JP2003108362A - Communication supporting device and system thereof - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a communication supporting device that prevents the reduction of usage frequency by eliminating monotony in a dialog to induce user's interest in the dialog. SOLUTION: A dialog processing means 20 interacts with a user by a natural language via a voice input 11 and a voice output 12. A character assumed as a respondent who responds to the user can be selected from a plurality of types with a dialog processing means 20, and an appropriate character is selected according to the interaction environment with an overall control unit 10. The character, depending on the tone and accent, discriminates the respondents.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a communication support device.

[0002]

2. Description of the Related Art Generally, an elderly person has few opportunities to go out and has little conversation with other persons, and therefore his family is often the other party. In addition, elderly people often suffer from accidents at home due to weakness in their physical strength, and it is necessary for the family to constantly watch over them in order to prevent accidents and to promptly deal with any accidents. Under such circumstances, it is difficult for a family living with an elderly person to leave the elderly person at home and go out, and it is often difficult to go shopping or even learn lessons, not to mention traveling.

On the other hand, recently, a type of robot called a pet robot, which behaves as if it had its own intention, has been commercialized. As this type of robot, for example, a dog type pet robot is called "AIBO" ( It is commercialized under the name Sony's product name). By monitoring and learning the tone, facial expression, and touching method of a human voice, this robot does not formulate the response to stimuli like conventional mechanical devices, and the response to stimuli changes depending on the situation. It becomes possible to behave as if they are growing up, and it is possible to make them illusion as if they are communicating with a person, and it has a kind of healing function. However, this pet robot aims at entertainment by imitating the movement of a dog, and is not intended to support communication with the elderly.

On the other hand, cat-shaped or bear-shaped robots have been proposed by Matsushita Electric Industrial Co., Ltd. as pet robots whose main purpose is to communicate with the elderly. To contribute, in addition to being able to perform simple daily conversations and behaving like a pet, it also has the function of indirectly grasping the usage status of the pet robot from a remote location. Therefore, this pet robot can serve as a conversation partner for the elderly person, mentally care for the elderly person, and confirm the safety of the elderly person's living alone from a remote location. This pet robot is connected to a center facility installed in a welfare service support center via a communication path (a communication path that combines digital communication system technology and mobile communication technology in a two-way CATV facility), When confirming the safety of the elderly, the pet robot transmits a message from the center equipment to the pet robot through the communication path, and the pet robot gives the message to the elderly by voice. The response of the elderly at this time is returned to the center facility, and the reasoning device provided in the center facility infers the living situation to confirm the safety of the elderly. The reasoning device infers that an abnormality may have occurred when, for example, the elderly person does not respond to the speech of the pet robot.

[0005]

Since the former pet robot described above does not have an interactive function, it is not possible to sufficiently alleviate the feeling of loneliness. The latter pet robot cannot communicate with the center facility, which is a public facility, via a communication path. It is not provided with the function of transferring the user's terminal 2b to the center equipment so that privacy is not infringed because the connection is established.

[0006] Therefore, the present inventors alleviate the feeling of loneliness of the user by providing an interactive function for simulating the interaction with a person, and can transfer the image of the user to the portable terminal within the range where privacy is not violated. As a system, we previously proposed a system that enables the user to see whether there is any abnormality (Japanese Patent Application No. 200
0-393678). However, the current technology of the dialogue function can only make a mechanical response, and the dialogue tends to be monotonous, so the freshness fades after several uses, and the frequency of use decreases and mental care functions are reduced. It has a problem that it cannot be fully exhibited.

The present invention has been made in view of the above circumstances, and an object thereof is to provide a communication support device in which the frequency of use is prevented from decreasing by eliminating the monotonous nature of the dialog and making the dialog interesting. To do.

[0008]

According to a first aspect of the present invention, a voice input section and a voice output section are connected to each other, a dialogue by a voice using a natural language is carried out with a user, and a response is made to respond to the user. A dialogue processing means capable of selecting a character assumed as a person from a plurality of types, and a central control unit having a character selecting means for selecting an appropriate character from the characters in the dialogue processing means according to a dialogue environment. Characterize. With this configuration, it is possible to eliminate the monotonousness of the dialogue by selecting a plurality of types of characters, and it is possible to keep the dialogue interesting and prevent the frequency of use from decreasing. can do.

According to a second aspect of the invention, in the first aspect of the invention, since the voice color is included as an attribute of the character, only a relatively simple structure for controlling the voice color is adopted.
It is possible to reduce the monotonousness of the dialogue by giving the impression of talking to multiple people.

According to a third aspect of the present invention, in the first aspect of the present invention, a driving source for driving a robot having a movable portion expressing an expression is added, and the character selecting means is expressed by the movement of the movable portion. The character is selected according to. According to this configuration, since the character is expressed by the motion, it is possible to reduce the monotonousness caused only by the dialogue by attracting the interest by the motion.

According to a fourth aspect of the present invention, in the first aspect of the present invention, since the operation unit for selecting the character is provided, the character can be selected according to the wish of the user.
By allowing the user to select a favorite character, it becomes possible to maintain the interest in the dialogue.

According to a fifth aspect of the present invention, in the first aspect of the present invention, there is a function of calling out to the user through the dialogue processing means so as to prompt the user to speak, and the character selecting means responds to the content of the user's response. And selecting the character. With this configuration, since the character is selected according to the utterance content of the user, the character selection is performed interactively, and it is possible to prevent monotonousness due to the formalization of the character selection.

According to a sixth aspect of the present invention, in the first aspect of the present invention, user information input means for inputting user information about the user is added, and the character selecting means is operable to generate the character according to the user information. Is selected. According to this configuration, since the character is selected based on the user information provided by the user, the character suitable for the user is selected.

According to a seventh aspect of the invention, in the sixth aspect of the invention, the user information input means stores the user information and is carried by the user, and the user information is input from the previous user information input means. It is characterized by comprising a signal receiving unit for receiving and delivering to the overall control unit. According to this configuration, since the user information is given from the user information input means possessed by the user, it is possible to easily input even user information such as biometric information of the user whose contents change from time to time.

According to the invention of claim 8, in the invention of claim 7, since the transmission path between the user information input means and the signal receiving section is a wireless transmission path, it can be used even if the user is away from the user. It becomes possible to input personal information.

According to a ninth aspect of the invention, in the seventh aspect of the invention, since at least a part of the transmission line between the user information input means and the signal receiving section is a human body, it can be touched by a user. It becomes possible to input user information.

According to a tenth aspect of the invention, in the invention of the sixth to ninth aspects, the user information is the age of the user.
Since the gender is included as the unique information, it is possible to select the character according to the age and gender of the user.

According to the invention of claim 11, in the invention of claims 6 to 9, since the user information includes biometric information of the user, the health condition of the user can be managed.

According to a twelfth aspect of the present invention, in the sixth to ninth aspects of the invention, the user information includes identification information for identifying the user, so that the user can be identified.

According to a thirteenth aspect of the present invention, in the inventions of the sixth to ninth aspects, the user information includes system setting information required when connecting to an external device connected via a communication path. You can easily set the system setting information required to connect to an external device through the communication path, and even if you lose this type of system setting information, you can easily reset it using the user information input means. become.

According to a fourteenth aspect of the present invention, in the first to thirteenth aspects of the present invention, the character selecting means is capable of selecting a character which is supposed to be a responder and a character which makes a relationship during the dialogue. To do.
According to this configuration, the user's utterance is urged due to the fact that the user is accused during the dialogue, and as a result, the decrease in the frequency of use can be reduced.

