Conversation Device

Abstract

To provide a conversation device that allows even an elderly person to enjoy conversation, gain useful information, and be comforted.SOLUTION: A conversation device 1 includes: a storage unit 2; a plurality of answer buttons 31, 32, 33, and 34; a speaker 5; a microphone 6; and a control unit 4 that recognizes speech uttered by a user, outputs from the speaker 5 one of questions about daily life, one's body, one's mental health, and a condition, according to the recognized speech, accepts answers for the questions by pressing the plurality of answer buttons 31, 32, 33, and 34, and outputs next questions or answers from the speaker 5 according to the type of the answer buttons 31, 32, 33, and 34 having been pressed. When outputting a question regarding the condition, the control unit 4 asks a question that distinguishes whether the condition is good or bad. If the condition is bad, the control unit 4 asks a question that distinguishes whether a physical or mental condition is bad.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a conversation device for conducting a pseudo conversation with a user. [Background technology]

[0002] The conversational robot described in Patent Document 1 is a robot that responds to human words, and is characterized by comprising a storage medium in which a plurality of response words are stored for each of a plurality of pre-specified words, a voice recognition means for recognizing input voice of the above-mentioned words from a microphone means, a means for generating situation information at the time of the response, and a means for, when the input voice is recognized by the voice recognition means, reading out from the storage medium and reproducing the optimal response word for the recognized voice based on information from the generation means.

[0003] In the simulated conversation system described in Patent Document 2, the user information acquisition unit 234 acquires a user profile and generates user data 516. The reply determination unit 232 determines an adult or child backchannel and a next question according to the user profile. Next, a question to start a conversation is output by the speaker 30, and when a voice is input in response to the output question, the voice recognition unit 210 recognizes the voice (input word) (determines a registered word corresponding to the voice). The reply determination unit 232 determines a backchannel and a next question according to the registered word from the determined backchannel group and next question group, and the voice synthesis unit 220 synthesizes the voice of the determined backchannel and next question and outputs it by the speaker 30, thus continuing the conversation. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Unexamined Patent Publication No. 167997 / 1983 [Patent Document 2] JP 2002-169591 A Summary of the Invention [Problem to be solved by the invention]

[0005] The devices described in Patent Documents 1 and 2 recognize the voice uttered by the user and determine questions and responses. However, elderly people are prone to decline in various functions related to language communication, such as unclear speech and reduced language comprehension. Therefore, it may be difficult for elderly people to respond by voice to questions posed by the devices described in Patent Documents 1 and 2.

[0006] In addition, conventional devices may not contain the information that the elderly want. For example, elderly people who live alone may feel lonely, physically weakened, or anxious about their own death. However, conventional devices do not contain conversations about these topics.

[0007] Therefore, an object of the present invention is to provide a conversation device that allows even elderly people to enjoy conversation, obtain useful information, and receive comfort. [Means for solving the problem]

[0008] In order to solve the above problems, the present invention has the following features. The present invention is a conversation device for conducting a simulated conversation with a user, and includes a memory unit for storing words that can be voice recognized, information on questions, answers, and responses regarding daily life, information on questions, answers, and responses regarding the body, information on questions, answers, and responses regarding mental health, and information on questions regarding one's condition, a plurality of answer buttons for inputting the user's answers, a speaker for outputting voice, a microphone for inputting the user's speech, and a control unit that recognizes the voice spoken by the user and outputs from the speaker one of a question regarding daily life, a question regarding the body, a question regarding mental health, or a question regarding one's condition depending on the recognized voice, accepts answers to questions by pressing one of a plurality of answer buttons, and outputs the next question or response from the speaker depending on the type of answer button pressed.

[0009] The control unit When outputting questions about your condition, ask questions that distinguish between good and bad conditions. If the condition is good, questions about daily life will be output from the speaker. If the person is not feeling well, ask questions to determine whether the problem is physical or mental. If the problem is related to physical condition, questions about the body are output from the speaker. If the problem is mental health issues, questions about mental health are output from the speaker.

[0010] Preferably, the conversation device further comprises a repeat playback button, and when the repeat playback button is pressed, the control unit re-outputs the previously output voice.

[0011] Preferably, the conversation device further comprises a voice mode button which, when pressed during the progress of a conversation, can return the operation of the control unit to the initial voice recognition operation.

