JP2006301968A - Conversation support device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a conversation support device capable of improving the convenience in foreign language conversations or the like. <P>SOLUTION: The conversation support device is connected to a database storing a response sentence table in which spoken sentence identification information identifying a spoken sentence is related to a response sentence list including assumed responses to the spoken sentence so as to be accessible thereto. The device acquires a response sentence list related to spoken sentence identification information of an inputted spoken sentence and displays response sentences contained in the response sentence list to make a responder selectively read out them. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a conversation support apparatus that supports conversation between a questioner and a responder.

  Today, international transportation networks have been improved, making it easy to travel to foreign countries, and increasing opportunities to talk with foreigners of various nationalities. Under these circumstances, the types of foreign languages used for conversations are increasing, and devices that support such foreign language conversations are desired.

Patent Document 1 discloses an electronic dictionary that questions foreign language problems in order to support the acquisition of foreign languages. Research on so-called machine translation, in which sentences are grammatically analyzed and translated, is also underway.
JP 2004-294993 A

  However, the conventional electronic dictionary and machine translation technology do not always support conversation smoothly. For example, in an electronic dictionary, it is useful when there are unknown words in the content to be spoken, but it is not useful when assembling sentences as conversations. Also, if there is an unknown word spoken by the other party, the meaning cannot be drawn if the spelling is unknown. In the case of machine translation, it is possible to assemble sentences as conversations, but there is a problem that it is impossible to translate when the content of the other party is not heard.

  As described above, the conventional electronic dictionary and machine translation technology have a problem in that they are not very convenient in foreign language conversation. This situation is not limited to foreign language conversations.

  The present invention has been made in view of the above circumstances. For example, it is an object of the present invention to provide a conversation support apparatus that can improve convenience in foreign language conversation.

  The present invention for solving the problems of the conventional example described above is a response in which an utterance sentence identification information for identifying an utterance sentence is associated with an answer sentence list enumerated as an assumed response to the utterance sentence in the conversation support device A response sentence acquisition unit that is connected to a database that stores a sentence table and is associated with the utterance sentence identification information of the input utterance sentence, and included in the response sentence list And a response sentence display unit that enumerates and displays the response sentences.

  Here, the response sentence table further stores specific response availability information indicating whether or not a response by a word indicating affirmation or denial is associated with the utterance sentence identification information, and the question identification of the input utterance sentence A response sentence adding unit that adds an affirmative or negative word to the acquired response sentence list when a response by a word indicating affirmation or negative is possible with reference to the specific response availability information associated with the information , May further be included.

  Further, the database stores an instruction table in which an instruction to the speaker side and an instruction to the responder side are associated with the utterance sentence identification information for identifying the utterance sentence, and the input utterance sentence An instruction presenting unit that presents an instruction to the speaker associated with the question identification information and further presents an instruction to the responder may be included.

  In addition, the database stores a commanded word list corresponding to the command word, and when the input utterance contains a command word, the commanded word list is presented to the question side You may further include the list display part shown to a user.

  Furthermore, the conversation support method according to an aspect of the present invention accesses a database that stores a response sentence table that associates utterance sentence identification information for identifying an utterance sentence and a response sentence list listed as an assumed response to the utterance sentence. Using a computer capable of acquiring the response sentence list associated with the utterance sentence identification information of the input utterance sentence and listing and displaying the response sentences included in the response sentence list It is said.

  Further, a program according to another aspect of the present invention accesses a database that stores a response sentence table that associates an utterance sentence identification information for identifying an utterance sentence and a response sentence list enumerated as an assumed response to the utterance sentence. A computer capable of acquiring the response sentence list associated with the utterance sentence identification information of the input utterance sentence, and listing and displaying the response sentences included in the response sentence list; It is characterized by being executed.

  Embodiments of the present invention will be described with reference to the drawings. The conversation support apparatus according to the embodiment of the present invention can be realized using a portable game machine, for example. As shown in FIG. 1, the portable game machine includes a control unit 11, an interface unit 12, an operation unit 13, a wireless communication module 14, a disk drive 15, a storage unit 16, and a power control unit 17. The image processing unit 18, the display 19, the sound processing unit 20, the sound output unit 21, and the sound input unit 22 are configured.