According to a fifteenth aspect of the present invention, in the first to fourteenth aspects, when the character selection means obtains a plurality of recognition candidates for a recognition result for a voice input of a user, the recognition candidates are respectively recognized. Different characters are assigned, and the dialogue processing unit has a function of asking the user to confirm the recognition candidate and having a function of determining the content of the voice input by the user when a positive response from the user is received. To do. According to this configuration, it is possible to increase the voice recognition rate by using the acoustic model suitable for the user's voice, and since it responds to the question, the interest in the dialogue is maintained, It is possible to reduce the decrease in the frequency of use by the user.

According to a sixteenth aspect of the present invention, in the fifteenth aspect of the present invention, the dialogue processing means is provided with a voice recognizing section for recognizing the utterance content of the user, and the voice recognizing section is collated with a voice feature amount series. A plurality of acoustic models are provided and a character is assigned to each acoustic model, and a character suitable for the voice of the user is extracted according to the result of confirmation of the recognition candidate, and an acoustic model corresponding to the extracted character is extracted. It is characterized by being used in a voice recognition unit. According to this configuration, it is possible to narrow down the acoustic model suitable for the user from a plurality of acoustic models through the interaction with the user, so that an appropriate acoustic model is automatically selected, and as a result, it is used. The recognition rate of the person's voice becomes high.

The invention according to claim 17 comprises the communication support device according to any one of claims 1 to 16 and a voice database in which voices of a plurality of contents are registered for a plurality of the characters. The dialogue processing means corresponds to the response sentence generation unit that generates the content of the voice that responds to the user, the character that is selected by the character selection unit, and the content of the voice that is generated by the response sentence generation unit. And a voice search unit that searches the voice database for a voice of the content, and when the voice search unit extracts the relevant voice from the voice database, the voice is used for a dialogue with the user. According to this configuration, since the voice database is provided, the family of the user can be set as a character, and when the user is an elderly person living alone, a conversation can be performed using the voice of the family. This can give the user a sense of security.

According to an eighteenth aspect of the present invention, in the seventeenth aspect of the present invention, the voice database includes a voice communication section that enables communication with a voice terminal capable of voice input, and a voice terminal through the voice communication section. A registration processing unit for registering an input voice is additionally provided. According to this configuration, any voice can be registered in the voice database through the voice terminal, so that, for example, the family of the user can easily register the voice in the voice database.

The invention of claim 19 is characterized in that, in the invention of claim 18, the voice terminal is a telephone terminal. According to this configuration, since the telephone terminal is used as the voice terminal, it is possible to easily register the voice in the voice database using the mobile phone or the fixed telephone.

According to a twentieth aspect of the invention, in the eighteenth or nineteenth aspect of the invention, the registration processing unit has a function of instructing the voice terminal about the content of the voice to be registered in the voice database. And According to this configuration, since the content of the voice to be registered in the voice database is instructed from the registration processing unit, it is possible to easily register the voice by registering the voice in accordance with the instruction and also to perform the dialogue in the communication support device. By previously instructing the content of the voice used for, the possibility that the voice stored in the voice database can be used for the dialogue becomes high.

According to a twenty-first aspect of the present invention, in the eighteenth or nineteenth aspect of the invention, the voice terminal includes a display section for displaying an image, and the registration processing section registers the voice terminal with the voice database. It is characterized in that it has a function of displaying the contents of the voice to be displayed on the display unit in a menu format. According to this configuration, the content of the voice to be registered in the voice database is instructed in the menu format from the registration processing unit, so that the content of the instruction is easy to understand, and the voice is easily registered by registering the voice according to the instruction. In addition to being able to perform, by instructing in advance the content of the voice used in the dialogue in the communication support device,
The voice stored in the voice database is more likely to be available for dialogue.

A twenty-second aspect of the present invention is the user terminal according to any one of the seventeenth to twenty-first aspects, wherein the communication support device interacts with the user through the voice input unit, the voice output unit, and the dialogue processing unit. In preparation for the above, the voice database is provided in a server capable of data communication with a user terminal via a communication path, and the voice search unit corresponds to the character when the character is selected by the character selecting means. When the voice extracted from the voice database is included in the voice extracted from the voice database when the voice having the content corresponding to the content of the voice generated by the response sentence generation unit is included in the voice extracted from the voice database at once. It is characterized by being used for the dialogue of. According to this configuration, since the voice database is provided in the server, protection and maintenance of the voice database can be surely performed by monitoring the voice database by a server administrator who is an expert.

The invention of claim 23 is characterized in that, in the invention of claims 17 to 21, the voice database is provided at the same place as the communication support device. According to this configuration, a server is not required, so that the system configuration is simplified.

According to a twenty-fourth aspect of the present invention based on the seventeenth to twenty-first aspects, the communication support device is provided with a user terminal that interacts with the voice input unit and the voice output unit, and The interactive processing means is provided in a server capable of data communication with the user terminal through a communication path, and the voice database is provided in the server. According to this configuration, since the voice database and the dialogue processing means are provided in the server, the voice database and the dialogue processing means are protected by monitoring the voice database and the dialogue processing means by a server administrator who is an expert. And maintenance will be performed reliably.

According to a twenty-fifth aspect of the present invention, the server according to the twenty-second aspect or the twenty-fourth aspect, wherein the server registers a history of a dialogue in which the communication support device has a dialogue with the user, and a server. And a billing processing unit that performs billing processing when the content of the dialogue record storage unit is read from a terminal connected via a communication path. According to this configuration, since the history of the dialogue is registered in the dialogue record storage unit and the history of the dialogue can be read out through the communication path, it can be known from the outside whether or not the dialogue by the user is normally performed. It will be possible. For example, when the user is an elderly person living alone, the safety of the elderly person living alone can be confirmed even if the family cannot directly communicate with the elderly person living alone by reading the conversation record. In addition, since the fee is charged when reading the dialogue record storage unit, the contents of the dialogue record storage unit are not unnecessarily read, and it is possible to suppress an increase in the traffic of the server and generate the operating cost of the server.

[0033]

BEST MODE FOR CARRYING OUT THE INVENTION (First Embodiment) As shown in FIG. 1, the present embodiment is a communication support device arranged in a house where an elderly person living alone and a sick person (hereinafter referred to as a user) are present. 1 shows an example in which data transmission is possible with the mobile terminal 2a or the terminal 2b composed of a personal computer via a communication path 3 including a network such as the Internet.

The communication support device 1 has a voice interactive function, a photographing function for photographing the surroundings, and a communication function for transmitting data including images and voice. In the present embodiment, the communication support device 1 is described as a robot having a dog-like appearance, but any appearance is acceptable as long as the robot can express facial expressions. Further, it is assumed that the mobile terminal 2a is a mobile phone having an image display function, and an image / sound server 4 for temporarily storing image data and sound data is connected to the communication path.

The communication support device 1 is provided with a voice input unit 11 in which an amplification circuit is added to a microphone as an ear and a voice output unit 12 in which an amplification circuit is added to a speaker as a mouth in order to perform a dialogue by voice. Further, in order to take a picture of the surroundings, an image pickup unit 13 including a small TV camera (a CCD image pickup element or a CMOS image pickup element in which a light receiving optical system and an image signal amplification circuit are combined) is provided as an eye.

The voice input unit 11 and the voice output unit 12 are connected to the dialogue processing unit 20, and the dialogue processing unit 20 enables communication with the user in natural language through the voice input unit 11 and the voice output unit 12. Also,
The imaging unit 13 and the audio input unit 11 are the AV signal processing unit 14.
The image signal photographed by the image pickup unit 13 and the audio signal obtained from the audio input unit 11 are connected to the digital camera, and are converted into digital signals and data compressed. AV signal processing unit 1
The image data and the audio data obtained by compressing the image signal and the audio signal in 4 are sent to the communication path 3 through the communication means 15, and the communication means 15 uses the predetermined communication protocol to convert the image data and the audio data into the image and sound server 4. Transfer to. The operation of each unit of the communication support apparatus 1 is managed by the general control unit 10.

