JP2002215183A - Method and apparatus for creating sensibility, and software - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for creating sensibility by which sensibility in a state similar to a human being is outputted, and provide an apparatus for creating sensibility, and a recording medium. SOLUTION: Personality information subjected to sensibility developing which determines at least reason and the will is previously held. Status information which indicates the emotion of an opponent or an environmental status is inputted. Instinctive motivation information which includes at least a first instinct parameter indicating a degree of comfort, a second instinct parameter indicating a degree of risk, and a third instinct parameter indicating a degree of achievement or a degree of change is created based on the status information. Emotional information including at least a basic emotional parameter of pleasure, anger, and grief is created based on the created instinctive motivation information. The created emotional information is controlled based on the personality information.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a kansei generating method, a kansei generating device, and a recording medium, and can be used in various systems for various uses as kansei control of a virtual human or robot.

[0002]

2. Description of the Related Art The related art related to the present invention includes:
For example, an “emotional generator and an emotion generating method” disclosed in Japanese Patent Application Laid-Open No. H11-265239 is known. The emotion representing the internal state of a human or the like changes variously depending on the situation at that time. Japanese Patent Laying-Open No. 11-265239 discloses a technique for realizing emotion generation in an unpredictable situation.

That is, a situation is evaluated in the light of a predictable situation, and an emotion of the device itself is generated. In addition, it analyzes past emotions that have actually occurred and situations at that time, and learns unpredictable incidental conditions specific to each situation and emotions corresponding thereto. If the newly input situation satisfies the learned incidental condition, the emotion corresponding to the incidental condition is output.

[0004] The state of emotion generated by such a device is reflected on, for example, output audio and video.

[0005]

Conventional devices simply generate emotions directly from input situational information. In a real human being, various parameters such as instinct, reason, and personality affect each other in a complex manner, resulting in behavior,
Remarks, facial expressions, etc. change. The conventional device cannot accurately reflect the instinct, reason, individuality, and the like of the human in the result.

[0006] Instinct and emotion can be considered as emotions. Instinct also motivates basic biological emotions and their emotional development. Also, it is considered that humans do not output emotions as they are, but output sensibility controlled by reason and personality. An object of the present invention is to provide a kansei generating method, a kansei generating device, and a recording medium that can output kansei in a state closer to a human.

[0007]

According to a first aspect of the present invention, there is provided a method for generating a sensibility, wherein at least personality information for determining at least a reason, a predetermined characteristic, and a will of the sensibility generating object is stored in advance, and the emotion or the state of the environment of the partner is generated. The first instinct parameter indicating at least the degree of pleasure, the second instinct parameter indicating the degree of risk, and the third indicating the degree of achievement and the degree of change based on the situation information. Instinct motivational information including instinct motivational information, and based on the generated instinct motivational information,
An emotion information including basic emotion parameters of anger and sadness is generated, and the generated emotion information is controlled based on the personality information.

In the present invention, instinctual motivation information for generating an emotion is generated based on input situation information (e.g., emotion, intention, surrounding situation, etc. of a partner). That is, instinctual motivation information is generated from the situation information, and emotion information is generated based on the instinctual motivation information. The generated emotion information is controlled according to the personality information. For this reason, it is possible to output emotion controlled by the individual's reason and intention, that is, information on sensibility.

Further, since emotion information is generated through instinctual motivational information, the generated emotion can be controlled more precisely and easily. For example, the emotions that occur when a human encounters a dangerous state while already recognizing the danger, and the emotions that occur when a person who does not feel danger suddenly encounters a dangerous state Is different,
Such emotional differences can also be reproduced.

[0010] Instinct parameters other than the above, it is desirable to further add attention (rejection), confidence (confused), follow-up (declaration), and the like. In addition, it is desirable to further add to the basic emotion parameters constituting emotion information, in addition to the above, surprise, fear, suffering, disgust, disdain, approach, escape, jealousy, envy, subordination, frustration, anxiety, and the like. The emotion generating device according to claim 2 inputs episode situation information indicating a state of a partner's emotion, environment, and will, and based on the episode situation information, at least a first instinct parameter indicating a degree of pleasure and a danger. Instinct determining means for generating instinct motivation information including a second instinct parameter indicating a degree and a third instinct parameter indicating a degree of achievement or a change, and an instinct motivation output from the instinct determining means. Emotion generating means for generating emotion information including at least basic emotion parameters of joy, anger, and sadness based on attachment information, and personality information providing means for providing personality information for determining at least the reason and the emotional intention of the sensibility generating target And an emotion control means for controlling emotion information output from the emotion generating means based on the personality information provided from the personality information providing means. Characterized in that a and.

According to the second aspect, the instinctive determining means, the emotion generating means, the personality information providing means, and the emotion controlling means are provided so that the method for generating a feeling according to the first aspect can be implemented. Therefore, it is possible to output the emotion controlled by the individual's reason and will, that is, information on the sensitivity. Further, since emotion information is generated through instinctual motivation information, the generated emotion can be controlled more precisely and easily.

According to a third aspect of the present invention, in the emotion generating apparatus according to the second aspect, the emotion generating means includes a life rhythm generating means for generating information representing a periodically changing environment or a living rhythm of a living body; And a spontaneous emotion control means for controlling a spontaneous emotion in the emotion generating means according to a probability according to the information of the life rhythm outputted by the means.

For example, natural environmental conditions such as temperature and humidity are irregular, but change periodically with changes in weather, season, time and the like. In addition, it is considered that each human has a physical rhythm, an emotional rhythm, an intelligence rhythm, and the like individually. Such a periodically changing rhythm is considered to have various effects on actual human emotions.

According to a third aspect of the present invention, the spontaneous emotion control means controls spontaneous emotion in the emotion generating means in accordance with a probability corresponding to the information of the living rhythm outputted by the living rhythm generating means. Therefore, it is possible to change the output emotion in accordance with the environment or the living rhythm of the living body. In a fourth aspect, the instinct-emotion correspondence information holding unit holds pattern information for associating the basic emotion parameter with the instinctual motivation information in the emotion generating unit. A matching probability learning means for outputting information indicating a probability of matching / mismatch between the instinct motivational information output from the determining means and the pattern information of the instinct-emotion correspondence information holding means; .

According to the fourth aspect, the probability of matching between the instinctual motivation information and the pattern information can be obtained from the matching probability learning means and used as an emotion judgment element. For example, when the human mental state changes from the first state to the other second state, the state changes via the third state on the way, so that the third state includes certain pattern information. May be temporarily inconsistent. However, in this case the third
The pattern information matched in the state is not very useful. By using the matching probability obtained by the matching probability learning means, it is possible to suppress the generation of the emotion of the pattern information with low probability.