  Here, the operation unit 13, the wireless communication module 14, the disk drive 15, the storage unit 16, and the voice input unit 22 are connected to the control unit 11 via the interface unit 12. The power control unit 17, the image processing unit 18, and the sound processing unit 20 are connected to the control unit 11, and a display 19 is further connected to the image processing unit 18. An audio output unit 21 is connected to the audio processing unit 20.

  The control unit 11 is an MPU (Micro Processing Unit) or the like, and operates according to an external storage medium M set in the disk drive 15 or a program stored in the storage unit 16. In the present embodiment, a program for realizing a conversation support device by this portable game machine is stored and supplied in the external storage medium M, and the control unit 11 reads the program from the external storage medium M. Execute.

  The external storage medium M stores a database used in processing as a conversation support device. The contents of the processing of the control unit 11 and the contents of the database will be described in detail later.

  The interface unit 12 distributes the signal output from the control unit 11 to any one of the wireless communication module 14, the disk drive 15, or the storage unit 16 in accordance with an instruction input from the control unit 11 and outputs the signal. Further, the interface unit 12 outputs a signal from the operation unit 13, the wireless communication module 14, the disk drive 15, the storage unit 16, or the voice input unit 22 to the control unit 11 while adjusting the signal to the control unit 11. Yes.

  The operation unit 13 includes game operation buttons, setting buttons, and the like. By operating various buttons and the like included in the operation unit 13, the player operates the game. The operation unit 13 outputs a signal representing the content of the player's operation to the interface unit 12.

  The wireless communication module 14 is a wireless communication device defined as, for example, 802.11b or g, and radiates and transmits data to be transmitted wirelessly in accordance with an instruction input from the control unit 11. Further, a signal arriving at an antenna (not shown) is received and output to the control unit 11.

  The disk drive 15 reads data from the set external storage medium M according to an instruction input from the control unit 11 and outputs the data to the control unit 11. The storage unit 16 is a storage element such as a RAM or a ROM, holds a program executed by the control unit 11, and also operates as a work memory of the control unit 11.

  The power control unit 17 supplies power to each unit. The image processing unit 18 draws a screen to be displayed on the display 19 in accordance with an instruction input from the control unit 11, and outputs a video signal for displaying the drawn screen to the display 19. The display 19 is a liquid crystal display, for example, and controls the liquid crystal based on the video signal input from the image processing unit 18 to display a video.

  The voice processing unit 20 synthesizes voice according to instructions input from the control unit 11 and outputs a synthesized voice signal to the voice output unit 21. Here, the sound represents various sounds such as a beep sound, a musical instrument sound, and a human voice. The audio output unit 21 is, for example, a speaker or a headphone terminal, and sounds the speaker based on the audio signal input from the audio processing unit 20 or outputs the audio signal to the headphone terminal.

  The voice input unit 22 is, for example, a microphone, and converts surrounding voice into a voice signal that is an electrical signal and outputs the voice signal.

Database contents
In this embodiment, the database stores at least an utterance sentence database, a response sentence database, a response sentence table, and a commanded word list. Here, each database is exemplified below.

(1) Sentence sentence database First, the utterance sentence database includes an utterance sentence entity table group G for each national language such as Japanese, English, etc., as shown in FIG. The utterance sentence entity table group G further includes an utterance sentence entity table T for each scene identifier such as “airport / in-flight”, “shopping”, etc.

In the utterance sentence entity table, an utterance sentence entity, voice data, and a keyword are recorded in association with an utterance sentence identifier. Here, the substance of the spoken sentence is a character string that describes a sentence to be actually spoken, and the voice data is data that records a voice when the character string is spoken in each language. This audio data is recorded in various audio data formats such as MP3 and AIFF. Furthermore, the keyword is necessary when voice input is performed as a method of accepting an utterance, and this information is not necessarily required when voice input is not performed. As a specific example, this keyword may be phoneme data to be compared during speech recognition. Moreover, this keyword may be a part of the substance of the utterance sentence, for example. In other words, the substance of the spoken sentence is
“When is this shop closed?”
As a keyword,
"Store", "Regular holiday", "When"
Etc. Further, the keyword may be a word that may be uttered in relation to the utterance sentence. For example, in the example of “When is this shop's regular holiday?”, As a synonym for “regular holiday”,
"Closed", "Closed day",
Such expressions may be included as keywords.

  Each utterance sentence entity table T registers in advance information related to an utterance sentence assumed to be used for each scene identified by each scene identifier.