Further, in the communication support device 1, parts corresponding to a neck, an ear, an arm, a mouth, etc. are movable in order to express a facial expression, and the movable part is moved by a driving part 31 provided in the driving means 30. Is given. The drive unit 31 includes a DC motor and gears, and is housed in the housing of the communication support apparatus 1 near each movable part.
The drive unit 30 is also provided with a drive unit control unit 32 for controlling the DC motor provided in the drive unit 31 and a drive unit control sensor unit 33 such as an encoder for detecting the rotation angle and the rotation direction of the DC motor. The drive unit control unit 32 feedback-controls the drive unit 31 using the instruction from the overall control unit 10 and the detection value of the drive unit control sensor unit 33. The communication support device 1 also includes a sensor unit 17 that detects a surrounding situation. As the sensor unit 17, a contact sensor that detects that the user has touched the communication support apparatus 1, a distance measurement sensor that detects the approach of the user, and an illuminance sensor that detects the ambient brightness are provided.

The overall control unit 10 detects the surrounding situation (contact or approach to the communication support device 1, change in ambient brightness, etc.) obtained from the sensor unit 17 and the input sound obtained from the AV signal processing unit 14. From the user based on the feature amount (such as the sound pressure level of the user's speech), the feature amount of the captured image (such as the distinction between a person and an artifact due to the difference from the background image), and the voice recognition in the dialogue processing means 20. Information (such as estimation of a speaker's intention by analyzing a keyword) and the internal state of the communication support device 1 ascertained by the overall control unit 10, a character to be described later is selected, and preset control is performed. The control command and the accompanying control data are transmitted to the drive unit control unit 32, the AV signal processing unit 14, the dialogue processing unit 20, the communication unit 15, the disk unit as if they were executed. Ray unit 16, voice input unit 11, audio output unit 12, controls the imaging unit 13.

That is, in the overall control unit 10, by instructing the drive unit control unit 32 about the region where the drive unit 31 is arranged and the direction / position / time of the operation, each unit of the communication support apparatus 1 is controlled. To perform the desired action. In addition, for the AV signal processing unit 14, the imaging unit 13
Instructing to turn on / off the acquisition of the image signal from the device, and also instructing to turn on / off the power source of the imaging unit 13. Furthermore, the dialogue control unit 23 is instructed to speak the content and voice color corresponding to the selected character, and the voice output unit 12 controls the amplification factor according to the selected character. Here, since the sound pressure level due to the user's utterance input to the voice input unit 11 is not fixed, the integrated control unit 10 outputs the voice signal so that the voice signal input to the AV signal processing unit 14 becomes an appropriate level. The amplification factor is also instructed to the input unit 11. In addition, the overall control unit 10 also gives an instruction to control the communication on / off to the communication unit 15 and an instruction to turn on / off the power to the display unit 16.

The image / sound server 4 temporarily stores image data and sound data. The image data and the audio data accumulated in the image / audio server 4 are transmitted from the mobile terminal 2a or the terminal 2b via the communication path 3 to the image / audio server 4
You can browse by accessing. Here, it is assumed that the mobile terminal 2a and the terminal 2b can access the video / audio server 4 through the Internet connection service.

The image / sound server 4 includes a conversion means 4a for converting the size and color tone of the image data into a format that can be displayed on the screen of the portable terminal 2a. That is, the mobile terminal 2
When browsing the image data stored in the image and sound server 4 by a, the image stored in the image and sound server 4 is converted into a format that can be displayed on the screen of the mobile terminal 2a, and the image is captured by the communication support apparatus 1. The image can be confirmed by the mobile terminal 2a. It is desirable to attach the shooting date and time to the image data, and the user's situation at the shooting date and time can be confirmed using the mobile terminal 2a or the terminal 2b even if the user's family is far away. .

On the other hand, when the user's family sends a message to the user, at least one of the image data and the audio data is sent to the image / audio server 4 using the mobile terminal 2a or the terminal 2b and temporarily sent. Accumulate. Here, the image data and the audio data are stored in a data compressed form. When the communication support apparatus 1 accesses the image / sound server 4, if there is image data or sound data addressed to the communication support apparatus 1, this is acquired via the communication path 3 and the dialogue processing unit 2
At 0, the voice output unit 12 utters a message such as "I have heard from my family." Here, the acquired image data or audio data is decompressed by the AV signal processing unit 14 and converted into an analog signal,
The image signal corresponding to the image data is displayed on the display unit 16, and the audio signal corresponding to the audio data is output from the audio output unit 12. With this function, the user can see and hear the faces and voices of the family using the communication support device 1 even if the family is far away.

The access right to the image data and voice data stored in the image / voice server 4 is set to the mobile terminal so that the image data / voice data stored in the image / voice server 4 may not be viewed by others. 2a, terminal 2b,
It is restricted according to the identification information from the communication support device 1.

By the way, the communication support device 1
In the above, there is a possibility that the privacy of the user may be impaired depending on the timing of photographing by the image pickup unit 13. Therefore, it is configured so that the user can determine the timing of shooting with his or her own will. However, since the user of the communication support device 1 is often an elderly person or a sick person, there is a high possibility that it will be cumbersome to stop operating if a configuration for manually instructing the timing of imaging is adopted. It is desirable to indicate the timing of shooting by a specific movement. As an example, in the present embodiment, the fact that the imaging unit 13 is provided in the eye area of the robot is used to instruct the photographing by opening and closing the eyelid of the user. That is, the user is the communication support device 1
If you open the eyelid after closing the eyelid by looking at the part corresponding to
The voice input unit 11 starts capturing the voice and the image capturing unit 13 starts shooting, and the storage unit (not shown) stores the voice data and the image data. After that, when the eyelids are closed for a certain time or longer, the capturing of the sound and the image is stopped. Considering the communication speed on the communication path, it is desirable to define a time limit for capturing audio and images. That is, even if the eyelids are not closed, the capturing of the sound and the image is automatically stopped at the time limit.
This control is performed by the integrated control unit 10. With this configuration, it is possible to automatically stop capturing voice and images even when the user leaves the communication support apparatus 1 without instructing to stop. Moreover, since the user intentionally indicates the timing of capturing the voice and the image, privacy is not impaired.

By the way, the dialogue processing means 20 basically generates a voice recognition section 21 for recognizing the user's voice input through the voice input section 11 and a synthesized voice to be output from the voice output section 12. The speech synthesis unit 22 and the dialogue control unit 23 that analyzes the meaning of the speech recognized by the speech recognition unit 21 and instructs the speech synthesis unit 22 about the content to be responded to. More specifically, as shown in FIG. 2, the voice recognition unit 21 includes a sound processing unit 21a and a language decoding unit 2.
1b, and the audio signal input through the audio input unit 11 is A / D processed by the audio processing unit 21a.
After the conversion is performed, spectrum analysis is performed by a method such as FFT. The spectrum-analyzed speech waveform is encoded into a feature quantity series that is time series data. The language decoding unit 21b uses the acoustic model indicating the appearance probability of the feature amount sequence of the voice waveform when the prepared word sequence is uttered, and the language model which is a database of the appearance probability of the word sequence, to perform the acoustic processing unit 21a. Estimate the word sequence from the output feature sequence of the speech. That is, the voice recognition unit 21 uses a statistical voice recognition technique.