[0012] Preferably, the control unit recognizes the voice spoken by the user and compares it with words registered in the memory unit to determine whether the recognized voice is related to daily life, the body, or mental health, or whether it is a question that requires asking about the user's condition, and outputs from the speaker one of a question about daily life, a question about the body, a question about mental health, or a question about the user's condition. Effect of the Invention

[0013] According to the present invention, the conversation device can ask questions, answer, and respond in accordance with the situation that the elderly person is currently facing. Therefore, a conversation device is provided that allows even the elderly person to enjoy conversation, obtain useful information, and receive comfort.

[0014] These and other objects, features, aspects and advantages of the present invention will become more apparent from the following detailed description taken in conjunction with the accompanying drawings. [Brief description of the drawings]

[0015] [Figure 1] FIG. 1 is a block diagram showing the functional configuration of a conversation device 1 according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a flowchart showing the operation of conversation device 1. [Diagram 3] FIG. 3 is a diagram showing an example of information contained in the recognizable voice database stored in the storage unit 2. As shown in FIG. [Figure 4] FIG. 4 is a flowchart showing the operation of the daily life topic QAR process in FIG. [Diagram 5] FIG. 5 is a flowchart showing the operation of the body topic QAR processing in FIG. [Figure 6] FIG. 6 is a flowchart showing the operation of the mental topic QAR process in FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] 1 is a block diagram showing the functional configuration of a conversation device 1 according to an embodiment of the present invention. In FIG. 1, conversation device 1 includes a storage unit 2, an operation unit 3, a control unit 4, a speaker 5, and a microphone 6.

[0017] The storage unit 2 includes a daily QAR database 21, a physical QAR database 22, a mental QAR database 23, a condition-related question database 24, and a recognizable voice database 25. Here, Q means a question. A means an expected answer to the question. The expected answers are typically the number that can be answered by pressing the first answer button 31, the second answer button 32, the third answer button 33, and the fourth answer button 34. However, if there are five or more expected answers, it is preferable to provide an expected answer called "others" and, for example, to assign "others" to the fourth answer button so that when the fourth answer button is pressed, another expected answer can be specified by the button. R means a response to the expected answer.

[0018] The daily QAR database 21 stores a large number of questions related to daily life, expected answers to the questions, and replies to the expected answers. It is not necessary that all questions, expected answers, and responses are associated with one set. For example, only expected answers corresponding to questions may be prepared, and information with no response to the expected answers may be stored in the database 21. In the case of information with no response, the control device 4 may ask a question immediately after the expected answer without responding. In addition, the responses include a response that generally describes daily life. The general response should not cause negative feelings in the elderly as much as possible, such as, for example, content that conveys the wonderfulness of continuing to live alone, but is not limited to this. The general response does not need to be associated with a question or an expected answer. A plurality of patterns of expected answers to questions may be prepared so that the user does not get bored. A plurality of patterns of responses to the expected answers may also be prepared.

[0019] The body QAR database 22 stores a large number of questions related to the body, expected answers to the questions, and responses to the expected answers in advance. It is not necessary that all questions, expected answers, and responses are associated with one set. For example, only expected answers corresponding to questions may be prepared, and information with no response to the expected answers may be stored in the database 22. In the case of information with no response, the control device 4 may ask a question immediately after the expected answer without responding. In addition, the responses include a response that generally describes the body. The general response may be, but is not limited to, a proposal for a self-management method for health. The general response may not be associated with a question or an expected answer. A plurality of patterns of expected answers to questions may be prepared so that the user does not get bored. A plurality of patterns of responses to the expected answers may also be prepared.

[0020] The mental QAR database 23 stores a large number of questions related to mental health (mental health), expected answers to the questions, and responses to the expected answers. It is not necessary that all questions, expected answers, and responses are associated with one set. For example, only expected answers corresponding to questions may be prepared, and information with no response to the expected answers may be stored in the database 23. In the case of information with no response, the control device 4 may ask a question immediately after the expected answer without responding. In addition, the responses include a response that generally describes mental health. The general response may be a message such as "It's a problem that everyone has" or "It was happy in the end," but is not limited thereto. The general response may not be associated with a question or an expected answer. A plurality of patterns of expected answers to questions may be prepared so that the user does not get bored. A plurality of patterns of responses to the expected answers may also be prepared.