(2) Response sentence database The response sentence database is similar to the utterance sentence database, and includes a response sentence entity table for each national language such as Japanese, English, and so on. In the response sentence entity table, the response sentence entity is recorded in association with the response sentence identifier. Here, the entity of the response sentence is a character string describing a sentence to be actually spoken. In the present embodiment, for each national language such as Japanese, English,..., A table (response voice data table) in which voice data corresponding to a response sentence (response voice data) and an identifier of the response voice data are associated with each other. May be recorded in a database. In this case, an identifier of response voice data corresponding to each response statement identifier may be included in the response statement entity table in association with each other (FIG. 3).

(3) Response sentence table As shown in FIG. 4, the response sentence table includes an utterance sentence identifier as utterance sentence identification information, a response sentence list enumerated in advance as an assumed response to the utterance sentence identified thereby, Alternatively, it is associated with specific response availability information indicating whether or not a response with a word representing negation is possible.

Specific examples include utterances,
“When is this shop closed?”
The spoken sentence identifier for
A list of response sentences such as “It is Sunday” or “I am always open” is associated. In the present embodiment, a response statement identifier is associated as a list of response statements, not the response statement itself.

  Furthermore, in the above example, “When is this store's regular holiday?”, It is not possible to respond with a word indicating affirmation or denial such as “Yes” or “No”. Information (“N”) indicating that it is impossible is associated.

For example, another example is an utterance that can be answered with “yes” or “no”,
“Would it be a little cheaper?”
The spoken sentence identifier for
A list of response sentences (response sentence identifiers corresponding to each response sentence) such as “No” or “I will consult with the store manager” is associated, and a response with a word indicating affirmation or denial as specific response availability information Information indicating that it is possible ("Y") is associated.

(4) Instructed word list As shown in FIG. 5, the instructed word list includes an instruction word in each scene for each same scene identifier in the speech database such as “airport / in-flight”, “shopping”, etc. Words corresponding to those that may be expressed by are enumerated and registered. Each word holds a representation in each language as a character string in association with each other.

  Here, the instruction word is a word that is replaced by a word such as “this”, “that”, or “it”. In addition, the instruction word is a word that can be expressed by an instruction word. For example, in the “hotel” scene, facilities such as “television”, “telephone”, “internet connection terminal”, etc. can be expressed as instructions in the hotel, so these words are designated words. Register in the list.

(5) Database Storage In the present embodiment, these databases are stored in the external storage medium M and can be accessed from the control unit 11 via the disk drive 15.

[Accept speech]
In the conversation support apparatus according to the present embodiment, conversation parties operate alternately to establish a conversation. In general, an operation is alternately performed between a side that speaks to start a conversation (speaker side) and a side that responds to the side (responder side). The speaker side performs in advance an instruction operation for selecting the language used by the speaker and the language used by the responder. In response to the instruction operation, the control unit 11 stores information representing the speaker side language (question side language information) and information representing the responder side language (response side language information) in the storage unit 16.

  In addition, the speaker selects a scene to be used. This selection shall select one of the scenes corresponding to the scene identifier in the utterance sentence database. The control unit 11 stores scene identification information for specifying the scene selected by the speaker in the storage unit 16.

  In the present embodiment, the processing performed by the control unit 11 as a conversation support device is functionally performed as shown in FIG. 6, an utterance sentence accepting unit 31, an utterance sentence candidate selection unit 32, and an utterance sentence candidate presentation unit. 33, a response sentence candidate search unit 34, a response sentence candidate addition unit 35, a response sentence candidate presentation unit 36, and a response output unit 37.

  For example, the speaker utters to the voice input unit 22 and inputs an utterance sentence. The utterance sentence receiving unit 31 compares the voice signal input from the voice input unit 22 with the phoneme data related to the keyword registered in the utterance sentence database, and the keyword included in the input utterance sentence. Is detected. Since this process can use a general process as a voice recognition process as it is, a detailed description thereof is omitted here.

  Note that the utterance sentence accepting unit 31 may accept an input of an utterance sentence as character string data via the operation unit 13. Also in this case, the utterance sentence receiving unit 31 detects a keyword included in the utterance sentence input as the data of the character string. Since this process is also widely known as a process for searching for a character string match, a detailed description thereof will be omitted.

  Then, the utterance sentence candidate selection unit 32 receives the detected keyword, refers to the utterance sentence entity table corresponding to the selected language (the language corresponding to the question side language information) and the selected scene, and uttered. For each sentence, the number of detected keywords among the associated keywords is counted.