On the other hand, the dialogue control section 23 includes the language decoding section 21.
The dialog management unit 23a is provided to which the word sequence from b is input. The dialogue management unit 23a includes an inference engine, and uses a natural language processing technique that matches an input word sequence with a rule (that is, knowledge) stored in the dialogue rule base 23c, and performs syntactic analysis and keyword extraction of the word sequence. And then
Based on the result, semantic analysis and context analysis are performed. That is, the meaning of the utterance by the user and the intention of the user are analyzed and expressed in a predetermined format, and the response content is generated according to the analysis result of the meaning of the utterance and the user's intention to generate the utterance content in the predetermined format. It is delivered to the section 23b. Here, the predetermined format is a format of semantic expression. The utterance content generation unit 23b generates a word sequence according to the response content passed from the dialogue management unit 23a and delivers it to the voice synthesis unit 22. The voice synthesizing unit 22 performs voice synthesis from a word sequence or edits a voice registered in advance according to the word sequence to generate a voice signal. When this voice signal is given to the voice output unit 12, the synthesized voice is sent to the outside.

Since it is desirable that the voice output from the voice output unit 12 be close to a natural voice, the following technique is used. That is, “Isao Tokuda, Tohru Ikeda, Naoya Miyano, Kazuyuki Aihara,“ Speech perception psychological experiment based on surrogate method ”, IEICE Technical Report, No. NLP99-150
~, Pp. 17-21, 2000 "," a vowel holding a sound pressure fluctuation sounds more natural ". Therefore, the speech synthesis unit 22 has a function of adding chaotic fluctuation to a voice signal. By having it, it is closer to a natural voice.

By the way, in the present embodiment, a plurality of types of characters are set in the dialogue processing means 20, and the overall control section 10 comprises a character selecting means for selecting the character set in the dialogue processing means 20. Here, the character is assumed as a responder who responds to the user, and if the character is different, not only the voice color, but also the tone and accent of words are changed. That is, the attribute of the character includes a voice color, tone, word usage, accent, and the like. Therefore, if the dialogue processing unit 20 is to perform rakugo, a character is set for each character. Also, for the elderly, "son",
Characters corresponding to "daughter" and "grandchild" can also be set.

Here, as an example of character selection, an example in which the communication support device 1 executes a rakugo story will be described. That is, in the present embodiment, it is assumed that the overall control unit 10 is programmed so that it can execute rakugo. At the start of the rakugo story, the user instructs the communication support device 1 to perform whatever the communication support device 1 does without specifying the purpose. For example, say "Do something". As described above, since the communication support device 1 of the present embodiment can play rakugo, the communication support device 1 responds with "Shall I do rakugo?" here,
If the user makes a positive response such as "please", the dialogue processing means 20 notifies the central control unit 10 that the instruction to start rakugo is given. When the start of the rakugo story is instructed, the overall control part 10 causes the drive part control part 32 to perform the motion of the parts corresponding to the neck and the arm based on the procedure of the movement set in advance according to the content of the theme of the rakugo story. The voice processing unit 20 gives an instruction to the dialogue processing unit 20 to give a voice color and utterance content suitable for the character being presented, and gives an instruction to the voice output unit 12 to adjust the amplification factor according to the utterance content.
In accordance with such an instruction, the voice synthesizer 22 of the dialogue processing means 20 generates a voice color of each character. Now, if there are two characters (that is, characters) who talk during the rakugo story, the head of the robot turns to the left in the utterance of one character, and the robot head in the utterance of the other character. Control the neck so that the head turns to the right. In addition, the motion of the robot is given a facial expression by appropriately instructing the motion of other parts according to the utterance content. Also, the characters are differentiated by making the voice color different for each character.

In the above example, the communication support device 1 automatically sets the character according to the content of the storyteller, but it is also possible to select the character according to the intention of the user. For example, “son”, “daughter”, “grandchild” are placed on the chest part of the communication support device 1.
By providing a plurality of selection buttons described as and pressing the desired selection button, the communication support apparatus 1 can speak in the voice color of the selected character. In this case, not only the voice color is changed for each character, but also the utterance content (word use) and tone are controlled so as to be different depending on the character. For example,
When it is determined from the voice of the user input from the voice input unit 11 that the user is not feeling well and is in a bad mood,
When "son" is selected, he speaks in a strong tone with excitement, and when "grandchild" is selected, he speaks in a gentle tone.

By the way, there are cases where the speech recognition unit 21 in the dialogue processing means 20 cannot uniquely recognize the utterance content of the user. In such a case, the voice recognition unit 21
Then, a plurality of recognition candidates are generated for the utterance content, and the certainty factor is output for each recognition candidate. When there is a significant difference in the certainty factor, the content with the higher certainty factor may be the recognition content, but when there is no significant difference in the certainty factor, the following processing is performed.

Now, when the user says "wake up at 5 o'clock" during a dialogue with the user, the voice recognition section 21 indicates "wake up at 5 o'clock" and "wake up for construction". It is assumed that two recognition candidates are obtained and there is no significant difference in the certainty factors of both recognition candidates. In such a case, the overall control unit 10 selects a character to be assigned to one of the recognition candidates, speaks out "Are you sure you want to wake up at 5?", And selects another character to be assigned to the other recognition candidate. Ask him, "Are you sure you want to wake it up for construction?" In this way, by selecting and uttering a different character for each recognition candidate, the user can clearly recognize that the recognition candidate has been confirmed, and affirmatively answer the question regarding any recognition candidate. If you respond,
The recognition candidate will be confirmed. By using such a technique, the possibility of erroneous recognition can be reduced because confirmation is performed by dialogue even if there is no significant difference in the certainty factor for a plurality of recognition candidates.

As described above, the speech recognition and weaving unit 21 estimates the word sequence using the acoustic model indicating the appearance probability of the feature quantity sequence of the voice waveform when the prepared word sequence is uttered. Therefore, when the acoustic model is created by the voice waveform of a young person, the recognition rate is high if the user is a young person, but the recognition rate tends to be low if the user is an elderly person. In other words, it can be said that the recognition rate is higher when the acoustic model created by a person according to the supposed user is used. Here, it may be possible to use an average acoustic model based on voices of all ages and sexes, but the deviation increases, which leads to a decrease in the acceptance rate.

Therefore, a plurality of acoustic models such as an acoustic model by a young person and an acoustic model by an elderly person are prepared, different characters are assigned to each acoustic model, and a confirmation response to the plurality of recognition candidates described above is performed. As a result, an acoustic model suitable for the user's voice is extracted. In this way, it is possible to reduce false recognition by making the integrated control unit 10 learn by the information for selecting the acoustic model.

(Second Embodiment) In the first embodiment, the user is identified by the tone of the voice input from the voice input unit 11, but in the present embodiment, as shown in FIG. An example will be given in which user information for identifying the age / sex of the user is given from the user information input means 40.

The user information input means 40 is provided with a memory 41 storing user information, and transmits a signal that enables data transmission to the signal receiving unit 18 provided in the robot which is the main body of the communication support apparatus 1. The unit 42 is provided. In the present embodiment, the signal transmitter 42 and the signal receiver 18
The human body is used for at least part of the transmission line between
By sending a minute current through the human body, the signal transmitter 4
The data transmission from 2 to the signal receiving unit 18 is possible. The user information input means 40 can be attached to a wrist (for example, a bracelet type) or a finger (for example, a ring type) of a human body, and the signal transmitting unit 42 and the signal receiving unit 18 can be obtained by touching a predetermined portion of the robot with a fingertip or the like. A transmission line between the and is formed.

In the memory 41, as user information, unique information of the user such as age and age of the user, biometric information such as body temperature, pulse rate and blood pressure of the user, and for identifying the user Any one or a combination of identification information and system setting information described later can be recorded. Therefore, when the user wearing the user information input means 40 touches a predetermined portion of the robot as the main body of the communication support apparatus 1, the user information stored in the memory 41 is transferred to the signal receiving unit 18, and is integrated. The user information is handed over to the control unit 10. In this way, the user information can be handed over to the central control unit 10 simply by touching the robot, which is the main body of the communication support apparatus 1, so that even an elderly person who is unfamiliar with the operation of the device can use it.

