows a data structure of dialog DB unit 1670. The dialogue DB unit 1670 includes an input phrase 2110, a place 2120, and an output phrase 2120. The dialogue DB unit 1670 holds data for performing filtering based on the location estimated by the topic. For example, in the voice interaction system 1600, when the user gives an utterance “Tell me about trivia” to the communication terminal 200 (input phrase 2110), the server 1620 extracts a location (for example, “Hokkaido”) associated with the utterance. . In this case, the topic estimation module 1690 extracts four output phrases 2120 whose place 2120 is “Hokkaido”.

[Effect of the embodiment]
As described above, according to the voice interaction system according to the present embodiment, when the conversation between the user and the communication terminal continues, the immediately preceding topic is understood, so that the communication terminal 200 responds to the user's utterance. Detailed responses. For example, when a user traveling in Kyoto visits a souvenir, the communication terminal can respond with a souvenir associated with Kyoto. Since such a response can be made, the dialogue between the user and the communication terminal becomes a more natural dialogue.

[Third Embodiment]
The third embodiment will be described below. The spoken dialogue system according to the present embodiment is different from the spoken dialogue system according to the above-described embodiment in that the tone of the utterance output by voice is different according to the familiarity with the user.

<Intimacy>
First, a voice interaction system 2200 according to the third embodiment will be described with reference to FIG. FIG. 22 is a diagram illustrating an example of the configuration of the voice interaction system 2200. The voice interaction system 2200 includes a communication terminal 200 and a server 2220. The server 2220 includes a dialog analysis unit 2250 instead of the dialog analysis unit 250 and a dialog generation unit 2280 instead of the dialog generation unit 280 with respect to the configuration of the server 220. The dialog analysis unit 2250 includes a closeness calculation module 2251. Other configurations are the same as those shown in FIG. Therefore, the description of the same configuration will not be repeated.

  The voice interaction system 2200 is different from the voice interaction systems according to the above-described embodiments in that the intimacy changes based on the number of interactions with the user and the frequency thereof and the response changes. For example, the user 1501 utters “good morning” to the communication terminal 200 three months ago. At this time, since the voice interaction system 2200 and the user 1501 are not so intimate, the server 2220 returns a response to the utterance “good morning” based on the data stored in advance. Specifically, the communication terminal 200 utters in a polite tone to the user 1501 “Good morning. Good weather today”.

  On the other hand, when the intimacy between the user 1501 and the voice interaction system 2200 is increasing at the present time after three months have passed, the communication terminal 200 is more effective even when the user 1501 makes the same utterance as “good morning”. As a frank expression, he says, "Good morning. Why don't you go out because the weather is good today?" In this way, the response contents of the responses to the same utterance (good morning) by the same user 1501 change depending on the familiarity that can change over time.

<Calculation method of intimacy>
With reference to FIG. 23, a method of calculating the familiarity in voice interaction system 2200 according to the present embodiment will be described. FIG. 23 is a diagram conceptually showing the calculation of the familiarity by the familiarity calculating module 2251.

  Case (A) is a diagram showing a case where the intimacy increases due to the user continuously speaking to the voice interaction system. That is, as shown in the graph 2310, as the user 2002 speaks to the communication terminal 200 over time, the closeness between the voice interactive system 2200 and the user 202 increases. The degree of increase is linear (proportional) in a certain aspect, but the degree of increase is not necessarily limited to the level seen in the graph 2310. For example, the intimacy may increase step by step (step function). It can change based on the degree of emotion represented by the words that appear in the spoken phrase. For example, intimacy does not increase when the appearing word is a negative phrase. On the other hand, when the appearing word is a positive phrase, the familiarity can be increased according to the positive degree.

  In addition, when the user 202 is not continuously using the voice interaction system 2200, the intimacy of a certain value may be decreased within a predetermined period so that the intimacy is decreased. Good. In a certain situation, the familiarity calculation module 2251 calculates the familiarity for each user using the conversation history stored in the conversation history storage unit 260. For example, the familiarity calculation module 2251 calculates the familiarity of each user by sequentially adding a preset value as the degree of increase in familiarity or subtracting the value in the case of a negative phrase.

  Case (B) is a diagram showing a mode in which an increase in closeness is suppressed when utterances having the same phrase are performed by the same user 202. That is, as shown in the graph 2320, when the user 202 speaks “tell me the weather”, the intimacy may initially increase by a certain fixed degree. However, if the user 202 does not speak only with the same phrase, the voice interaction system 2200 suppresses an increase in familiarity for the user. More specifically, the familiarity calculation module 2251 refers to the data stored in the dialogue history storage unit 260 and determines whether phrases (nouns) included in the utterances by the user 202 are the same. . When the closeness calculation module 2251 detects that the utterance by the user 202 is repetition of the utterance including the same phrase, the closeness calculation module 2251 maintains the closeness for the user 202 at a constant value.