For example,
"When is this shop closed?"
When the conversation-like question is input, the control unit 11 detects the keywords “store”, “when”, and “rest” from here. At this time,
“When is this shop closed?”
If the keywords such as “store”, “regular holiday”, “vacation”, and “when” are associated with the utterance statement, the utterance sentence of “When is this shop regular holiday?” Will have a count of “3”.

Also,
“Is this store also in Japan?”
Is associated with keywords such as “Store”, “Japan”, and “Yes”, only “Store” matches the keywords detected from the above question. “1”.

  Also, here, an example has been described in which an utterance sentence candidate is selected according to the number of detected keywords. However, among the pronunciation contents that appear in the sentence, a part that is easy to recognize is set as a keyword, and the keyword is detected. An utterance sentence candidate may be selected depending on whether or not. That is, in this case, for example, keywords such as “bar”, “can drink”, and “where's the bar” are keywords for speech sentence candidates such as “where is the bar”. Is assigned.

  The utterance sentence accepting unit 31 accepts an input of an utterance sentence as character string data via the operation unit 13 and detects a keyword included in the utterance sentence input as the character string data. Then, the utterance sentence candidate selection unit 32 receives the detected keyword, refers to the utterance sentence entity table corresponding to the selected language (the language corresponding to the question side language information) and the selected scene, and uttered. For each sentence, the information is compared with the associated keyword, and information specifying the utterance sentence that matches the detected keyword is output to the utterance sentence candidate presentation unit 33.

[Presentation of spoken sentence candidates]
The utterance sentence candidate presentation unit 33 selects an utterance sentence based on the information input from the utterance sentence candidate selection unit 32, and uses the selected utterance sentence as the attention utterance sentence, and converts the character string of the attention utterance sentence into the utterance sentence. Read from the entity table. Then, the image processing unit 18 is instructed to display the read character string on the display 19.

For example, when the utterance sentence candidate selecting unit 32 outputs the keyword count, the utterance sentence having the maximum count is selected as the attention utterance sentence. In this example, for the question “When is this shop closed?”, If the maximum number of utterances of “When is this shop closed?” When is a regular holiday? "
Is displayed on the display 19.

Similarly, when an utterance sentence associated with the detected keyword is found and information specifying the found utterance sentence is output, the utterance sentence specified by the information is selected as an attention utterance sentence. In this example, for example, in response to the question “Where can I drink?”, The keyword “I can drink” is detected, and the utterance sentence “Where is the bar” is found, and this “Where is the bar?” "Is specified. So this "where is the bar"
Is displayed on the display 19.

  At this time, the utterance sentence candidate presenting unit 33 causes the utterer to select whether or not the selected attention utterance sentence is in line with the intention of the question. Here, when the speaker performs an instruction operation that does not conform to the intention of the question, the control unit 11 interrupts the process and waits for the input of the question.

  On the other hand, when the speaker performs an instruction operation that the attention utterance is in line with the intention, the control unit 11 refers to the responder side language information and corresponds to the language on the responder side. The voice information associated with the attention utterance is read out. Then, the audio information is output to the audio processing unit. For example, if “English” is set as the responder's language, the English voice corresponding to the utterance sentence “When is this store's regular holiday?” Is played and played through the speaker. The

[Processing for instructions]
The utterance sentence candidate presenting unit 33, when the keyword detected by the utterance sentence accepting unit 31 includes a predetermined instruction word such as “this”, “that”, “it”, etc. You may perform the process relevant to. This instruction word is also predetermined for each language.

  In this process, the utterance sentence candidate presenting unit 33 displays an instruction word list corresponding to the scene identifier of the selected scene when a predetermined instruction word is included in the detected keyword. 19 is displayed. This processing may be performed, for example, after the user performs an instruction operation that the noticed utterance is as intended.

For example, when the user selects “Hotel” as the scene,
"Is this there?"
Is input as a keyword, and an utterance sentence such as “Is this available?” Is selected as an attention utterance sentence.

  At this time, the utterance sentence candidate presenting unit 33 displays a list of words stored in the database as a directed word list corresponding to the scene identifier of “hotel” on the display 19. As a result, words such as “TV”, “telephone”, “Internet connection terminal”,... Are listed on the display 19 as facilities in the hotel. The speaker performs an instruction operation to select one of them.