According to a fifth aspect of the present invention, in the sensibility generating device according to the second aspect, at least emotion feedback generated by inputting at least the last generated own emotion information to the emotion generating means is reflected on the next generated self emotion information. A control means is provided. Human emotions can be considered to change in a chain according to various motive inputs. For example, the degree of emotional anger that occurs when a person in a normal state is motivated to be angry, and the degree of emotional anger that occurs when a person who is already angry is further motivated to be angry. Is very different.

According to the fifth aspect of the present invention, by providing the emotion feedback control means, the state of the emotion which has occurred immediately before can be returned to the input and reflected on the next emotion. Therefore, emotions closer to humans can be generated. According to a sixth aspect of the present invention, in the emotion generating apparatus according to the second aspect, the emotion control means reflects information on a life rhythm, which is a personality of the emotion generation target, in the input emotion information.

According to the sixth aspect, the information of the life rhythm can be reflected on the sensitivity. For example, there is a difference in the result of judgment based on reason or the like between a case where a person is reluctant and a case where a person is not reluctant. Such a difference in sensitivity can be reproduced by reflecting the life rhythm. According to a seventh aspect of the present invention, in the kansei generating device according to the second aspect, a knowledge database for storing situation information indicating past situations, episodes, and the results thereof, and the past situation information similar to newly input situation information is stored in the knowledge database. Knowledge matching means for searching and extracting from a database, and providing past situation information to the emotion control means,
While updating the content of the knowledge database with the newly input status and status information indicating the result,
Data update control means for automatically deleting status information stored in the knowledge database, which has a lower priority, automatically in chronological order according to the weight of the content is further provided. .

In the present invention, situation information indicating past situations and their results is stored in the knowledge database. For example, information indicating the status of a certain episode and whether or not the final result was successful is accumulated. For this reason, past situation information similar to the current situation can be acquired from the knowledge database and used for emotion control.

By the way, information newly generated with the passage of time must be sequentially added to the knowledge database. However, the storage capacity of a device constituting the knowledge database is limited. Also, the processing speed decreases as the amount of stored information increases. However, according to the seventh aspect, under the control of the data update control means, the status information having the lower priority is automatically deleted from the knowledge database in chronological order. Therefore, a result similar to that of human forgetting can be realized, and shortage of storage capacity and reduction of processing speed can be prevented.

According to an eighth aspect of the present invention, in the sensibility generator according to the second aspect, voice input means for inputting a voice signal, intensity detection means for detecting voice strength from the voice signal input by the voice input means, and Tempo detection means for detecting the appearance speed of voice as a tempo from the voice signal input by the input means, and intonation detection means for detecting intonation representing an intensity change pattern in a word of voice from the voice signal input by the voice input means; A change amount detecting means for obtaining a change amount for each of the sound intensity detected by the strength detecting means, the tempo of the sound detected by the tempo detecting means, and the intonation of the sound detected by the intonation detecting means; Emotion detecting means for outputting a signal representing at least anger, sadness, and joy based on the amount of change detected by the means. Characterized in that was.

According to the eighth aspect, the emotional state of the other party can be detected based on the feature amount extracted from the voice. Therefore, it is possible to generate a self-emotion according to the emotion of the other party. According to a ninth aspect of the present invention, in the sensibility generating apparatus according to the eighth aspect, a voice recognition unit that recognizes voice input from the voice input unit and outputs character information, and outputs the voice information recognized by the voice recognition unit in a natural language. Natural language processing means for generating semantic information representing the meaning of the input speech after processing is further provided.

According to the ninth aspect, since the semantic information is obtained with respect to the words spoken by the other party, the result of understanding the semantic information can be reflected on one's own sensitivity. Claim 1
The recording medium of No. 0 is software including a computer-executable program and data used for control of emotion generation, and the program inputs status information representing the emotion or environment state of the other party, Based on the information, an instinct motivation including at least a first instinct parameter indicating a degree of pleasure, a second instinct parameter indicating a degree of risk, and a third instinct parameter indicating a degree of achievement or a change. Based on the procedure for generating information and the generated instinctual motivational information,
A procedure for generating emotion information including basic emotion parameters of anger and sadness, a procedure for providing personality information that determines at least the reason and will of the sensibility generation target, and generating the emotion information based on the personality information. And a control procedure.

By inputting the software of claim 10 and executing the program on a predetermined computer, the method of generating sensibility of claim 1 can be implemented.

[0025]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of a feeling generating method, a feeling generating device and a recording medium according to the present invention will be described with reference to FIGS. This form corresponds to all claims.

FIG. 1 is a block diagram showing an example of the configuration of a system using the sensibility generating device. FIG. 2 is a block diagram showing a configuration of the instinct information generation unit. FIG. 3 is a block diagram illustrating a configuration of the emotion information generation unit. FIG. 4 is a schematic diagram showing an example of a reaction pattern model in the emotion reaction pattern DB. FIG. 5 is a block diagram showing the configuration of the emotional thought recognition unit. FIG. 6 is a block diagram showing a configuration of the emotion detection device. FIG. 7 is a block diagram showing a configuration of the intonation detecting unit.
FIG. 8 is a graph showing the relationship between the change in the emotional state and the sound intensity, tempo, and intonation.

FIG. 9 is a time chart showing a process of audio signal processing in the intonation detecting section. FIG. 10 is a flowchart showing the operation of the forgetting processing unit. FIG. 11 is a schematic diagram illustrating a configuration example of information stored in the emotional emotion storage DB. In this embodiment, the instinct determining means, the emotion generating means, the personality information providing means and the emotion controlling means of the present invention are the instinct information generating section 110, the emotion information generating section 130, and the personal D
B (abbreviation of database, the same applies hereinafter) 163 and kansei thinking recognition unit 150. The life rhythm generating means and the spontaneous emotion control means in claim 3 correspond to the multivariate analysis section 136 and the spontaneous emotion control section 137, respectively.

The instinct-emotion correspondence information holding means and the matching probability learning means in claim 4 correspond to the emotion reaction pattern DB 141 and the learning processing unit 135, respectively. The emotion feedback control means according to claim 5, wherein the basic emotion parameter generation unit 13
3 and the temporary storage DB 142. The knowledge database, the knowledge matching unit, and the data update control unit in claim 7 correspond to the knowledge DB 161, the matching processing unit 152, and the update processing unit 156, respectively.

The voice input means, the intensity detection means,
The tempo detecting means, intonation detecting means, change amount detecting means, and emotion detecting means are provided by the microphone 11, the intensity detecting unit 17,
Tempo detector 18, intonation detector 19, emotion change detector 2
2 and the voice emotion detection unit 23. Claim 9
The voice recognition unit and the natural language processing unit correspond to the voice recognition unit 20 and the sentence recognition unit 26, respectively.

The system shown in FIG. 1 is configured on the assumption that a natural emotional dialogue between any human and a computer (virtual human) is realized. In this example, an emotion detection device 200 is provided to detect the emotion of a human being a computer, and the emotion generation device 1 is used to reflect the personality and sensitivity of the computer itself in a dialog.
00 is provided.