Memory 4 provided in user information input means 40
If the user information stored in 1 is unique information such as the age and sex of the user, the overall control unit 10 selects a character having the same sex (or opposite sex) of the age group and calls the user. The language is also decided based on the unique information. For example, if the user is a male senior, the user is called "grandpa", and if the user is a female senior, it is called "grandma".

When the user information is biometric information such as body temperature, heart rate, and blood pressure, when the ecological information deviates from the normal range, the overall control unit 10 selects the character of the doctor and is slightly dignified. Speaking advice in a voice. Here, the character is also displayed on the display unit 16.

When the user information is identification information, the overall control unit 10 collates the identification information and selects a character to be displayed on the display unit 16 depending on whether the identification information matches. For example, if the identification information matches, a smiling character with a smile is displayed, and the voice output unit 12 utters that the identification information matches. If the identification information does not match, a character like Enma Daio is displayed. The character is displayed, and the voice output unit 12 speaks to indicate that the identification information is incorrect. In this way, the security function can be realized and the enjoyment can be realized by a simple operation.

By the way, the communication support device 1
Is connected to the video / audio server 4 via the communication path 3 including the Internet, the system setting information necessary for connection with the video / audio server 4 is also used as the user information. The system setting information includes the telephone number of the provider, the user name, the password, the IP address of the server, etc.
It is registered in the memory 41 of the user information input means 40 in the manufacturer that provides the communication support device 1.
Therefore, the user can connect to the Internet simply by touching the robot, which is the main body of the communication support apparatus 1, and touch the robot even if the system setting information set in the communication means 15 is lost due to some reason. Only then can the system setting information be easily reset. Also in this case, when the connection to the communication path 3 is successful, a smiling character is displayed on the display unit 16. Other configurations and operations are similar to those of the first embodiment.

In the present embodiment, an example in which the human body is used for at least a part of the transmission path between the user information input means 40 and the signal receiving section 18, but the transmission path is wireless transmission using infrared rays or the like as a transmission medium. It may be a road.

(Third Embodiment) In the present embodiment, the central control unit 10 can select a character that makes a relationship during the conversation, in addition to the character assumed as a responder who interacts with the user. Is. In other words, during the dialogue between the character as the responder and the user, a motion of slightly raising and lowering the neck together with the utterance of "Umm" or a motion of clapping hands with an utterance such as "I see, that is good" is performed. The characters are selected at appropriate timings such as when going, and at the same time, a character expressing a positive attitude is selected. Other configurations and operations are similar to those of the first embodiment.

In the present embodiment, a plurality of characters are selected by one robot, but a plurality of robots may be used so that robots other than the robot in dialogue with the user can make a joint. Good.

(Fourth Embodiment) In this embodiment, as shown in FIG. 4, the configuration of the dialogue control unit 23 and the voice synthesis unit 22 in the communication support apparatus 1 is different from that of the first embodiment. In addition, the dialogue control unit 23
Uses the voice stored in the voice database 43. In the present embodiment, the voice database 43 is provided in the image / voice server 4 which is a server, but the voice database 43 may be provided in a server provided separately from the image / voice server 4, and as will be described later, the voice database 4 is provided.
3 does not necessarily have to be provided in the server.

FIG. 4 shows only the main part of the communication support device 1, and the components having the same functions as those of the other embodiments described above are omitted. The dialogue control unit 23 of the present embodiment collates the word sequence output from the voice recognition unit 21 with the rules stored in the dialogue rule base 23c to generate the content of the voice response to the user as text data. The sentence generation unit 23d and the response contents generated by the response sentence generation unit 23d are stored in the voice database 4
3 and a voice search unit 23e that matches with No. Here, for simplification, the rules stored in the dialogue rule base 23c are shown in Table 1 as if-then-type rules in which the utterance content of the user is the condition part and the response content is the response part. However, actually, as described in the first embodiment, the dialogue rule base 23c includes various rules for syntax analysis and semantic analysis. In addition, Table 1
The contents of the condition part and the response part of the above are written in katakana, which means that they are phoneme of text data.

[0067]

[Table 1]

In the response sentence generator 23d, as shown in FIG. 5, the text data extracted from the word sequence which is the recognition result of the voice recognizer 21 is collated with the condition part of the rule as shown in Table 1 (S1). If there is a rule in which the text data matches the condition part (S2), the response part of the rule is output as text data of the content of the voice to be answered (S3). If there is no rule in which the text data matches the condition part, the text data indicating that the user's utterance content cannot be understood is output (S4).

By the way, the voice database 43 is a database in which voices having a plurality of contents are registered for a plurality of types of characters. Basically, the voice database 43 is a character whose voices are registered. I am assuming a family like grandchildren. That is, the voice database 43 is usually registered with a real voice, not a synthetic voice. The data stored in the voice database 43 is not a real voice itself but a voice file converted into a digital signal and converted into a predetermined format (for example, WAVE format). Each record of the voice database 43 includes voice files and voice contents. It also includes identifying information and user information for identifying users and their families. In addition, a plurality of types of everyday greetings and callouts are registered as the content of the voice. Table 2 shows an example of data registered in the voice database 43. In Table 2, the "text" column is text data that serves as a search key, and the "voice" column shows the contents of the registered voice file.

[0070]

[Table 2]

The voice search unit 23e, as shown in FIG.
It has a function of searching the data of the voice database 43 through the communication path 3 when the text data which is the content of the response voice is delivered in the response sentence generation unit 23d (S1), and is provided in the overall control unit 10. The character selected by the character selection means (that is, the user's family)
The content of the voice database 43 is searched with a search condition as to whether or not a voice that matches the content of the voice generated by the response sentence generation unit 23d is registered in the voice database 43. S2). In the voice database 43, when there is a record satisfying the search condition designated by the voice search unit 23e, the communication path 3
The voice file is transferred to the voice search unit 23e via the, and when there is no record satisfying the search condition, the voice search unit 23e is notified that the voice file does not exist.

The voice synthesizing unit 22 of the present embodiment outputs the synthesized voice obtained by performing the text synthesizing according to the content of the voice generated by the response sentence generating unit 23d.
And a voice reproduction unit 22b that reproduces the voice file searched by the voice search unit 23e from the voice database 43. Therefore, when there is no voice file satisfying the search conditions given by the voice search unit 23e to the voice database 43, the text synthesizing unit 22a generates a synthetic voice corresponding to the content of the voice output from the response sentence generating unit 23d (S4). ), When a voice file satisfying the search condition is transferred from the voice database 43, the voice reproducing unit 22
The voice which converted the digital signal into the analog signal in b is output from the voice output unit 12 (S3).

[0073] For example, if the collation of the text data is requested as "Hello" to the audio database 43 from the speech retrieval unit 23e, since it is possible to extract the audio file named "Hello" by Table 2, the speech synthesis unit A voice is output from the voice output unit 12 through the voice reproduction unit 22b in 22.

As is clear from the above description, the communication support device 1 of the present embodiment has the image / audio server 4
And functions as a user terminal used by the user. Moreover, as described above, it has a function of retrieving the voice of the family from the voice database 43, so that it can be regarded as an agent of the family and functions as a so-called agent type terminal.

By the way, in order to realize the above operation, it is necessary to register the voice in the voice database 43. In the present embodiment, the audio database 4 is provided through the audio terminal 2c connectable to the image / audio server 4 via the communication path 5.
The voice can be registered in 3. The communication path 5 may have any form and can be shared with the communication path 3. As the voice terminal 2c, a personal computer or the like can be used as long as it can be connected to the communication path 5, but by using a telephone terminal such as a mobile phone or a fixed line telephone as the voice terminal 2c, voice can be easily stored in the voice database 43. You will be able to register at. In other words, as a matter of course, telephone terminals have a built-in microphone for inputting voice, and moreover, mobile phones as well as fixed-line telephones are equipped with a display unit. By displaying a message instructing the content of the voice to be registered or the image and voice server 4 on the display unit, it becomes possible to register the voice with the proper content in the voice database 43.