  In this case, the control unit 11 refers to the responder side language information, and reads out the speech information associated with the noticed utterance corresponding to the responder side language. Then, the voice information is output to the voice processing unit 20, and the word corresponding to the language on the responder side is displayed on the display 19.

  For example, in the above hotel example, when the user selects “TV”, the speaker responds to the spoken sentence “Do you have this?” The word corresponding to “TV” is displayed on the display in the responder side language.

[Multiple candidates]
Further, the utterance sentence candidate presenting unit 33, when there are a plurality of utterance sentence candidates to be selected, such as the utterance sentence having the maximum count or the utterance sentence matching the detected keyword, the utterance sentence of the plurality of candidates. May be presented to the speaker for selection.

  Furthermore, for a plurality of utterances that are easy to be selected at the same time in the selection based on keywords (hereinafter referred to as similar utterance sentence groups), each of the similar utterance sentence groups on the utterance sentence entity table of the database is previously specified. Group identifiers may be associated.

For example, in the “Airport / Inflight” scene,
“I want to move to the seat on the aisle.”
“I want to move to the window seat”
“I want to move to another vacant seat”
In these three utterance sentences, keywords such as “seat” and “want to move” are associated with each other, and therefore, the counts relating to the detected keywords are likely to be the same. Therefore, a common group identifier is associated with them.

  Further, a table (opposite question sentence table) in which the opposite question sentence in each language, the options, and the utterance sentence identifier for each option are associated is recorded in the database (FIG. 7).

  Then, when there are a plurality of utterance sentences with the maximum count, the utterance sentence candidate presenting unit 33 refers to the group identifier associated with each utterance sentence and checks whether they are common to each other. Here, if they are common to each other, the utterance sentence candidate presenting unit 33 selects the one corresponding to the speaker side language among the opposite question sentences and options registered in the opposite question sentence table in association with the group identifier. The counter question message is displayed on the display 19 and / or the voice of the counter question message is output to the voice processing unit 20 to sound. Also, the options are displayed on the display 19 and the speaker is selected.

For example, if the speaker
"I want to move my seat"
When a question is input with such voices, the keywords “seat” and “want to move” are detected. And the counts for these detected keywords are as in the example above:
“I want to move to the seat on the aisle.”
“I want to move to the window seat”
“I want to move to another vacant seat”
For all three utterances, “2” is the maximum.

  Since all the group identifiers associated with these utterances are the same (common), the utterance sentence candidate presenting unit 33 reads the opposite question sentence and the option associated with the group identifier. At this time, if the language on the speaker side is Japanese, the opposite question sentence in Japanese and the option are read out.

  For example, the opposite question sentence is written in advance according to the utterance sentence candidate, such as “Which seat do you want to move to?”. Each option is displayed in correspondence with the above three utterances, such as “passage side”, “window side”, “other vacant seats”, and the like.

  When the speaker performs an instruction operation to select one of these options, the utterance sentence candidate presenting unit 33 reads an utterance sentence identifier registered in association with the selected option. Further, the utterance sentence candidate presenting unit 33 refers to the responder side language information, and reads out speech information associated with the read utterance sentence identifier and corresponding to the responder side language. Then, the audio information is output to the audio processing unit.

Thus, for example, when the speaker selects “Aisle side” from the above options,
“I want to move to the seat on the aisle.”
Voice information corresponding to the utterance sentence is sounded.

[Emotion level expression]
Further, the utterance sentence candidate presenting unit 33 may add an expression that emphasizes the utterance sentence according to an instruction of the speaker. In this case, in the database stored in the external storage medium M, the emphasized expression word list and the corresponding voice data are stored in association with each other as the emphasized expression word list. In this case, as a voice expression, even if it is the same word meaning “extremely”, it may be expressed cheerfully or positively as “very delicious” (such as physical condition) It is different from expressing it negatively or negatively like “very bad”. Therefore, in this emphasis expression word list, as shown in FIG. 8, for each language, the emphasis expression word and the corresponding sound data include sound data in the positive case, sound data in the negative case, Are associated with each other. In FIG. 8, four types of words “high”, “so-so”, “pretty”, and “very” are shown as highlighted words.

  The utterance sentence candidate presenting unit 33 selects these emphasized expressions on a screen for inputting whether or not the attention utterance sentence is a desired one. If the user does not select the emphasized expression, the operation described above is performed. Also, when selecting an emphasis expression, first select either positive or negative, then present a list of emphasis expression words registered in the database and determine which emphasis expression to use. Let them choose.