  Thus, for example, if a user asks only the weather forecast every day, or asks only general topics such as news, if the phrases included in the utterance by the user are the same, the user's Intimacy remains constant.

<Changes in responses by spoken dialogue system>
With reference to FIG. 24, the change of the response by the voice interaction system 2200 will be described. FIG. 24 is a diagram for explaining a mode in which the response changes according to the familiarity.

  As shown in the case (A), the phrase of the dialogue DB unit 270 can change according to the familiarity. The dialogue DB unit 270 includes a table 2400. Table 2400 includes user utterances 2410, intimacy 2420, and reply phrases 2430. User utterance 2410 represents an utterance to the voice interaction system 2200 by the user. The familiarity 2420 is divided, for example, by a plurality of sections. The reply phrase 2430 represents a phrase stored in advance according to each familiarity. For example, when the user utterance 2410 is “good morning” and the closeness 2420 is determined to be “high”, the response phrase 2430 is “good morning. I will do my best today”.

  Case (B) is a diagram showing a manner in which the wording changes according to the familiarity. In one aspect, the spoken dialogue system 2200 responds to the news question from the user according to the familiarity that has been set: “The government is planning to build a new base in Japan in Antarctica. It seems to have become clear, "he utters (utterance 2440). At this point, since the intimacy is set to be low, the tone of the utterance is also a relatively polite tone.

  In a certain aspect, the dialogue generation unit 2280 converts the expression that conveys the content of the utterance 2440 according to the familiarity of the user who is going to interact with the voice dialogue system 2200 without changing the content of the utterance 2440. To do. For example, when it is detected that the familiarity of the authenticated user is higher than a preset standard, the dialogue generation unit 2280 phrase “is doing” the phrase that constitutes the utterance content. Change to "tega". In addition, the dialogue generation unit 2280 changes the phrase “obtained” to the phrase “understood”. That is, the dialogue generation unit 2280 changes the expression of the utterance content that can be output by default to a simple expression based on the determination result that the familiarity is higher than the standard. Furthermore, the dialog generation unit 2280 converts the phrase “Yes” to the phrase “Yes”. That is, the dialogue generation unit 2280 converts a phrase to be uttered from a polite tone into a normal tone.

  When such a conversion rule is defined in the dialog generation unit 2280, the dialog generation unit 2280 responds with a phrase having a high familiarity as the response phrase 2150. That is, the dialogue generator 2280 translates the phrase “It seems that the government is planning to build a new Japanese base in Antarctica”. The communication terminal 200 can output the response phrase 2450 generated based on the conversion.

[Effect of the embodiment]
As described above, the voice interaction system 2200 according to the present embodiment calculates the familiarity according to the history of conversations with each user so far, and changes the tone of the utterance according to the familiarity. As a result, the user feels more familiar with the communication terminal 200 constituting the voice interaction system 2200.

[Fourth Embodiment]
In the first to third embodiments described above, a case where the voice input / output function (communication terminal) and the speech generation function (server) are realized by separate devices is illustrated. However, the technical idea according to the present disclosure can be realized by other device configurations. For example, the voice input / output function and the speech generation function may be realized by a single device. For example, a device in which the communication terminal 200 and the server 220 shown in FIG. 2 are integrated may be realized as a voice interaction device.

  Therefore, with reference to FIG. 25, a voice interactive apparatus 2500 according to the fourth embodiment will be described. FIG. 25 is a block diagram showing an outline of the configuration of the voice interaction apparatus 2500. The voice interactive apparatus 2500 has a configuration provided in the communication terminal 200 and the server 220 shown in FIG. With this configuration, the voice interaction apparatus 2500 can perform voice recognition, voice authentication, and utterance generation based on a user's utterance without using a communication line, and thus is not affected by the communication line. Quick conversation is possible.

[Fifth Embodiment]
In still another aspect, voice recognition and voice authentication may be performed by a communication terminal that accepts an utterance from a user. In this case, the communication terminal transmits the voice recognition result and the voice authentication result to the server. The server uses each result to generate a response to the utterance.

  Therefore, with reference to FIG. 26, the configuration of a voice interaction system 2600 according to this aspect will be described. FIG. 26 is a block diagram showing an outline of the configuration of the voice interaction system 2600. The voice interaction system 2600 includes a communication terminal 2610 and a server 2620.