An environment information input device 300 is provided for inputting various environment information. The environment information input device 300 outputs, for example, date, time, weather information, location, video information, and the like. The sensibility generating device 100 can be used in a system that operates autonomously. For example, information of a scenario created in advance is sent to
, A response corresponding to the scenario can be obtained from the output of the sensibility generating device 100. In that case, the emotion detection device 200 is unnecessary.

In the example shown in FIG. 1, the devices necessary for realizing the dialogue are connected to the output of the kansei generating device 100, but the kansei data output from the kansei generating device 100 may be used for various purposes. Can be. For example, when the sensibility generating device 100 is used for data communication, it is only necessary to output character information, and there is no need to output sound. In addition, it is also possible to reflect the emotion data output from the emotion generating apparatus 100 on video and music, information retrieval, and machine control.

Next, the configuration and operation of the sensibility generating device 100 will be described. The emotion detecting device 200 will be described later. The system shown in FIG. 1 can be actually configured by a computer system and a software program executed on the computer system, or can be realized as dedicated hardware. The software program and the data to be used can be stored in an arbitrary recording medium, and can be read from the recording medium into a computer and executed. In the following description, the system itself in FIG. 1 is referred to as a computer.

The input of the sensibility generator 100 is roughly divided into two types of data D1 and D2. Data D
1 is information indicating the emotion of the other party. The data D2 is character information subjected to natural language processing, and includes information on the will, situation, and environment of the other party. By natural language processing, data D
2 is input as information indicating “5W3H”, that is, “who”, “what”, “when”, “where”, “why”, “how”, “how much”, and “how much”.

In practice, the following various information can be used as inputs to the sensibility generating device 100.
(A) Changes in vocal strength, rhythm, tempo, pause, scale, pitch, melody, sound harmony, frequency, etc. related to temporal properties and basic emotions (anger, joy, sadness, disgust, startle, fear, etc.)
Degree.

(B) Accent, thickness, fineness, brightness, roughness, timbre (JIS-Z8) of utterance related to tonal properties
109), formants, intonation, prominence (prominence that emphasizes certain parts of spoken language and clarifies meaning). (C) Words related to the nature of stress, phrase content, stress distribution in the sentence, hyperphrasal feature information, feature information generated by artificial intelligence, and the like.

(D) Discourse-analyzed text information, episode information (including semantic information, information recognized by artificial intelligence), and the like. Of these pieces of information, (A) and (B) are affected by the intention and emotion of the speaker. Such an emotion can be detected using the emotion detection device 200. Figure 1
As shown in FIG. 2, the instinctive information generating unit 1
10, prosody pattern DB 121, instinct language meaning dictionary 122, emotion information generation unit 130, emotion reaction pattern DB
141, temporary memory DB 142, emotional thinking recognition unit 150,
It has a knowledge DB 161, a sensitivity DB 162, a personal DB 163, and a moral hazard DB 164.

The functions of the sensibility generating device 100 can be basically divided into three functional elements: an instinct information generating unit 110, an emotion information generating unit 130, and a kansei thinking recognition unit 150. First, the instinct information generation unit 110 will be described. As shown in FIG. 2, the instinct information generation unit 110 includes a prosody pattern synchronization degree recognition unit 111 and an instinct parameter generation unit 1.
12 and a dictionary search unit 113.

The prosody pattern DB 121 referred to by the prosody pattern synchronization degree recognition unit 111 stores a dictionary of prosody patterns input to the computer (virtual person) in advance. Prosody is the rhythmic element of speech, syllable, word, phrase, sentence, whole speech (continuous speech longer than word)
Represents the phonetic and phonological properties that appear for. That is, the pattern information of the computer itself equivalent to the input information of (A) and (B) is used as the personality information as the prosody pattern DB12.
It is held at 1.

The prosody pattern synchronization degree recognition section 111 compares the emotion analysis data D1 of the other party input from the emotion detection device 200 or the like with the prosody pattern stored in the prosody pattern DB 121 to recognize the degree of synchronization and the degree of synchronization between the two. I do. In the output of the prosody pattern synchronization degree recognition unit 111, information indicating the presence or absence of strong vocabulary, a change in emotional rhythm, and the like appear. On the other hand, information about instinct stimuli is registered in the instinct language meaning dictionary 122 in advance. Specifically, words related to the nature of stress stress, stress distribution patterns in the sentence, and various pieces of information indicating hyper-phrasal features are held as a dictionary in association with instinct stimuli.

The dictionary search unit 113 compares data D2 (will and situation of the other party) input as character information with the contents of the instinct language meaning dictionary 122, and generates instinctual reaction information from the contents of the conversation. I do. The instinct parameter generation unit 112 generates instinct motivational information D4 based on information input from the prosody pattern synchronization degree recognition unit 111, information input from the dictionary search unit 113, and data D3. The data D3 is information fed back from the output of the sensibility generating device 100, and is an episode or a desired reaction pattern proposed by the computer.

In this example, the instinct motivational information D4 includes the degree of certainty (or degree of embarrassment), degree of pleasure (or degree of discomfort), degree of danger (or degree of security), degree of attention (or degree of rejection), degree of achievement ( Or the degree of change) and the degree of follow-up (or degree of assertion). The instinct parameter generation unit 112
The value of each instinct parameter is determined as follows.

Pleasure (discomfort): The degree of pleasure is added when approaching the content suggested by the computer or the desired situation episode, and subtracted when the opposite is true. In addition, the degree of pleasure is added when approaching a prosody that is determined to be comfortable in advance, and is subtracted when the opposite is true. Danger level (security level): The risk level is added when the computer approaches a content that the computer has previously considered to be dangerous or a situation episode that is assumed to be dangerous, and subtracts when the situation is the opposite. In addition, the risk is added when approaching a prosody that is regarded as dangerous in advance, and is subtracted when the prosody is regarded as dangerous.

Achievement (degree of change): Achievement is added when the content approaches a content that is determined to be successful or achieved in advance or a situation episode assumed to be successful or accomplished, and is subtracted when the opposite is true. In addition, the degree of change is added when approaching a specific prosody regarded as rapid modulation, and is subtracted when the opposite is true. Attention (rejection level): Content that was previously rejected / denied or rejected
When approaching a situation episode that is assumed to be negative, the rejection level is added (subtraction level is subtracted), and vice versa. In addition, when a strong claim or a repeated claim is detected, or when approaching a strong prosody, the attention level is added. In addition, a rejection degree is added when approaching the prosody defined as unpleasant.

Follow-up degree (degree of assertion): When approaching the content determined in advance as self-deprecation or self-denial or an assumed situation episode, the degree of follow-up is added (the degree of assertion is subtracted), and the content determined in advance as a specialty appears. In this case, the degree of assertion is added (the degree of following is subtracted). If a prosody determined as anxiety appears in advance, the claim level is added. In addition, when approaching a strong prosody, the degree of repulsion or the degree of self-denial may be added.