  According to this, for example, when the attention utterance is “delicious”, if the speaker is positive and selects “very”, the sentence in the responder side corresponding to “delicious” is uttered. At the same time, a sound is produced based on the speech data of the responder side language corresponding to “very” in a positive expression.

  In the utterance sentence table, for each utterance sentence, a flag indicating whether or not such emphasized expression is possible is associated, and the utterance sentence associated with the flag that the emphasis expression is possible is the attention utterance sentence. The above processing may be performed only in the case where

  However, for example, by adding an emphasis expression of “very” to the utterance sentence “Is there a TV?”, It becomes a comical conversation, and the atmosphere between the parties can be guided in a better direction. Therefore, it is possible to select a word of emphasized expression regardless of whether the emphasized expression is possible grammatically or semantically.

[Present response]
When the utterance sentence candidate presentation unit 33 completes the process of ringing voice information corresponding to the utterance sentence, the process of the response sentence candidate search unit 34 is started.

  The response sentence candidate search unit 34 receives the input of the utterance sentence identifier of the utterance sentence sung by the utterance sentence candidate presentation unit 33, refers to the database, and is recorded in the response sentence table in association with the utterance sentence identifier. The response sentence list corresponding to the responder side language is acquired.

  The response sentence candidate adding unit 35 receives the input of the utterance sentence identifier of the utterance sentence sung in the utterance sentence candidate presenting part 33, and refers to the database to acquire the specific response availability information associated with the utterance sentence identifier. . If the specific response availability information is information (“Y”) indicating that a response with a word indicating affirmation or denial is possible, the responder is included in the response message list acquired by the response message candidate search unit 34. Add words that indicate affirmation or denial in the side language and output. Here, the word indicating affirmation or negation is a word such as “Yes” or “No” in English, for example.

  In addition, when the acquired specific response availability information is information (“N”) indicating that the response with a word indicating affirmation or denial is impossible, the response sentence candidate adding unit 35 determines that the response sentence candidate searching unit 34 The acquired response sentence list is output as it is.

  The response sentence candidate presenting unit 36 acquires the response sentence character string from the response sentence database based on the response sentence list output by the response sentence candidate adding unit 35 and causes the display 19 to display the response sentence character string.

  Then, the responder selects one of the response sentences from the character string of the response sentence displayed on the display 19 and speaks. At this time, an instruction operation for selecting one of the response sentences is performed, and the response output unit 37 responds to the speaker side language among the response voice data corresponding to the selected response sentence. The data may be read out and output to the voice processing unit 20, and a voice based on the response voice data may be played through a speaker.

  Also, the responder reads out one of the response sentences to the voice input unit 22, the control unit 11 determines which response sentence is input from the input voice, and displays the determination result. 19 may be displayed. In this case as well, for speech recognition, a speech recognition keyword is associated with each response sentence, and among the displayed response sentence candidates, the response sentence associated with the most keywords included in the input speech is read out. You may make it judge that it was.

[Operation]
The conversation support apparatus according to the embodiment of the present invention has the above-described configuration and operates as follows. Here, for example, it is assumed that a speaker who speaks Japanese as a native language has a conversation with a responder whose native language is English. The speaker side operates the conversation support device to set “Japanese” as the speaker side language and “English” as the responder side language. In addition, a scene is selected according to the place to be used. Here, for example, it is assumed that “meal” is selected as the scene.

For example, when the speaker wants to know what is recommended by the store among the items listed in the menu, the speaker speaks to the voice input unit 22 of the conversation support device, such as “Please tell me the recommendation”. Do. Here, the conversation support apparatus according to the present embodiment selects the most likely one from a predetermined utterance sentence group using a keyword included in the input utterance sentence instead of the entire inputted utterance sentence. It is not necessary to input the same utterance sentence registered in the database. For example, the conversation support device itself is treated as an interpreter like an interpreter,
“Ask me for recommendations”
In this way, input may be performed using an instructional expression.

The conversation support device detects a keyword from the input information. Here, “recommended” is detected from a predetermined keyword list, and the utterance sentence associated with the detected keyword is searched from the utterance sentence list corresponding to the “meal” scene. Here is "Do you recommend it?"
Utterance sentence is found.