  The communication terminal 2610 includes a voice recognition unit 240 in addition to the configuration of the communication terminal 200. The voice recognition unit 240 includes a voice recognition module 241 and a speaker identification module 242. The speech recognition module 241 executes speech signal recognition processing accepted by the speech input unit 210. The speaker specifying module 242 specifies a speaker using the voice signal, voice data and user identification information registered in a memory (not shown) of the communication terminal 2610.

  The voice content recognized by the voice recognition unit 240 and the specified user information are transmitted to the server 2620. The server 2620 uses the audio content and user information to generate a response corresponding to the audio content while referring to the user's past conversation history.

  Other processes are the same as the processes in the voice interaction system according to the above-described embodiment. Therefore, detailed description will not be repeated.

[Sixth Embodiment]
The servers according to the first to third embodiments are configured to realize a speech recognition function, a dialog generation function, and a speech synthesis function. In other aspects, each function may be implemented in a separate computer device.

[Seventh Embodiment]
Each of the above-described embodiments is exemplified as being realized by a processor (not shown) of a computer included in a communication terminal or server executing instructions included in a program stored in a memory. However, at least a part or all of the functions included in the communication terminal or the server according to the present embodiment may be realized by a circuit or other hardware that realizes the function.

<Configuration>
The configuration according to the present disclosure can be summarized as follows.

[Configuration 1]
A voice recognition unit (241) configured to recognize an utterance;
A voice authentication unit (242) configured to identify a speaker based on the recognized utterance and pre-registered user information;
A topic estimation unit (280) configured to generate a topic in which the speaker is interested based on the recognized utterance and the identified speaker;
A voice interactive device (2500) comprising a voice output unit (211) configured to output the topic in a voice.

[Configuration 2]
An audio signal input unit configured to accept an input of an audio signal based on an utterance;
A voice recognition unit (241) configured to recognize the utterance based on the input voice signal;
A voice authentication unit (242) configured to identify a speaker based on the input voice signal and user information registered in advance;
A topic estimation unit (280) configured to generate a topic in which the speaker is interested based on the recognized utterance and the identified speaker;
A voice interaction device (220) comprising: an output unit (230) configured to output a topic signal for outputting the topic in voice.

[Configuration 3]
A terminal (2610);
A server (2620) capable of communicating with the terminal,
The terminal
A voice recognition unit (241) configured to accept an utterance and recognize the utterance;
A voice authentication unit (242) configured to identify a speaker based on the recognized utterance and pre-registered user information;
A transmission unit configured to transmit an audio signal based on the utterance and an identification signal of the identified speaker to the server;
The server
A topic estimation unit configured to generate a topic in which the speaker is interested based on the voice signal and the identification signal;
A topic transmission unit configured to transmit a topic signal for outputting the topic as a voice to the terminal,
The terminal
A spoken dialogue system comprising an output unit configured to output the topic by voice based on the topic signal received from the server.

[Configuration 4]
A terminal used in the voice interaction system according to Configuration 3,
A speech recognizer configured to recognize utterances;
A voice authentication unit configured to identify a speaker based on the recognized utterance and pre-registered user information;
A transmitter configured to transmit an audio signal based on the utterance and an identification signal of the identified speaker to a server;
A terminal comprising: an output unit configured to receive a topic signal for outputting a topic of interest of the speaker by voice from the server and output the topic by voice.

[Configuration 5]
A storage unit configured to store a history of interaction with each user of the voice interaction device;
The speech conversation apparatus according to Configuration 1 or 2, wherein the topic estimation unit is configured to generate the topic based on a history of conversations with the user.

[Configuration 6]
A closeness calculation unit configured to calculate a closeness between the user and the voice interaction device based on a history of interaction with the user of the voice interaction device;
The spoken dialogue apparatus according to Configuration 1 or 2, wherein the topic estimation unit is configured to adjust the tone of the topic according to the familiarity.

[Configuration 7]
Recognizing the utterance,
Identifying a speaker based on the recognized utterance and pre-registered user information;
Generating a topic in which the speaker is interested based on the recognized utterance and the identified speaker;
A voice dialogue method including the step of outputting the topic by voice.

[Configuration 8]
Receiving an input of an audio signal based on an utterance;
Recognizing the utterance based on the input audio signal;
Identifying a speaker based on the input audio signal and pre-registered user information;
Generating a topic in which the speaker is interested based on the recognized utterance and the identified speaker;
Outputting a topic signal for outputting the topic by voice.