Here, in the image / sound server 4, a sound communication unit 44 which enables communication with the sound terminal 2c,
The voice database 43 is additionally provided with a registration processing unit 45 for registering the voice input from the voice terminal 2c through the voice communication unit 44 in the voice database 43. Now, a mobile phone is used as the voice terminal 2c, and various messages are displayed on the display part 2d of the mobile phone as shown in FIG. 7, and the buttons displayed on the display part 2d can be operated by key operation of the mobile phone. And Voice terminal (mobile phone) 2
When the voice terminal 2c and the voice communication unit 44 can communicate with each other when the voice is registered in the voice database 43 from "c", as shown in FIG. 7A, it is described in the "voice" column shown in Table 2. (in the illustrated example, "Hello") content of the message is displayed together with the instruction of the registration procedure ( "Please saying the" in the illustrated example). Since the "recording start" button B1 is displayed at the bottom of the screen of the display unit 2d, the person who wants to register a voice turns on the "recording start" button B1 by key operation of the voice terminal 2c which is a mobile phone. Operate and speak the voice of the specified content.

When the "start recording" button B1 is turned on, FIG.
As shown in (b), "Recording" is displayed in the center of the screen of the display unit 2d, and a "Recording end" button B2 is displayed instead of the "Recording start" button B1. When the operation of turning on the “recording end” button B1 is performed after the speech of the instructed content is finished, the spoken speech is temporarily stored in the image / audio server 4. "Recording end" After the operation of the button B2, as shown in FIG. 7 (c), the in the center portion of the screen of the display unit 2d, designated contents of the message (in the illustrated example, "Hello") is registered in the confirmation message (In the illustrated example, "Are you sure you want to register?") Is displayed.
Furthermore, as the buttons for responding to the confirmation message, there are "play" button B3, "cancel" button B4, and "OK".
Button B5 is displayed. By operating the "Play" button B3, you can check the spoken voice, and the "Cancel" button B
4 is operated, the voice temporarily stored in the image / voice server 4 is erased. When the voice temporarily stored in the image / voice server 4 is registered in the voice database 43, the “OK” button B5 is turned on.
The series of processes is performed by the registration processing unit 45.

According to the above procedure, the voice database 43
You can register your voice using a mobile phone. In addition to the voice file, the record of the voice database 43 needs information for identifying the contents of the voice and user information for identifying the user and family. As the information for identifying the content of the voice, the instruction content at the time of registering the voice may be used. In addition, user information for identifying the user and family is prepared in advance as a data file in association with the telephone number of the mobile phone, and the telephone number of the mobile phone is accessed when the image / voice server 4 is accessed from the mobile phone. The user information for identifying the user or family member corresponding to the telephone number is extracted by collating with the data file. Here, an example in which a mobile phone is used as the voice terminal 2c is shown, but the form of the voice terminal 2c does not matter as long as it has a function of inputting voice and a function of communicating with the image / voice server 4. Therefore, the information associated with the user information for identifying the user or family by the above-mentioned data file does not have to be a telephone number, and for example, an electronic mail address or a password can be used.

In the above-mentioned example, the instruction of the contents of the voice to be registered is displayed one by one on the display section 2d. However, since many kinds of the voice contents are usually required, they are shown in the left column of Table 2. The text data may be divided into a plurality of pieces and displayed collectively on the display unit 2d. That is, the contents of a plurality of types of voices may be displayed in a list on the display unit 2d in a menu format, and the voices may be sequentially registered, or desired contents may be selected from a plurality of types of contents for road recording. Specifically, a button is provided on the screen of the display unit 2d for instructing whether to sequentially register or select and register, and when sequentially registering, each time one content is registered, the next registration is performed. There should be a button for this. Further, when selecting and registering, a number may be given to a plurality of options and desired contents may be selected by the number. Subsequent operations are based on the case of registering the voice contents one by one. Furthermore, in the above example, the audio terminal 2c is provided with the display unit 2d for displaying an image, but the instruction of the registration procedure can also be given by voice instead of the image.

(Fifth Embodiment) In the fourth embodiment, the communication support device 1 responds to the user's utterance by using the voice of the user's family or the like stored in the voice database 43. , The user can simulate a conversation with the family. On the other hand, the user's family can feel at ease if they can obtain information on the safety of the user. Therefore, in the present embodiment, as shown in FIG. 8, a server provided with the voice database 43 (for example,
The image / sound server 4) is provided with a dialogue record storage unit 46 for registering a history of the user's dialogue with the communication support apparatus 1. That is, when the user has a dialogue with the communication support apparatus 1, the voice database 43 is accessed, so that the history of the contents of the dialogue is stored in the dialogue recording storage unit 46, and if this is read, the safety of the user can be obtained. You can get a guide to know the user's living condition.

The recording of data in the dialogue record storage section 46 may be performed when the voice database 43 is accessed, or may be periodically performed in some cases. On the other hand, in order to read out the data from the dialogue record storage unit 46, a server provided with the dialogue record storage unit 46 (for example,
A terminal 2e such as a mobile phone or a fixed phone or a personal computer equipped with a speaker, which is connected to the image / sound server 4) via the communication path 5, may be used. Like the voice terminal 2c described in the fourth embodiment, the communication path 5 between the terminal 2e and the server may be the same as the communication path 3. If a mobile phone is used as the terminal 2e, it can also be used as the voice terminal 2c. Here, if the recorded contents of the dialogue record storage unit 46 are read out by the terminal 2e, it is possible to generate the operating cost of the server (for example, the image / audio server 4). For billing, a billing processing unit 47 for detecting access from the terminal 2e to the dialogue record storage unit 46 may be provided. The charging system by the charging processing unit 47 can be set appropriately.

In the above configuration, when the user's voice is input from the voice input unit 11, the response sentence generation unit 23d generates the content of the response voice, and this content is collated with the voice database 43 to generate one voice. However, when a character is selected by the character selecting means, voices corresponding to the selected character are collectively fetched from the voice database 43 and stored, and then a response sentence is generated. When the voice extracted from the voice database 43 includes the voice corresponding to the voice generated by the unit 23d, the voice may be used as a response to the user. In this way, if the voices of a plurality of types of contents are collectively extracted from the voice database 43, the time from the user's utterance to the response can be made relatively short without being affected by the traffic on the communication path 3. You can

In the fourth and fifth embodiments, the communication support device 1 is used synonymously with the user terminal installed in the user's house, but the user terminal is referred to as the communication support device 1. Synonyms are not required. Therefore, in addition to the configuration in which the voice database 43 is installed far from the user terminal as in the above example, the voice database 43 may be installed in the user's home or incorporated in the communication support device 1. That is, by providing the voice database 43 in the same place as the communication support device 1, it is possible to construct a communication support system without using a separate server (for example, the image / voice server 4). A server in which the voice input unit 11 and the voice output unit 12 of the configuration of the communication support apparatus 1 are provided in the user terminal and the dialogue processing means 20 of the communication support apparatus 1 is installed far from the user terminal (for example, , The image / sound server 4) may be provided together with the sound database 43.

[0084]

According to the first aspect of the present invention, the voice input unit and the voice output unit are connected to each other, and it is assumed that the user has a voice dialogue using natural language with the user and is a responder who responds to the user. The character includes a dialogue processing unit capable of selecting from a plurality of types, and a central control unit having a character selection unit for selecting an appropriate character from the characters in the dialogue processing unit according to a dialogue environment. It is possible to eliminate the monotonousness of the dialogue by selecting the character, and it is possible to keep the dialogue interesting and suppress the decrease in the frequency of use.