Confidence (degree of embarrassment): When approaching the episode of the content of embarrassment or an assumed situation, if the recognition rate of various stimuli (inputs) related to instinct is low (for example, 70% or less),
The degree of embarrassment occurs in inverse proportion to the recognition rate. The recognition rate is determined based on the tone of the voice and the content of the conversation.

In order to realize the above control, the contents desired by the computer and the prosody of the situation episode are determined in advance as individuality. As described above, the emotional information of the partner stimulates the personal instinct of the computer, and the value of each instinct parameter changes. The instinct motivational information D4 output from the instinct information generation unit 110 is input to the emotion information generation unit 130. Next, the emotion information generation unit 130 will be described.

As shown in FIG. 3, the emotion information generation section 130 includes a reaction pattern search section 134, a learning processing section 135, a multivariate analysis section 136, a spontaneous emotion control section 137, and a basic emotion parameter generation section 133. In addition, the reaction pattern search unit 134, the learning processing unit 135, and the emotion reaction pattern DB 141 constitute a response system 131,
The multivariate analysis unit 136 and the spontaneous emotion control unit 137 constitute the operant system 132.

[0049] The response system 131 is provided to generate an emotion by stimulation induction. The operant system 132 is provided for generating spontaneous emotions (libido). In the emotion reaction pattern DB 141 used in the response system 131, information of a reaction pattern model representing the correspondence between the instinct motivational information D4 and the basic emotion parameters is held in advance. This reaction pattern model can be represented, for example, as shown in FIG.

When one computer selectively reproduces the personalities of a plurality of persons, a reaction pattern model corresponding to each of a plurality of persons or each type of personality is registered in the emotion reaction pattern DB 141 in advance. The reaction pattern model may be selected according to the personality of the selected person. In this example, instinct motivational information D
The instinct parameters input as 4 include the above-mentioned confidence (or embarrassment), pleasure (or discomfort), danger (or safety), attention (or rejection), achievement (or change) and Six degrees of follow-up (or degree of assertion) are assumed.

The following 15 types of parameters are assumed as basic emotion parameters output from emotion information generating section 130. However, parentheses indicate the instinct parameters that are affected. 1. Anger (discomfort) 2. Joy and merry (pleasant) 3. Sorrow (not achieved / stagnation / discomfort) Surprise (achievement / shock) Fear (danger, tension) 6. suffering (danger, tension, discomfort) Disgust (rejection / rejection / discomfort) 8. Disdain (rejection / relaxation) Approach (pleasant / safe) 10. Escape / Avoid (Danger / Tension / Discomfort) Jealousy (discomfort, anger, envy, attention) Active (safe, pleasant, convinced) Dependency (achievement / follow) 14. Irritability / Struggle (Claim / Stagnation / Discomfort / Danger) Anxiety (danger, tension, embarrassment, discomfort) The emotion response pattern DB 141 holds a reaction pattern indicating the relationship with one or more basic emotion parameters for each of the 15 types of basic emotion parameters.

The reaction pattern search unit 134 searches the emotion reaction pattern DB 141, checks whether the input instinct motivational information D4 matches the reaction pattern of each basic emotion parameter, and checks the matched basic emotion parameter. Is output as data D6. Learning processing unit 13
5 is information D3 output from the emotional thought recognition unit 150
Then, the probability of pattern matching is learned based on the next reaction emotion of the partner output from the reaction pattern search unit 134, and the contents of the emotion reaction pattern DB 141 are changed according to the learning result.

On the other hand, environmental information (D2) including, for example, weather information, seasonal information, and time information is input to the input of the operant system 132. The multivariate analysis unit 136
Performs multivariate analysis on various input environmental information,
As a result, life rhythm information is output. The life rhythm information includes a short-term rhythm (for example, one-hour cycle), a life rhythm (for example, 24-hour cycle), and a long-term emotion rhythm (for example, 2
Regular (sinusoidal) rhythm with a constant cycle such as an 8-day cycle, a physical long-term rhythm (for example, a 23-day cycle), an intelligent rhythm (for example, a 33-day cycle), and irregularities such as temperature, humidity, and weather There is a rhythm.

The spontaneous emotion control unit 137 includes the multivariate analysis unit 1
The spontaneous emotion (libido) is output according to the probability within a predetermined range in the life rhythm information rhythm output from 36. The basic emotion parameter generation unit 133 comprehensively determines the information on the basic emotion parameter and the matching rate output from the response system 131 and the spontaneous emotion output from the operant system 132 (15 types of basic emotion parameters). Is output as self-emotion information D5.

The output self emotion information D5 is temporarily stored in the temporary storage DB 142, and is fed back to the input of the basic emotion parameter generation unit 133. The basic emotion parameter generation unit 133 inputs the information fed back from the temporary storage DB 142 as the immediately preceding self-emotion and reflects it in the next emotion determination result. Further, when the basic emotion parameter generation unit 133 makes a comprehensive judgment, the priority and the degree of influence of each unit are determined according to the personality defined as the personality information 143.

For example, when the personality of the intuition type is reproduced, the influence of the response system 131 is increased (8
0% or more) and the influence of the previous self-emotionality is also increased. In the case where the thinking-type personality is reproduced, the influence of the response system 131 is reduced in an environment where the output of the operant system 132 is stable (3
0% or less) and reduce the influence of the previous self-emotional feeling.

The self-emotional information D5 output from the emotion information generating section 130 is input to the sensible thinking recognition section 150. As shown in FIG. 5, the emotion information generating unit 130 includes a weighting processing unit 151, a matching processing unit 152, a multivariate analysis unit 153,
A total intuition determination unit 154 and an update processing unit 156 are provided. The weighting processing unit 151 weights the input self-emotional information D5 according to the personality information 155. Weighted self-emotion information is weighted by processing unit 1
51.

On the other hand, to the input of the collation processing unit 152, character information (5W3H) including an episode indicating an environment, a situation, a partner's will, and the like and data of the result are input as data D2. The knowledge DB 161 referenced by the matching processing unit 152 includes
Past episodes, their results, and semantic information indicating their meaning are accumulated as knowledge in the form of character information (5W3H). The knowledge on the knowledge DB 161 includes information on the time when each data is obtained, and is arranged according to the order of the time.

In this example, the knowledge on the knowledge DB 161 is
It can be classified into long-term memory, declarative memory, and procedural memory. Declarative memory is verbal memory, which represents episodic information that is an event in a specific temporal / spatial context and semantic information as general knowledge. The procedural memory represents a memory related to a method or technique. Episode information includes time, place, content, will (approve, disagree, favor, etc.), personal name, quantity, weight, situation, state, partner's personal information, emotion, intention (purpose), attitude, interpersonal relationship, etc. . The semantic information corresponds to a language dictionary and a sensitivity dictionary. Personal information includes personal temperament, personality, emotion, social adaptability (sociability), desire, conflict, attitude, superiority, complex, interest, adequacy, morality, thinking pattern, emotional singularity, attachment content (and Degree), taboo, preference, good / bad criteria, etc.