[Configuration 9]
Recognizing the utterance,
Identifying a speaker based on the recognized utterance and pre-registered user information;
Transmitting an audio signal based on the utterance and an identification signal of the identified speaker to a server;
Receiving from the server a topic signal for outputting a topic of interest that the speaker is interested, estimated based on the voice signal and the identification signal;
A voice dialogue method including the step of outputting the topic by voice based on the topic signal.

[Configuration 10]
A program for causing a computer to function as a voice interaction device, wherein the program is stored in one or more processors,
Recognizing the utterance,
Identifying a speaker based on the recognized utterance and pre-registered user information;
Generating a topic signal for outputting a topic in which the speaker is interested based on the recognized utterance and the identified speaker;
A program for executing the step of outputting the topic by voice based on the topic signal.

[Configuration 11]
A program for causing a computer to function as a voice interaction device, wherein the program is stored in one or more processors,
Receiving an input of an audio signal based on an utterance;
Recognizing the utterance based on the input audio signal;
Identifying a speaker based on the input audio signal and pre-registered user information;
Generating a topic in which the speaker is interested based on the recognized utterance and the identified speaker;
A program for executing a step of outputting a topic signal for outputting the topic by voice.

[Configuration 12]
A program for causing a computer to function as a voice interaction device, wherein the program is stored in one or more processors,
Recognizing the utterance,
Identifying a speaker based on the recognized utterance and pre-registered user information;
Transmitting an audio signal based on the utterance and an identification signal of the identified speaker to a server;
Receiving from the server a topic signal for outputting a topic of interest that the speaker is interested, estimated based on the voice signal and the identification signal;
A program for executing the step of outputting the topic by voice based on the topic signal.

[Configuration 13]
A terminal,
With a server,
The terminal includes a voice recognition unit configured to recognize an utterance;
An audio signal converter configured to convert a recognized utterance into an utterance signal;
A transmitter configured to transmit the speech signal to the server,
The server
A speech recognition unit configured to recognize the utterance based on the utterance signal received from the terminal;
A voice authentication unit configured to identify a speaker based on the utterance signal and user information registered in advance;
A topic estimation unit configured to generate a topic in which the speaker is interested based on the recognized utterance and the identified speaker;
A transmission unit configured to transmit a topic signal for outputting the topic as a voice to the terminal,
The terminal
A receiver configured to receive the topic signal from the server;
And a voice dialogue system including an output unit configured to output the topic in a voice based on the topic signal.

[Configuration 14]
A terminal used in the system according to Configuration 13,
A speech recognizer configured to recognize utterances;
An audio signal converter configured to convert the recognized utterance into an utterance signal;
A transmitter configured to transmit the utterance signal to a server;
A receiver configured to receive from the server a topic signal generated based on the utterance signal;
A terminal comprising: an output unit configured to output a topic corresponding to the utterance by voice based on the topic signal.

  The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

  110,241 Voice recognition module, 120 Voice authentication module, 130,1690 Topic estimation module, 140 Dialogue generation module, 200,500,1200,2610 Communication terminal, 210 Voice input part, 211 Voice output part, 220,520,820, 920, 1620, 2220, 2620 Server, 230, 530 Control unit, 240, 540 Speech recognition unit, 242 Speaker identification module, 250, 550, 850, 2250 Dialog analysis unit, 260, 560, 1660 Dialog history storage unit, 280 , 580, 880, 980, 1680, 2280 Dialogue generation unit, 290, 590 Speech synthesis unit, 400 computer, 501 mobile phone, 510 RFID tag information transmission unit, 541 user identifier unit, 990 interest estimation module, 1110, 2110 Input phrase, 1120, 1240 Interest noun, 1130, 2120 Output phrase, 1200, 1300, 1800, 1900, 2000, 2400 Table, 1230, 1930 Speaker, 1250, 1960 Time stamp, 1501 User.

Claims (14)