According to the invention of claim 2, in the invention of claim 1, since a voice color is included as an attribute of the character, only a relatively simple structure for controlling the voice color is adopted.
It is possible to reduce the monotonousness of the dialogue by giving the impression of talking to multiple people.

According to a third aspect of the present invention, in the first aspect of the present invention, a driving source for driving a robot having a movable part expressing an expression is added, and the character selecting means is expressed by the motion of the movable part. The character is selected according to the action, and the character is expressed by the action. Therefore, it is possible to reduce the monotonousness caused only by the dialogue by attracting the interest by the action.

According to a fourth aspect of the present invention, in the first aspect of the invention, since the operation unit for selecting the character is provided, the character can be selected according to the wish of the user.
By allowing the user to select a favorite character, it becomes possible to maintain the interest in the dialogue.

According to a fifth aspect of the present invention, in the first aspect of the invention, there is a function of calling out to the user through the dialogue processing means so as to prompt the user to speak, and the character selecting means responds to the content of the user's response. Since the character is selected according to the content of the user's utterance, the character is selected interactively, and monotonousness due to the formalization of the character selection is prevented. be able to.

According to a sixth aspect of the present invention, in the first aspect of the present invention, a user information input means for inputting user information regarding the user is added, and the character selecting means causes the character according to the user information. Is selected, the character is selected based on the user information provided by the user, so that the character suitable for the user is selected.

According to the invention of claim 7, in the invention of claim 6, the user information input means stores the user information and is carried by the user, and the user information is input from the previous user information input means. Since the user information is provided from the user information input means possessed by the user, since the user information is received and passed to the integrated control unit, the user information whose contents like biometric information of the user fluctuates from time to time. But you can enter it easily.

According to the invention of claim 8, in the invention of claim 7, since the transmission path between the user information input means and the signal receiving section is a wireless transmission path, it can be used even if the user is away from the user. It becomes possible to input personal information.

According to a ninth aspect of the invention, in the seventh aspect of the invention, since at least a part of the transmission path between the user information input means and the signal receiving section is a human body, it can be touched by a user. It becomes possible to input user information.

According to a tenth aspect of the invention, in the inventions of the sixth to ninth aspects, the user information is the age of the user.
Since the gender is included as the unique information, it is possible to select the character according to the age and gender of the user.

According to the invention of claim 11, in the invention of claims 6 to 9, since the user information includes biometric information of the user, it is possible to manage the health condition of the user.

According to a twelfth aspect of the present invention, in the inventions of the sixth to ninth aspects, since the user information includes identification information for specifying the user, it is possible to confirm the identity of the user.

According to a thirteenth aspect of the present invention, in the sixth to ninth aspects, the user information includes system setting information required when connecting to an external device connected via a communication path. You can easily set the system setting information required to connect to an external device through the communication path, and even if you lose this type of system setting information, you can easily reset it using the user information input means. become.

According to a fourteenth aspect of the present invention, in the first to thirteenth aspects of the present invention, since the character selecting means can select a character that is supposed to be a responder and a character that makes a relationship during the dialogue, As a result, the user's utterance is prompted, and as a result, it is possible to reduce the decrease in the frequency of use.

According to a fifteenth aspect of the invention, in the inventions of the first to fourteenth aspects, when the character selection means obtains a plurality of recognition candidates for a recognition result for a voice input of a user, the recognition candidates are respectively recognized. Different characters are assigned, and the dialogue processing unit has a function of asking the user to confirm the recognition candidate and determining the content of the voice input by the user when receiving a positive response from the user, By using the acoustic model suitable for the user's voice, the recognition rate of the voice can be improved, and since the user can answer the question, the interest in the dialogue is maintained and the frequency of use by the user is increased. The decrease can be reduced.

According to a sixteenth aspect of the present invention, in the fifteenth aspect of the present invention, the dialogue processing means is provided with a voice recognition section for recognizing the utterance content of the user, and the voice recognition section collates with a voice feature quantity series. A plurality of acoustic models are provided and a character is assigned to each acoustic model, and a character suitable for the voice of the user is extracted according to the result of confirmation of the recognition candidate, and an acoustic model corresponding to the extracted character is extracted. It is used in the voice recognition unit, and since it is possible to narrow down the acoustic model that suits the user from a plurality of acoustic models through the dialogue with the user, an appropriate acoustic model will be automatically selected,
As a result, the recognition rate of the user's voice increases.

The inventions of claims 17 to 25 relate to a communication support system using the communication support device according to the invention of claims 1 to 16.

The invention according to claim 17 comprises the communication support device according to any one of claims 1 to 16 and a voice database in which voices of a plurality of contents are registered for a plurality of the characters, The dialogue processing means corresponds to the response sentence generation unit that generates the content of the voice that responds to the user, the character that is selected by the character selection unit, and the content of the voice that is generated by the response sentence generation unit. A voice search unit for searching the voice database for a voice of the content is provided, and when the voice search unit extracts the relevant voice from the voice database, the voice is used for dialogue with the user, and the voice database is provided. This makes it possible to set the user's family as a character, and when the user is an elderly person living alone, Of using the voice brings a sense of security to the user by can be performed interactively, yet monotony when compared to the mechanical voice that it is possible to continue the conversation remembering interested in dialogue because is reduced.

According to an eighteenth aspect of the present invention, in the seventeenth aspect of the present invention, the voice database includes a voice communication section capable of communicating with a voice terminal capable of inputting voice, and a voice communication section from the voice terminal through the voice communication section. Since a voice input terminal is provided with a registration processing unit for registering voices, any voice can be registered in the voice database through a voice terminal. Voice can be registered.

According to the nineteenth aspect of the invention, in the eighteenth aspect of the invention, since the voice terminal is a telephone terminal, the voice can be easily registered in the voice database using a mobile phone or a fixed telephone.

According to a twentieth aspect of the present invention, in the eighteenth or nineteenth aspect of the present invention, the registration processing unit has a function of instructing the voice terminal about the content of voice to be registered in the voice database. Since the content of the voice to be registered in the voice database is instructed by the registration processing unit, the voice can be easily registered by registering the voice according to the instruction, and the content of the voice used for dialogue in the communication support device. By instructing in advance, there is a high possibility that the voice stored in the voice database can be used for dialogue.

According to a twenty-first aspect of the invention, in the eighteenth or nineteenth aspect of the invention, the voice terminal includes a display section for displaying an image, and the registration processing section registers the voice terminal with the voice database. It has a function to display the contents of the voice to be displayed on the display unit in a menu format. Since the contents of the voice to be registered in the voice database are instructed in the menu format from the registration processing unit, the content of the instruction is easy to understand and By registering the voice according to
In addition to the ability to easily register the voice, the voice stored in the voice database is more likely to be used for the dialogue by instructing the content of the voice used for the dialogue in the communication support device in advance.

A twenty-second aspect of the present invention is the user terminal according to the seventeenth to twenty-first aspects, wherein the communication support device interacts with the user through the voice input section, the voice output section, and the dialogue processing means. In preparation for the above, the voice database is provided in a server capable of data communication with a user terminal via a communication path, and the voice search unit corresponds to the character when the character is selected by the character selecting means. When the voice extracted from the voice database is included in the voice extracted from the voice database when the voice having the content corresponding to the content of the voice generated by the response sentence generation unit is included in the voice extracted from the voice database at once. The voice database is used for the dialogue of the Protection and maintenance of the speech database is ensured by the administrator of a server is listening.