In this example, knowledge information is stored in the knowledge DB 161 according to the following grammar. However, the configuration of the database is changed according to the purpose. Story = Scene + Plot + Resolution Scene = Character + Location + Time Theme = (Event) + Goal Plot = Episode Episode = Lower Goal + Trial + Result Trial = Event + Episode Result = Event + State Resolution = Event + State Lower Goal , Goal = desired state Character, place, time = state Further, new information is sequentially added to the knowledge DB 161 by the operation of the update processing unit 156. Unnecessary information is automatically deleted from the knowledge by repeated forgetting processing. That is, the data is deleted in order from the data that is older in time except for the data with the higher priority. For example, data that has been repeatedly used and determined to have a strong impression is given priority, and is not deleted even when it becomes old. Also, the degree of forgetting and the priority of each data can be changed according to individuality.

The collation processor 152 receives the input data D
2 and retrieves and extracts past episodes and results from the knowledge DB 161 that are close to the previous episode and collates the input data with the extracted knowledge. The learning processing unit 157 generates, based on the result of the input episode, self-value information for the episode by learning. That is, the learning processing unit 157 assigns a degree of satisfaction and a degree of pleasure or discomfort based on the result of the input episode.

The multivariate analysis unit 153 includes the weight processing unit 1
51, the episode information and the result information input from the matching processing unit 152, the self-value information input from the learning processing unit 157, and the self-will input from the personal DB 163. And the information of the self-instinct is multivariately analyzed, and the result is output to the overall intuition determination unit 154.

The total intuition determination unit 154 stores the personal DB1
Using the contents of the moral hazard DB 164 and the moral hazard DB 164 as a determination dictionary, the information input from the multivariate analysis unit 153 is comprehensively determined, what is spontaneously executed, and the result is output as data D3. The personal DB 163 stores various types of information as dictionary information as described below.

1. Individuality information: (a) Judgment criteria according to the degree of individuality type: stereotype, others-oriented type, internal-oriented type,
Traditional, attack-oriented, collaborative, stress-overcoming, and stress-dissipating types are possible. Further, the achievement motivation degree and the reactance degree can also be used as criteria.

(B) Judgment criteria for cognitive style: A cognitive style based on a distinction between “contemplation type” and “impulsive type” and a distinction between “field-dependent type” and “field-independent type” is used as a judgment standard. (C) Judgment criteria based on personality: In the case of Japanese, individual temperament, personality, emotion, social adaptability (sociability), classified by personality test method or TPI (Tokyo University version comprehensive personality test), etc. Desire, Conflict, Attitude, Complex, Interest, Appropriateness, Morality, Thinking Pattern, Emotion Singularity, Attachment Content (and Degree), Taboo, Preference, Good and Evil Criteria, Shame Criteria, Crime Criteria, Pleasure / Discomfort Is used as a criterion.

(D) Judgment Criteria for Negativeness Bias: Negative information is biased and used for character formation in order to make it easy to catch large amounts of negative information. (E) Criteria for sticking persistence time: Determine the degree of sticking to cognitive information, episodes, and emotion information of the other party and the response response time to it.

2. Id unconscious reaction criterion information: (a) Word dictionary and phrase dictionary of contents that stimulate instinct. (B) Criteria for various instinct reaction times for the degree of patience, stickiness, and intuition of each gender. (C) Self-instinct pattern corresponding to the emotion of the partner determined as personality. 3. Homeostasis Criteria Information: Criteria for trying to maintain the overall instinct output in harmony and stability.

4. Ego Consciousness Reaction Criteria Information: Judgment criteria information that expresses one's will by personality. In addition, as a judgment dictionary, information used for recognizing judgment and identification judgment such as true / false, right / wrong, adequacy, information used for instinct judgment such as pleasure and unpleasantness, and objects such as complexity and weight Information used for individual recognition judgment of information, information used for judgment of relative cognition between objects such as same, large, small, different, and similar, and meta-memory judgment such as certainty of memory and accuracy of knowledge Information and
It includes information used for abstract decisions such as true, good, and love, and information used for inductive decisions.

The moral hazard DB 164 holds dictionary information on occupational morals, personal morals, fundamental morals, and the like. For example, as an occupational moral, "a complete calculation is required as an architect", "work takes precedence over anything", and "we have pride as a professional" are registered. In addition, as personal morals, "I value a woman (I do not like a man)", "I am proud of the countryside", "I am proud of the Japanese" are registered. As a fundamental moral, "do not kill", "value parents", "I am a man (woman)" and the like are registered.

The total intuition determination unit 154 includes a weighting unit 151, a matching unit 152, and a multivariate analyzing unit 15.
3 and a moral hazard DB1 and a judgment dictionary on the personal DB 163 representing the personality and will of the computer.
The self-emotion response (sensitivity) is determined based on the suppression dictionary based on the judgment dictionary described above. When making a decision, reflect the environment and situation at that time and the will of the partner.

The sensibility thinking recognition unit 150 has the following functions. 1. When a strong impression or vocabulary or a sudden change in emotion is detected, the cycle of the judgment is changed according to the personality. For example,
When suddenly loud and strong content is claimed, the judgment cycle is shortened. 2. In response to his / her own biorhythm due to personality, different emotional judgments are made when he / she feels tired and when he / she does not feel good.

3. Different emotional judgments are made according to their own pleasure / discomfort and emotional amount. 4. A rational value judgment is performed on the information indicating the current situation using the knowledge in the knowledge DB 161, and a final will is determined by reflecting the influence of the emotion judgment result. 5. When making value judgments, social value, occupational value,
Judge from each standpoint, such as living value and personal value. In addition, each of social value, occupational value, living value and personal value is distinguished and judged. For example, regarding social value, the value is calculated for each viewpoint of religion, aesthetics, society, politics, economy, and ethics.

6. As a material for decision making, we judge the value of each of satisfaction, dissatisfaction, gain and loss, interest, safety and danger. When making a value determination regarding safety, for example, the following determination is made. (A) Generate value for hostility and defensive reactions if a third party is trying to add "discomfort" to themselves.

(B) Generate value for hostile emotions and offensive reactions if you are trying to add "discomfort" to a third party. (C) generate value for favor and cooperative attack response when trying to ally with a third party when something else is trying to add "discomfort" to the third party.

7. The generated value information is the sensitivity DB 162
And is used as a material for subsequent judgment. In addition, since the Kansei thinking recognition unit 150 has various learning functions similar to humans, the contents of the personal DB 163 and the Kansei DB 162 are sequentially updated by accumulating experience.