  The conversation support apparatus displays the result of this search and displays the inquiry about whether or not this is the desired content (FIG. 9). When the speaker confirms that the speaker is the desired one and performs an instruction operation to that effect, the conversation support device will receive the voice data in the responder side language “English” associated with the utterance. Read and sound a sound based on the sound data. At this time, a character string in the responder side language corresponding to the spoken sentence may be displayed on the display 19. When the word for emphasis is selected, it may be displayed similarly.

  Further, the conversation support apparatus reads a response sentence list in the response side language recorded in association with the identifier of the utterance sentence. Here, the question “Do you recommend it?” Can be answered with affirmative or negative words such as “Yes” or “No”, so “Yes” appears in the recorded response list. , Responding language (English) words corresponding to “No” are added and presented.

Thus, for example, on the display 19 in English, with an instruction to select a response, such as “Please select an answer from the following”,
"Yes"
"No"
"Recommended is sold out"
"It's written on the menu"
Etc. are displayed (FIG. 10).

  The responding user selects a response sentence from this option and reads it out.

And if the speaker wants a spoon while eating,
“Is this there?”
When the utterance sentence is input, the conversation support apparatus detects keywords such as “this” and “present” from the input utterance sentence. And the utterance sentence to which the detected keyword is most associated,
"Is this this?"
And ask the speaker if this is the desired content. When the speaker confirms that the speaker is the desired one and performs an instruction operation to that effect, the conversation support apparatus further detects that the instruction word “this” is included in the detected keyword, The instructed word list associated with the “meal” scene is displayed to prompt the speaker to select. The speaker selects the desired word “spoon” from the list. Then, the conversation support device reads the voice data in the responder side language “English” associated with the uttered sentence, sounds the voice based on the voice data, and selects “ An English character string corresponding to “spoon” is displayed on the display 19 (FIG. 11). Here is a sentence in English corresponding to the spoken sentence,
"Do you have this?"
Is also displayed.

  The responder obtains the content of the utterance from the word displayed on the display 19 and the uttered voice.

  Further, the conversation support apparatus reads a response sentence list in the response side language recorded in association with the identifier of the utterance sentence. The above questions can be answered with affirmative or negative words such as “Yes” or “No”, so the responding language corresponding to “Yes” or “No” in the recorded response sentence list Add (English) words to present.

Thus, for example, on the display 19 in English, with an instruction to select a response, such as “Please select an answer from the following”,
"Yes"
"No"
"I will bring it now."
And other response options are displayed.

  The responding user selectively reads out the response sentence from this option. The control unit 11 of the conversation support apparatus recognizes the content of the reading out by voice and displays on the display 19 which response sentence has been selected. Thereby, the conversation between the speaker and the responder is established via the conversation support device.

[Instructions for parties]
Further, the control unit 11 according to the present embodiment may ask the speaker and the responder for a predetermined work in relation to the utterance sentence presented by the utterance sentence candidate presentation unit 33. In this case, in the database stored in the external storage medium M, as shown in FIG. 12, the spoken sentence identifier, the work instruction to the speaker side in the national language, and the responder side in the national language An instruction table in which work instructions are associated is further held.

  When the control unit 11 performs an instruction operation to the effect that the speaker is the desired one for the attention utterance sentence presented by the utterance sentence candidate presentation unit 33, the control part 11 sets the utterance sentence identifier of the attention utterance sentence. It is checked whether or not the associated work instruction is registered in the instruction table.

  Here, when the work instruction is registered in the instruction table, the control unit 11 reads out the work instruction to the speaker side in the set speaker side language and displays the work instruction on the display 19. In addition, when the response sentence list is displayed on the display 19 in association with the utterance sentence, the control unit 11 sends a work instruction to the responder associated with the utterance sentence identifier of the utterance sentence. Of these, the language in the responder side is read and displayed together on the display 19.

  For example, in the “meal” scene as in the above example, for an utterance sentence such as “Which is recommended?”, “Indicates a menu as a work instruction to the speaker side in the instruction table. Please associate a work order such as “Please”. In addition, as a work instruction to the responder side, a work instruction such as “Please show with menu” is associated.

  Thus, for example, when the utterance is selected, the speaker shows the menu in accordance with the work instruction “Please show menu”, and the responder gives the work instruction “Please show in menu”. With reference to the menu, a response sentence such as “This is” can be selectively read out as a response sentence, for example, and conversation can be performed.