According to a twenty-third aspect of the invention, in the seventeenth to twenty-first aspects of the invention, the voice database is provided at the same place as the communication support device, and a server is not required, so that the system configuration is simple. become.

In a twenty-fourth aspect of the present invention based on the seventeenth to twenty-first aspects, the communication support device is provided with a user terminal that interacts with the voice input unit and the voice output unit, and The dialogue processing means is provided in a server capable of data communication with the user terminal via a communication path, the voice database is provided in the server, and the voice database and the dialogue processing means are provided in the server. Therefore, the voice database and the dialogue processing means are monitored and monitored by the server administrator who is an expert, so that the voice database and the dialogue processing means are surely protected and maintained.

According to a twenty-fifth aspect of the present invention, in the invention of the twenty-second aspect or the twenty-fourth aspect, the server stores an interaction record storage unit for registering a history of interaction of the communication support device with the user, and a server. And a billing processing unit that performs billing processing when the content of the dialogue recording storage unit is read from a terminal connected via a communication path. Since the history can be read out through the communication channel, it becomes possible to know from the outside whether or not the dialogue by the user is normally performed. For example, when the user is an elderly person living alone, the safety of the elderly person living alone can be confirmed even if the family cannot directly communicate with the elderly person living alone by reading the conversation record. In addition, since the fee is charged when reading the dialogue record storage unit, the contents of the dialogue record storage unit are not unnecessarily read, and it is possible to suppress an increase in the traffic of the server and generate the operating cost of the server.

[Brief description of drawings]

FIG. 1 is a block diagram showing a first embodiment of the present invention.

FIG. 2 is a block diagram showing a dialogue processing means used in the above.

FIG. 3 is a block diagram of a main part according to a second embodiment of the present invention.

FIG. 4 is a block diagram showing a fourth embodiment of the present invention.

FIG. 5 is an operation explanatory diagram of the above.

FIG. 6 is an operation explanatory diagram of the above.

FIG. 7 is an operation explanatory diagram of the above.

FIG. 8 is a block diagram showing a fifth embodiment of the present invention.

[Explanation of symbols]

1 Communication support device 2a Mobile terminal 2b terminal 2c voice terminal 2d display 2e terminal 3 communication channels 4 image and sound server 5 communication paths 10 Integrated control section 11 Voice input section 12 Audio output section 20 Dialog processing means 21 Speech recognition unit 23d Response sentence generator 23e Voice search unit 31 Drive source 40 User information input means 43 voice database 44 Voice communication unit 45 Registration Processing Department 46 Dialog record storage 47 Accounting unit

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ Identification code FI theme code (reference) G10L 3/00 561E 561D (72) Inventor Takeyuki Suzuki 1048, Kadoma, Kadoma-shi, Osaka Matsushita Electric Works Co., Ltd. Inner F term (reference) 5D015 KK02 LL06 5D045 AB11 AB30

Claims (25)

[Claims] 1. A voice input unit and a voice output unit are connected to perform a voice dialogue using a natural language with a user, and a character assumed as a responder who responds to the user can be selected from a plurality of types. And a general control unit having character selection means for selecting an appropriate character from the characters in the dialogue processing means according to the dialogue environment. 2. The communication support device according to claim 1, wherein a voice color is included as an attribute of the character. 3. A drive source for driving a robot having a movable part expressing an expression, and the character selecting means selects a character according to the expression expressed by the motion of the movable part. The communication support device according to claim 1. 4. The communication support device according to claim 1, further comprising an operation unit for selecting the character. 5. A function for calling the user to utter a speech through the dialogue processing means, wherein the character selecting means selects the character according to the content of the user's response. Item 1. The communication support device according to item 1. 6. The user information input means for inputting user information about the user is added, and the character selection means selects the character according to the user information. Communication support device. 7. A signal receiving unit for storing the user information and possessed by the user by the user information input unit, receiving the user information from the previous user information input unit, and delivering the user information to the general control unit. The communication support device according to claim 6, further comprising: 8. The communication support apparatus according to claim 7, wherein a transmission path between the user information input means and the signal receiving section is a wireless transmission path. 9. The communication support apparatus according to claim 7, wherein at least a part of a transmission path between the user information input means and the signal receiving unit is a human body. 10. The communication support device according to claim 6, wherein the user information includes the age / sex of the user as unique information. 11. The communication support device according to claim 6, wherein the user information includes biometric information of the user. 12. The communication support device according to claim 6, wherein the user information includes identification information for identifying a user. 13. The user information according to claim 6, wherein the user information includes system setting information required when connecting to an external device connected via a communication path. The communication support device described. 14. The character selecting means is capable of selecting, together with a character assumed as a responder, a character that makes a relationship during a dialogue.
The communication support device according to claim 13. 15. When the character selection means obtains a plurality of recognition candidates for a recognition result for a voice input by a user, a different character is assigned to each recognition candidate, and the dialogue processing means confirms the recognition candidates. 15. The method according to claim 1, further comprising a function of determining the content of the voice input by the user when the user is asked a question and a positive response by the user is received.
The communication support device according to any one of 1. 16. The dialogue processing means is provided with a voice recognizing unit for recognizing a user's utterance content, and the voice recognizing unit is provided with a plurality of acoustic models to be compared with a feature amount series of voice and each acoustic model A character is assigned to each of the characters, a character suitable for the voice of the user is extracted according to the result of confirmation of the recognition candidate, and an acoustic model corresponding to the extracted character is used in the voice recognition unit. 15. The communication support device according to item 15. 17. The communication support device according to claim 1, and a voice database in which voices of a plurality of contents are registered for the plurality of characters, and the dialogue processing means. A response sentence generation unit that generates the content of a voice response to the user,
A voice search unit for searching a voice database for a voice corresponding to the character selected by the character selection unit and corresponding to the voice generated by the response sentence generation unit, and the voice search unit corresponds to the voice database. A communication support system characterized in that when a voice to be extracted is extracted, the voice is used for dialogue with a user. 18. The voice database includes a voice communication unit capable of communicating with a voice terminal capable of voice input, and a registration processing unit for registering voice input from the voice terminal through the voice communication unit. The communication support system according to claim 17, wherein the communication support system is attached. 19. The communication support system according to claim 18, wherein the voice terminal is a telephone terminal. 20. The communication support system according to claim 18, wherein the registration processing unit has a function of instructing the voice terminal about the content of voice to be registered in the voice database. 21. The voice terminal includes a display unit for displaying an image, and the registration processing unit has a function of displaying, in a menu format, the content of voice to be registered in the voice database on the voice terminal. The communication support system according to claim 18 or 19, characterized in that. 22. The communication support device comprises a voice input unit, a voice output unit, and a dialogue processing unit in a user terminal that interacts with a user, and the voice database communicates with the user terminal. Is provided in a server capable of data communication via, the voice search unit collectively retrieves voices corresponding to the character from the voice database when the character is selected by the character selection unit, and the dialogue processing unit, 2. When the voice extracted from the voice database includes a voice having a content corresponding to the content of the voice generated by the response sentence generation unit, the voice is used for a dialogue with a user.
The communication support system according to any one of claims 7 to 21. 23. The communication support system according to claim 17, wherein the voice database is provided at the same place as the communication support device. 24. The communication support device,
The voice input unit and the voice output unit are provided in a user terminal that interacts with a user, the interaction processing unit is provided in a server that can perform data communication with the user terminal via a communication path, and the voice database is The communication support system according to any one of claims 17 to 21, wherein the communication support system is provided in a server. 25. The server comprises a dialogue record storage unit for registering a history of dialogues with the user by the communication support apparatus, and a dialogue record storage unit for storing a dialogue record storage unit from a terminal connected to the server via a communication path. 25. The communication support system according to claim 22, further comprising: a billing processing unit that performs a billing process when the content is read.