The emotional thinking / recognition unit 150 outputs a result based on a comprehensive judgment based on numerical values such as various values, so that logical reasoning and judgment such as artificial intelligence are not performed. That is, the data D3 output from the emotional thought recognition unit 150 is emotional information obtained by intuitive judgment of the computer itself. Next, the sensitivity generating device 1 of FIG.
A specific example of the emotion detection device 200 connected to the input of 00 will be described.

As shown in FIG. 6, the emotion detection device includes a microphone 11, an A / D converter 12, a signal processing unit 13, a voice recognition unit 20, an intensity detection unit 17, a tempo detection unit 18, and an intonation detection unit 19. Temporary storage unit 21, emotion change detection unit 22, voice emotion detection unit 23, emotion pattern DB 24, keyboard 25, sentence recognition unit 26, television camera 31, image recognition unit 32, face pattern DB 33, face emotion detection unit 34, characters Recognition unit 39, emotion and emotion storage DB41, forgetting processing unit 42, synchronization processing unit 43, humanity information DB44, personal information DB4
5, a specialized information DB 46 and an emotion recognition unit 60 are provided.

The speech recognition unit 20 has a signal processing unit 1
3. Phoneme detector 14, word detector 15, and sentence detector 1
6 are provided. The speech recognition unit 20 also includes a function of a commercially available speech recognition (natural language) device. In FIG. 6, a voice recognition unit 20, an intensity detection unit 17, a tempo detection unit 18, a intonation detection unit 19, a temporary storage unit 21, an emotion change detection unit 22, and a voice emotion detection unit 23 are used to detect emotion from voice. Circuit.

This emotion detection device is provided as input means for reading information of a person to be detected as an emotion.
A microphone 11, a keyboard 25, and a television camera 31 are provided. That is, the voice input from the microphone 11,
The emotion of the other person is detected using character information input from the keyboard 25 and information such as facial expressions input from the television camera 31.

Note that it is actually possible to detect an emotion based only on the voice input from the microphone 11,
It is also possible to detect the emotion based on only the character information input from the keyboard 25, and it is also possible to detect the emotion of the other person based only on the facial expression input from the television camera 31. However, comprehensively judging information obtained from a plurality of information sources is more effective in increasing the accuracy of emotion detection.

First, processing related to audio will be described. The audio signal input from the microphone 11 is sampled by the A / D converter 12 and converted into a digital signal. The voice digital signal obtained at the output of the A / D converter 12 is input to the voice recognition unit 20. Signal processing unit 1
Reference numeral 3 extracts a frequency component required for voice intensity detection.
The intensity detector 17 detects the intensity of the signal extracted by the signal processor 13. For example, the result of averaging the amplitude of the audio signal can be used as the intensity.

The averaging period for detecting the sound intensity is set to, for example, about 10 seconds. However, if a break for each sentence is detected even within 10 seconds, averaging is performed from the beginning of the sentence to the time when the break is detected. That is, the strength of each voice sentence is detected. The phoneme detecting unit 14 provided in the speech recognizing unit 20 detects a segment for each phoneme of the input speech. For example, if a sentence “Today is fine weather” is input by voice, “Kyo / u / ha / i / i / te / n / ki / de / su / su /
”Is detected for each phoneme.

The word detecting section 15 provided in the voice recognizing section 20 detects a break of each word of the input voice. For example, when a sentence “Today is fine weather” is input by voice, a break for each word is detected, such as “Today / Ha / Good / Tenki / Ya”. Further, the sentence detection unit 16 provided in the speech recognition unit 20 detects a break of each sentence of the input speech. When a silent state longer than a specific length is detected, it is considered that a break for each sentence has appeared. A value on the order of (0.1 to 2) seconds is assigned to the threshold value of the length of the silent state. The threshold is not constant, but is automatically changed to reflect the state of the emotion detected immediately before.

The tempo detecting section 18 receives a signal for each phoneme output from the phoneme detecting section 14 and detects the number of phonemes appearing in a unit time as a tempo. For the tempo detection cycle, for example, a time of about 10 seconds is assigned. However, when a sentence break is detected, 1
Even if it is within 0 seconds, the counting of the number of phonemes is stopped up to that point and the value of the tempo is calculated. That is, the tempo is detected for each sentence.

The intonation detecting section 19 receives a speech signal which is divided for each word for which the word detecting section 15 has detected a break. The intonation detecting unit 19 detects the intonation representing the voice intensity change pattern in each word and in each sentence segment in the sentence detecting unit 16 from the input speech signal. As a result, the intonation detecting section 19 detects a characteristic intensity pattern in the break.

As shown in FIG. 7, a band-pass filter 51, an absolute value converter 52, a comparator 53, an area center detector 54 and an area interval detector 55 are provided inside the intonation detector 19.
Is provided. FIG. 9 shows an example of the waveforms of the signals SG1, SG2, SG3, and SG4 of each unit in the intonation detecting unit 19.
Is shown in Note that the vertical axis of each signal in FIG. 9 represents amplitude or intensity. In the example of FIG. 9, the length of one word extracted from the voice is about 1.2 seconds.

The band-pass filter 51 extracts only frequency components necessary for detecting inflection from the input signal SG1. In this example, only the frequency components within the range of 800 Hz to 1200 Hz appear at the output of the band-pass filter 51 as the signal SG2. Referring to FIG. 9, it can be seen that a pattern of intensity change due to intonation within a word appears in the signal SG2.

In order to facilitate the signal calculation processing, the intonation detecting section 19 is provided with an absolute value converting section 52. The absolute value converter 52 converts the amplitude of the input signal into its absolute value. Accordingly, a signal SG3 shown in FIG. 9 appears at the output of the absolute value converter 52. The comparing unit 53 compares the magnitude of the signal SG3 with a threshold and outputs only a component larger than the threshold as the signal SG4. That is, the comparison unit 53 outputs the signal S
Only a component having a large value in the power spectrum of G3 is output. Note that the threshold value applied to the comparison unit 53 is
It is determined adaptively using a method called discriminant analysis.

Referring to FIG. 9, a signal SG4 includes two areas A1 and A2 corresponding to the intonation pattern in a speech word.
A2 clearly appears. The area center detection unit 54 detects times t1 and t2 when a position corresponding to the center of each of the two areas A1 and A2 appears. Area interval detector 55
Are the two times t1, t detected by the area center detection unit 54.
2 is detected as the region interval Ty. The value of the area interval Ty corresponds to an intonation pattern in a voice word. In practice, the result of averaging the values of the area intervals Ty is used as the value of the intonation.

In some cases, three or more regions appear in signal SG4 in one word. When three or more areas appear, the area intervals Ty are calculated for two areas adjacent to each other, and the plurality of obtained area intervals Ty are calculated.
The result of averaging y is used as the value of the intonation. The state of human emotion changes, for example, as shown in FIG. Further, in order to correctly grasp emotions such as anger, sadness, and joy, it is important to detect a change in a feature amount such as intensity, tempo, and intonation.