[0021] The condition-related question database 24 stores one or more questions for determining whether the condition is good or bad. In addition, when the condition is bad, one or more questions for determining whether the condition is bad due to physical reasons or mental reasons are stored. Expected answers to the database 24 are also stored in the database 24.

[0022] Although not shown, conversation device 1 has a clock unit that indicates the date and time. Memory unit 2 stores past answers together with the date and time indicated by the clock unit. Control unit 4 may refer to the past answers stored in memory unit 2 to change the content of the question, but this is not essential to the present invention.

[0023] The recognizable voice database 25 stores information about voices that can be recognized by the control unit 4. Fig. 3 is a diagram showing an example of information contained in the recognizable voice database stored in the storage unit 2. Note that Fig. 3 shows only an example and does not limit the present invention. Recognizable voices are divided into four groups here, but this is not limiting.

[0024] In the set of questions about how one is feeling, recognizable sounds include, for example, "good morning" and "good night." In the set of questions about daily life, recognizable sounds include, for example, "myself," "health tips," "diet," and "hobbies." In the set of questions about the body, recognizable sounds include, for example, "my knee hurts," "high blood pressure," "I often choke," and "I got burned." In the set of questions about mental health, recognizable sounds include, for example, "worry," "loneliness," "anxiety," and "anger."

[0025] The operation unit 3 includes a first answer button 31, a second answer button 32, a third answer button 33, a fourth answer button 34, a repeat playback button 35, and an audio mode button 36. The first answer button 31, the second answer button 32, the third answer button 33, and the fourth answer button 34 are buttons for answering the expected answer to the question. For example, if the question is, "What do you do to stay healthy? If you want to 'get plenty of sleep,' press button 1; if you want to 'exercise more,' press button 2; if you want to 'be mindful of your diet,' press button 3; and if you want to 'quickly forget about unpleasant things,' press button 4.", the corresponding buttons are the buttons for answering.

[0026] The repeat playback button 35 is a button for re-outputting the previously output sound. When the repeat playback button 35 is pressed, the control unit 4 outputs the previously output sound from the speaker 5 again.

[0027] The voice mode button 36 is a button for returning to step S102 (start of voice recognition mode) in Fig. 2 described later. When the voice mode button 36 is pressed, the control unit 4 returns to step S102 (start of voice recognition mode) in Fig. 2.

[0028] Fig. 2 is a flowchart showing the operation of the conversation device 1. The operation of the conversation device 1 will be described below with reference to Fig. 2. First, when the power is turned on, the control unit 4 causes the speaker 5 to output a greeting such as "Hello" or "Good morning" that matches the date and time when the power was turned on (step S101). At this timing, the voice recognition mode starts (step S102). In the voice recognition mode, the control unit 4 can recognize voice input from the microphone 6.

[0029] In the voice recognition mode, suppose that the user speaks something. The control unit 4 recognizes the voice input from the microphone 6 and compares it with the recognizable voice database (step S103). If there is no information matching the recognized voice in the recognizable voice database, the control unit 4 waits as it is. If there is information matching the recognized voice in the recognizable voice database, the control unit 4 proceeds to process the question, answer, and response corresponding to the recognized voice (step S105, S106, S107, or S108 in FIG. 2).

[0030] If the recognized voice is related to daily life, the control unit 4 refers to the daily life QAR database 21 and executes daily life QAR processing (step S105). If the recognized voice is related to the body, the control unit 4 refers to the body QAR database 22 and executes physical topic QAR processing (step S106). If the recognized voice is related to mental matters, the control unit 4 refers to the mental QAR database 23 and executes mental topic QAR processing (step S107). The daily life QAR processing, physical topic QAR processing, and mental topic QAR processing are the processing shown in Figs. 4, 5, and 6, respectively, and will be described later.

[0031] If the recognized voice belongs to a set of questions about how the user is feeling, the control unit 4 refers to the health question database 24 and asks a question about how the user is feeling (step S108). The control unit 4 first outputs a question from the speaker 5 to distinguish whether the user is feeling good or bad. For example, in response to the user's voice input of "Good morning," the control unit 4 outputs the question "Good morning. How are you feeling today?".