[download]
Furthermore, the contents of the database in the present embodiment may be downloaded from a server on the network via the wireless communication module 14, for example. As a specific example, if an instruction word list can be downloaded, for example, not only words according to the scene but also words according to the times can be included.

  According to the present embodiment, the response sentences are presented as a list, and the conversation is established by selectively reading out to the other party. Therefore, it is possible to avoid a situation in which the contents spoken by the other party cannot be heard or understood. In addition, because the instruction word is detected and a list of the commanded words is presented and selected, a conversation can be established using the command word in combination with the list of commanded words. Need not be included in the database. This makes it possible to reduce the size of the database. Thus, according to the conversation support apparatus of the present embodiment, the convenience can be improved.

It is a configuration block diagram showing an example of a conversation support device according to an embodiment of the present invention. It is explanatory drawing showing the example of the speech sentence database which concerns on embodiment of this invention. It is explanatory drawing showing the example of the response sentence database which concerns on embodiment of this invention. It is explanatory drawing showing the example of the response sentence table which concerns on embodiment of this invention. It is explanatory drawing showing the example of the commanded word list which concerns on embodiment of this invention. It is a functional block diagram showing the example of the conversation assistance apparatus which concerns on embodiment of this invention. It is explanatory drawing showing the example of the opposite question sentence table which concerns on embodiment of this invention. It is explanatory drawing showing the example of the list | wrist of the emphasized word which concerns on embodiment of this invention. It is explanatory drawing showing the example of a screen of the conversation assistance apparatus which concerns on embodiment of this invention. It is explanatory drawing showing the example of a screen of the conversation assistance apparatus which concerns on embodiment of this invention. It is explanatory drawing showing the example of a screen of the conversation assistance apparatus which concerns on embodiment of this invention. It is explanatory drawing showing the example of the instruction | indication table which concerns on embodiment of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Portable game machine, 11 Control part, 12 Interface unit, 13 Operation unit, 14 Wireless communication module, 15 Disk drive, 16 Storage part, 17 Power supply control part, 18 Image processing unit, 19 Display, 20 Voice processing unit, 21 Voice output unit, 22 voice input unit, 31 utterance sentence acceptance unit, 32 utterance sentence candidate selection part, 33 utterance sentence candidate presentation part, 34 response sentence candidate search part, 35 response sentence candidate addition part, 36 response sentence candidate presentation part, 37 Response sentence output part.

Claims (6)

A database storing a response text table in which speech text identification information for identifying a speech text and a response text list enumerated as an assumed response to the speech text are associated with each other;
A response sentence acquisition unit that acquires the response sentence list associated with the utterance sentence identification information of the input utterance sentence;
A response text display unit that lists and displays the response texts included in the response text list;
A conversation support apparatus characterized by comprising: The conversation support device according to claim 1,
The response sentence table further stores specific response availability information indicating whether or not a response with a word representing affirmation or denial is associated with the spoken sentence identification information,
With reference to the specific response availability information associated with the question identification information of the input utterance sentence, when a response with a word representing affirmation or denial is possible, the acquired response sentence list is affirmed or Response sentence addition part for adding negative words,
Further comprising: a conversation support device. The conversation support device according to claim 1 or 2,
The database stores an instruction table that associates an instruction to the speaker side and an instruction to the responder side with respect to the utterance sentence identification information for identifying the utterance sentence,
An instruction presentation unit that presents an instruction to the speaker side associated with the question identification information of the input utterance sentence, and further presents an instruction to the responder side;
Conversation support apparatus characterized by including. The conversation support device according to any one of claims 1 to 3,
The database stores a commanded word list corresponding to a command word,
A list display unit that presents the instructed word list and presents it to the question side user when an instruction word is included in the input utterance sentence;
Further comprising: a conversation support device. Using a computer that can access a database that stores a response sentence table that associates an utterance sentence identification information that identifies an utterance sentence and a response sentence list listed as an assumed response to the utterance sentence,
Obtaining the response sentence list associated with the utterance sentence identification information of the input utterance sentence;
List and display the response texts included in the response text list.
A conversation support method characterized by that. To a computer that can access a database that stores a response sentence table that associates an utterance sentence identification information for identifying an utterance sentence and a response sentence list enumerated as an assumed response to the utterance sentence,
A procedure for acquiring the response sentence list associated with the spoken sentence identification information of the input utterance sentence, and a procedure for listing and displaying the response sentences included in the response sentence list;
A program characterized by having executed.

Images