In the emotion detection device shown in FIG. 6, in order to make it possible to refer to past feature values, the intensity output by the intensity detection unit 17, the tempo output by the tempo detection unit 18, and the intonation detection unit 19 output. The value of the intonation is temporarily stored in the temporary storage unit 21. In addition, the emotion change detection unit 22 outputs the current intensity output from the intensity detection unit 17 and the tempo detection unit 18.
Output the current tempo and the current intonation value output by the intonation detecting unit 19 and the past (at a slightly earlier time) intensity, tempo and intonation value stored in the temporary storage unit 21. Input to detect changes in emotional state. That is,
A change in voice intensity, a change in tempo, and a change in intonation are detected.

[0092] The voice emotion detecting section 23 includes the emotion change detecting section 2.
2, a change in the intensity of the voice output, a change in tempo, and a change in intonation are input, and the current state of emotion is estimated. In this example, three types of states of anger, sadness, and joy are estimated as emotion states. Emotion pattern DB24
Is related to the information that associates the pattern of changes in voice intensity, tempo, and inflection with the state of anger and the pattern of changes in voice intensity, tempo, and inflection is related to the state of sadness Information and information relating the pattern of the change of the voice intensity, the change of the tempo, and the change of the intonation to the state of joy are stored in advance.

[0093] The voice emotion detection unit 23 is provided with an emotion pattern DB.
While referring to the information held in 24 as an estimation rule,
The current state of emotion is estimated based on patterns of intensity change, tempo change, and intonation change output from the emotion change detection unit 22. The information representing each of the three states of anger, sadness, and joy estimated by the voice emotion detection unit 23 is input to the emotion recognition unit 60 and the emotion and emotion storage DB 41. The emotional emotion storage DB 41 sequentially stores and accumulates the current emotional state input from the voice emotional detection unit 23.

Therefore, the past emotional state can be reproduced by reading the information stored in the emotional emotion storage DB 41. On the other hand, the content of a sentence (speech content of the other party) input from the microphone 11 as voice is recognized by the sentence recognition unit 26. To the input of the sentence recognizing unit 26, character information corresponding to each phoneme recognized by the speech recognizing unit 20, information indicating a word break and a sentence break is input.
The character information input from the keyboard 25 is also input to the text recognition unit 26.

The sentence recognizing unit 26 performs recognition and syntax analysis of the input character string for each word, and grasps the contents of the sentence as a natural language. Actually, it recognizes utterance information indicating “5W3H”, that is, “who”, “what”, “when”, “where”, “why”, “how”, “how much”, “how much”. The utterance information recognized by the text recognition unit 26 is input to the emotion recognition unit 60.

Next, a process for detecting an emotion from the facial expression of the partner will be described. The television camera 31 captures an image of at least a face of a human being who is a subject of the emotion detection device in FIG. An image captured by the television camera 31, that is, an image including the expression of a human face, is
Is input to The information of the image captured by the television camera 31 is input to the character recognition unit 39. That is, when the video of the text is shot by the television camera 31, the character recognition unit 39 recognizes each character of the text from the shot video. The character information recognized by the character recognition unit 39 is sent to the text recognition unit 2
6 is input.

The image recognition section 32 recognizes characteristic elements from the input image. Specifically, the eyes, mouth, eyebrows, and cheekbones of the subject's face are respectively recognized, and the relative positions of the eyes, mouth, eyebrows, and cheekbones in the face are detected. Further, the image recognition unit 32 always tracks the position in order to detect changes in the positions of the eyes, mouth, eyebrows, and cheekbones due to changes in the facial expression, and expressions such as shaking the head.

The face pattern DB 33 holds in advance reference position information (information corresponding to the normal facial expression of the subject) regarding the positions of the eyes, mouth, eyebrows, and cheekbones in the face. Note that the content of the face pattern DB 33 can be arbitrarily changed. The face pattern DB 33 holds in advance rule information indicating a correspondence between a change in the facial expression and each of the six types of emotions (joy, anger, sadness, fear, enjoyment, and surprise).

The face emotion detecting section 34 uses the respective positions of the eyes, mouth, eyebrows, and cheekbones recognized by the image recognizing section 32 and the information of the reference position held in the face pattern DB 33, that is, the feature amount, that is, the normal state. The difference of the expression with respect to the position of is detected. In addition, the face emotion detection unit 34 detects six types of emotions (joy, anger, sadness, fear, enjoyment, fun, anger, etc.) based on the detected change amount and change speed of the feature amount and the rule information stored in the face pattern DB 33. Surprise) to estimate each state. Information representing the estimated six types of emotion states is output from the facial emotion detection unit 34 and input to the emotion recognition unit 60 and the emotion and emotion storage DB 41.

The emotion recognizing section 60 includes information indicating the state of the emotion (anger, sadness, joy) input from the voice emotion detecting section 23, utterance information input from the sentence recognizing section 26, and face emotion detecting section 34. The final emotional state is estimated by comprehensively judging the information representing the state of the emotion (joy, anger, sadness, fear, enjoyment, surprise) input from. For the statement information, the state of the emotion (joy, anger, sadness, fear, enjoyment, surprise) contained in the statement information is estimated by determining the content (5W3H) of the sentence according to a predetermined rule. Can be.

The information indicating the state of the emotion estimated from the voice by the voice emotion detection unit 23, the information on the speech content recognized by the text recognition unit 26 from the voice or the character input from the keyboard 25, and the face emotion detection unit 34 The information representing the state of the emotion estimated from the facial expression is stored in the emotional emotion storage DB 41.
And are sequentially stored. Each piece of information stored in the emotional emotion storage DB 41 is added with the time or time when the information was detected and the date.

Among the information input to the emotional emotion storage DB 41, information on the emotion input from the voice emotion detection unit 23, information on the utterance content input from the text recognition unit 26,
The emotion information input from the face emotion detection unit 34 must be grasped in association with each other. Therefore, the synchronization processing unit 43 associates a plurality of types of information stored in the emotional emotion storage DB 41 with each other based on the detected time (input time) and the date. For example, as shown in FIG. 11, the anger estimated by the voice emotion detection unit 23,
Information indicating the state of emotion of sadness and joy and information of the content of the statement (5W3H) are associated with each other by their time.

Incidentally, the emotional emotion storage DB 41 has a sufficient storage capacity for storing a relatively large amount of information. However, since the storage capacity is limited, it is necessary to suppress the amount of information to be stored in order to continue using this device for a long period of time. Therefore, a forgetting processing unit 42 is provided. The forgetting processing unit 42 automatically deletes outdated information from the emotional and emotional storage DB 41. However, information meeting specific conditions is stored without being deleted even when it becomes old.