[0032] The user answers the above question by voice. Alternatively, at this stage, the control unit 4 may switch the operation from the voice mode to a mode in which the user answers by button (called the "button mode"). The control unit 4 judges whether the recognized answer is indicative of a good or bad condition (step S109).

[0033] If it is acceptable, the control unit 4 proceeds to the daily life QAR process (step S110).

[0034] On the other hand, if the feeling is bad, the control unit 4 outputs a question to distinguish whether the feeling of unwell is related to the body or to the mind (step S111). For example, the control unit 4 outputs, "You say you are not feeling well. What is it that is bad about you? If you have pain in your body, press button 1. If you are not feeling well, press button 2. If you have a problem, press button 3. If something bad has happened, press button 4."

[0035] Although the mode has been switched to button mode here, it is also possible to leave the mode in voice mode and have the user respond by voice recognition.

[0036] In the above example, when buttons 1 and 2 are pressed, the control unit 4 recognizes that the feeling of unwell is related to the physical condition, and when buttons 3 and 4 are pressed, the control unit 4 recognizes that the feeling of unwell is related to the mental condition.

[0037] If the problem is related to a physical condition, the control unit 4 proceeds to a physical QAR process (step S112). If the problem is related to a mental condition, the control unit 4 proceeds to a mental QAR process (step S113).

[0038] In this way, according to the operation flow shown in FIG. 2, when the user directly speaks about daily life, the body, or mental health, the control unit 4 proceeds to QAR processing related to daily life, the body, or mental health.

[0039] On the other hand, when the user does not directly talk about daily life, the body, or mental health, such as "Good morning," the control unit 4 asks whether the user is feeling well, and if the user is not feeling well, proceeds to QAR processing for daily life, and if the user is feeling bad, asks further questions to distinguish whether the problem is physical or mental. If the problem is physical, the control unit 4 proceeds to QAR processing related to the body, and if the problem is mental, proceeds to QAR processing related to mental health.

[0040] This allows the conversation device 1 to perform QAR (question, answer, and response) that is suited to the situation the elderly person is currently facing. Therefore, the conversation device becomes one that allows even elderly people to enjoy conversation, obtain useful information, and receive comfort.

[0041] The daily life QAR processing, physical topic QAR processing, and mental topic QAR processing will be described with reference to Figures 4, 5, and 6, respectively. Except for the fact that the content of the question, the content of the expected answer, and the content of the response are based on the respective databases 21, 22, and 23, the flow of the processing is basically the same for the daily life QAR processing, physical topic QAR processing, and mental topic QAR processing, so they will be described together.

[0042] When each QAR process starts, the button mode is entered and the control unit 4 receives input of answers from the first answer button 31 to the fourth answer button 34.

[0043] In each QAR process, the control unit 4 refers to each database 21, 22, 23 and outputs voice questions related to daily life, body, or mental health (steps S201, S301, S401). In each database 21, 22, 23, a definition is set as to what questions should be output according to the contents of the user's speech. A plurality of patterns of the definition may be prepared, and questions may be output randomly or according to the date and time, the number of times the device has been used, etc.

[0044] For example, in the case of a topic related to the body, if the user utters, "My knee hurts," the control unit 4 will output a voice question such as, "That must be terrible. If it always hurts, press button 1, and if you have injured it, press button 2, etc."

[0045] Next, when the user presses any one of the first answer button 31 to the fourth answer button 34 (steps S202 to S205, S302 to S305, S402 to S405), the control unit 4 outputs a voice of a question or response corresponding to the pressed button (steps S206, S306, S406).

[0046] For example, in the knee example above, when button 1 is pressed, the control unit 4 outputs only the response "It must be tough to be in pain all the time," and may further output "Press any button," and may change the topic from step S211 depending on the button pressed next. Also, for example, the control unit 4 may output a voice question such as "It must be tough to be in pain all the time. Have you gone to the doctor? If you go regularly, press button 1, and if you don't go to the doctor but go to an osteopathic clinic, press button 2,...."

[0047] The question or answer and the answering action will branch out for each button. Such branching scenarios may be prepared in advance, may be chosen randomly, or may be chosen based on the date and time or the number of times of use. Also, as shown in Figures 4, 5, and 6, it may be possible to jump to a different topic midway.