A speech recognizer configured to recognize utterances;
A voice authenticator configured to identify a speaker based on the recognized utterance and pre-registered user information;
A topic estimation unit configured to generate a topic in which the speaker is interested based on the recognized utterance and the identified speaker;
A voice dialogue apparatus comprising: a voice output unit configured to output the topic in voice. An audio signal input unit configured to accept an input of an audio signal based on an utterance;
A speech recognition unit configured to recognize the utterance based on the input speech signal;
A voice authentication unit configured to identify a speaker based on the input voice signal and pre-registered user information;
A topic estimation unit configured to generate a topic in which the speaker is interested based on the recognized utterance and the identified speaker;
An audio dialogue apparatus comprising: an output unit configured to output a topic signal for outputting the topic in audio. A terminal,
A server capable of communicating with the terminal,
The terminal
A voice recognition unit configured to accept an utterance and recognize the utterance;
A voice authenticator configured to identify a speaker based on the recognized utterance and pre-registered user information;
A transmission unit configured to transmit an audio signal based on the utterance and an identification signal of the identified speaker to the server;
The server
A topic estimation unit configured to generate a topic in which the speaker is interested based on the voice signal and the identification signal;
A topic transmission unit configured to transmit a topic signal for outputting the topic by voice to the terminal;
The terminal further includes
A spoken dialogue system comprising: an output unit configured to output the topic by voice based on the topic signal received from the server. A terminal used in the voice interaction system according to claim 3,
A speech recognizer configured to recognize utterances;
A voice authenticator configured to identify a speaker based on the recognized utterance and pre-registered user information;
A transmission unit configured to transmit an audio signal based on the utterance and an identification signal of the identified speaker to a server;
A terminal comprising: an output unit configured to receive a topic signal for outputting a topic of interest of the speaker by voice from the server and output the topic by voice. A storage unit configured to store a history of interaction with each user of the voice interaction device;
The spoken dialogue apparatus according to claim 1, wherein the topic estimation unit is configured to generate the topic based on a history of dialogue with the user. A closeness calculation unit configured to calculate a closeness between the user and the voice interaction device based on a history of interaction with the user of the voice interaction device;
The spoken dialogue apparatus according to claim 1, wherein the topic estimation unit is configured to adjust a tone of the topic according to the familiarity. Recognizing the utterance,
Identifying a speaker based on the recognized utterance and pre-registered user information;
Generating a topic that the speaker is interested in based on the recognized utterance and the identified speaker;
A voice dialogue method including the step of outputting the topic by voice. Receiving an input of an audio signal based on an utterance;
Recognizing the utterance based on the input audio signal;
Identifying a speaker based on the input audio signal and pre-registered user information;
Generating a topic that the speaker is interested in based on the recognized utterance and the identified speaker;
Outputting a topic signal for outputting the topic by voice. Recognizing the utterance,
Identifying a speaker based on the recognized utterance and pre-registered user information;
Transmitting an audio signal based on the utterance and an identification signal of the identified speaker to a server;
Receiving, from the server, a topic signal for outputting a topic of interest of the speaker, which is estimated based on the voice signal and the identification signal;
And outputting the topic in a voice based on the topic signal. A program for causing a computer to function as a voice interaction device, wherein the program is stored in one or more processors,
Recognizing the utterance,
Identifying a speaker based on the recognized utterance and pre-registered user information;
Generating a topic signal for outputting a topic in which the speaker is interested based on the recognized utterance and the identified speaker;
A program for executing the step of outputting the topic by voice based on the topic signal. A program for causing a computer to function as a voice interaction device, wherein the program is stored in one or more processors,
Receiving an input of an audio signal based on an utterance;
Recognizing the utterance based on the input audio signal;
Identifying a speaker based on the input audio signal and pre-registered user information;
Generating a topic that the speaker is interested in based on the recognized utterance and the identified speaker;
A program for executing a step of outputting a topic signal for outputting the topic by voice. A program for causing a computer to function as a voice interaction device, wherein the program is stored in one or more processors,
Recognizing the utterance,
Identifying a speaker based on the recognized utterance and pre-registered user information;
Transmitting an audio signal based on the utterance and an identification signal of the identified speaker to a server;
Receiving, from the server, a topic signal for outputting a topic of interest of the speaker, which is estimated based on the voice signal and the identification signal;
A program for executing the step of outputting the topic by voice based on the topic signal. A terminal,
With a server,
The terminal includes a voice recognition unit configured to recognize an utterance;
An audio signal converter configured to convert a recognized utterance into an utterance signal;
A transmitter configured to transmit the speech signal to the server,
The server
A speech recognition unit configured to recognize the utterance based on the utterance signal received from the terminal;
A voice authentication unit configured to identify a speaker based on the speech signal and pre-registered user information;
A topic estimation unit configured to generate a topic in which the speaker is interested based on the recognized utterance and the identified speaker;
A transmission unit configured to transmit to the terminal a topic signal for outputting the topic by voice,
The terminal further includes
A receiver configured to receive the topic signal from the server;
And a voice dialogue system including an output unit configured to output the topic by voice based on the topic signal. A terminal used in the system according to claim 13,
A speech recognizer configured to recognize utterances;
An audio signal converter configured to convert the recognized utterance into an utterance signal;
A transmitter configured to transmit the speech signal to a server;
A receiving unit configured to receive a topic signal generated based on the utterance signal from the server;
A terminal comprising: an output unit configured to output a topic corresponding to the utterance by voice based on the topic signal.

Images