[0048] In Figs. 4, 5 and 6, after S207, S307 and S407, questions or responses and answers are repeated in the same manner, such as S211 to S215, S311 to S315 and S411 to S415.

[0049] If the user wishes to stop the question or reply and answer, he or she can press the voice mode button 36. When the voice mode button 36 is pressed, the control unit 4 returns to the operation of step S102.

[0050] When the question or response and answer scenario progresses, the control unit 4 refers to each database 21, 22, 23 and gives a general response. The general response is different for each topic of daily life, body, or mental health. For a user who is in good health and has no major problems in daily life, a generalization such as "living alone is wonderful" is registered in the database 21. For a user who is in bad health and has physical problems, a generalization such as introducing a method of self-management or a message of encouragement such as "keep up the good work" is registered in the database 22. For a user who is in bad health and has mental problems, a generalization such as "everyone gets weaker" or "more people feel happier alone" is registered in the database 23.

[0051] In this way, in each QAR process, the control unit 4 outputs a summary message such as encouragement or a message about how wonderful it is to somehow manage to live alone even when you are not feeling well, or a message that makes you feel more relaxed and happy being alone, while the elderly person is engaged in a question-and-answer session with the conversation device 1. In this way, the elderly person can become interested in conversation with the conversation device 1, and can sometimes receive encouragement or comfort from the conversation device 1.

[0052] Note that the summary may not be given suddenly, but may be given only to users who have used the conversation device 1 several times. This can prevent users from becoming bored with the conversation device 1.

[0053] Although the present invention has been described in detail above, the above description is merely illustrative of the present invention in every respect and is not intended to limit the scope of the present invention. It goes without saying that various improvements and modifications can be made without departing from the scope of the present invention. Each of the constituent elements of the invention disclosed in this specification is considered to be an independent invention. Inventions in which each of the constituent elements is combined in any combination method are also included in the present invention. [Industrial Applicability]

[0054] The present invention is a conversation device and has industrial applicability. [Explanation of symbols]

[0055] 1 Conversation device 2 Storage section 21 Daily QAR Database 22 Physical QAR Database 23 Mental QAR Database 24 Condition-related Questions Database 25 Recognizable Speech Database 3 Control section 31 First Answer Button 32 Second Answer Button 33 Third Answer Button 34 4th Answer Button 35 Repeat Playback Button 36 Audio mode button 4. Control Unit 5 Speakers 6. Microphone

Claims

1. A conversation device for conducting a pseudo conversation with a user, a storage unit for storing words and phrases that can be voice-recognized, information on questions, answers, and responses about daily life, information on questions, answers, and responses about the body, information on questions, answers, and responses about mental health, and information on questions about one's condition; a plurality of answer buttons for inputting user answers; A speaker for outputting audio; A microphone for inputting a user's speech; a control unit that recognizes a voice spoken by a user, and outputs from the speaker any one of a question regarding daily life, a question regarding the body, a question regarding mental health, and a question regarding a state of health in accordance with the recognized voice, and receives an answer to the question by pressing one of the plurality of answer buttons, and outputs from the speaker the next question or an answer in accordance with the type of the answer button pressed, The control unit is When outputting the question about the condition, a question is asked to distinguish between good and bad condition, If the patient is in good condition, questions about daily life are output from the speaker. If the person is not feeling well, ask questions to determine whether the problem is physical or mental. If the problem is related to a physical condition, a question about the physical condition is output from the speaker. If the feeling of unwell is related to mental health, a question about mental health is output from the speaker.

2. In addition, it has a repeat playback button. The conversation device according to claim 1 , wherein when the repeat playback button is pressed, the control unit re-outputs the previously output voice.

3. 3. The conversation device according to claim 1, further comprising a voice mode button which, when pressed during the progress of a conversation, causes the operation of the control unit to return to the initial voice recognition operation.

4. The conversation device according to any one of claims 1 to 3, wherein the control unit recognizes the voice spoken by the user and compares it with words and phrases registered in the memory unit to determine whether the recognized voice is related to daily life, the body, or mental health, or whether it is necessary to ask about the user's condition, and outputs from the speaker one of a question about daily life, a question about the body, a question about mental health, and a question about the user's condition.

Images