The operation of the forgetting section 42 will be described with reference to FIG. In step S11 of FIG. 10, the stored time (or detected time) and the date information are referred to for each of a large number of data stored in the emotional emotion storage DB 41. In step S12, based on the current time and the time of the data referred to in step S11, it is determined whether or not a predetermined period has elapsed since the corresponding data was stored. When processing old data for which a certain period has elapsed since the storage, the process proceeds to step S13 and subsequent steps. Relatively new data for which a certain period has not elapsed is stored as it is.

In step S13, if the data is information indicating the state of the emotion, the amount of change in the emotion (the difference between the previous and next emotions) is examined. If the amount of change in emotion exceeds a predetermined threshold, the process proceeds from step S13 to S17, so that even if the data is old, the data is stored as it is. If the emotion change amount is equal to or smaller than the threshold, the process proceeds from step S13 to S14.

In step S14, an emotion pattern relating to the data is detected, and it is determined whether or not the pattern matches a predetermined specific pattern. That is,
It is checked whether or not a combination of a plurality of emotion states and speech contents matches a specific pattern representing a “strong impression” state. If the detected pattern matches the specific pattern, the process proceeds from step S14 to S17, so that even if the data is old, the data is stored as it is. If the patterns do not match, the process proceeds from step S14 to S15.

In step S15, if the data is the content of a comment, it is determined whether or not the content matches a predetermined content of the comment (a message that is likely to leave an impression). In addition,
"Match" if there is high similarity, even if they do not match exactly
Can also be considered. If the content of the data utterance matches the content of the predetermined utterance, steps S15 to S1 are performed.
Since the process proceeds to step 7, even if the data is old, the data is stored as it is.

If the data does not match in step S15, the data is deleted in step SS16. The above processing is executed for all data on the emotional emotion storage DB 41. Further, the forgetting process shown in FIG. 10 is periodically and repeatedly executed. The period of execution of the forgetting process can be arbitrarily changed as the personality of the individual.
In steps S14 and S15, processing is performed with reference to a previously facilitated pattern DB (not shown). The contents of the pattern DB are automatically updated by learning the input information.

In FIG. 10, the processing is simplified. In practice, the amount of change in the emotion, the pattern of the emotion, and the content of the statement are all comprehensively determined. That is, when there is information with a large amount of change in emotion, information with a matching emotion pattern, and information with the same or similar utterance content, the overall priority is determined. In particular,
Information with the same or similar remark has the highest priority, and information with the same emotion pattern has the second priority.
The information with the second highest ranking and the largest change in emotion has the lower priority. Therefore, information with the same or similar remark contents is hard to be deleted by the forgetting process, and remains as a memory even when it becomes old.

As a result of the processing of the forgetting processing unit 42 as described above, old data in the emotional emotion storage DB 41 may have a large emotional change, a pattern that is regarded as having a “strong impression”, or a number of times. Only those that have been repeatedly input and those whose utterance content is likely to remain impressive are stored in that order according to their strength and content. As a result, old data in the emotional emotion storage DB 41 becomes incomplete data in which only a part remains.
Such data has the same content as the past vague memory in human memory.

By reading out the past emotional states and remark contents accumulated in the emotional emotion memory DB 41 and analyzing the data, for example, it is possible to judge whether or not this emotion detection device is operating correctly, or to estimate the emotions. It can also be updated to improve the contents of the database of each unit used for. The data stored in the emotional and emotional memory DB 41 is further sorted according to the content thereof, and is classified into the humanity information DB 44, the personal information DB 45,
6 is stored.

The humanity information DB 44 holds information that determines the subject's character and information on the behavior determination pattern, such as gender, age, aggression, coordination, and current emotion. The personal information DB 45 holds information such as the address of the individual, the current situation, the environment, and the content of the statement (5W3H). The specialized information DB 46 holds information such as occupation, career, occupational aptitude, and occupational behavior decision pattern.

Output from the personality information DB 44, the personal information DB 45, and the specialized information DB 46 are moral pattern information of an individual. Based on the moral pattern information and the past emotion of the partner, the sensitivity of the partner can be detected. The microphone 11 shown in FIG. 6 may be replaced by a telephone receiver, or a mouse may be provided as a means for inputting information such as characters.

Further, the television camera 31 shown in FIG. 6 can be replaced by any of various image pickup means such as an optical camera, a digital camera, and a CCD camera.

[0115]

As described above, in the present invention, instinctual motivation information for generating emotions is generated based on input situation information (emotions of the other party, surrounding situations, etc.). Then, emotion information is generated based on the instinctive motivation information. Further, the generated emotion information is controlled according to the personality information.

Therefore, it is possible to output emotion controlled by the individual's reason and will, that is, information on sensibility. Further, since emotion information is generated through instinctual motivation information, the generated emotion can be controlled more precisely and easily.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration example of a system using a sensibility generating device.

FIG. 2 is a block diagram illustrating a configuration of an instinct information generation unit.

FIG. 3 is a block diagram illustrating a configuration of an emotion information generation unit.

FIG. 4 is a schematic diagram showing an example of a reaction pattern model in an emotion reaction pattern DB.

FIG. 5 is a block diagram illustrating a configuration of a sensible thinking recognition unit.

FIG. 6 is a block diagram illustrating a configuration of an emotion detection device.

FIG. 7 is a block diagram illustrating a configuration of an intonation detecting unit.

FIG. 8 is a graph showing a relationship between a change in an emotional state and a sound intensity, tempo, and intonation.

FIG. 9 is a time chart showing a process of audio signal processing in the intonation detecting section.

FIG. 10 is a flowchart illustrating an operation of a forgetting processing unit.

FIG. 11 is a schematic diagram illustrating a configuration example of information stored in an emotional emotion storage DB.

[Explanation of symbols]

 Reference Signs List 11 microphone 12 A / D converter 13 signal processing unit 14 phoneme detection unit 15 word detection unit 16 sentence detection unit 17 intensity detection unit 18 tempo detection unit 19 intonation detection unit 20 voice recognition unit 21 temporary storage unit 22 emotion change detection unit 23 Voice emotion detection unit 24 Emotion pattern DB 25 Keyboard 26 Sentence recognition unit 31 TV camera 32 Image recognition unit 33 Face pattern DB 34 Face emotion detection unit 39 Character recognition unit 41 Emotional emotion storage DB 42 Forgetting processing unit 43 Synchronization processing unit 44 Human nature Information DB 45 Personal Information DB 46 Specialized Information DB 51 Bandpass Filter 52 Absolute Value Transformer 53 Comparison Unit 54 Area Center Detector 55 Area Interval Detector 60 Emotion Recognition Unit 100 Sensitivity Generator 110 Instinct Information Generator 111 Prosody Pattern Synchronization Recognition unit 112 Instinct parameter generation unit 113 Dictionary Cable part 121 prosodic pattern DB 122 instinct language meaning dictionary 130 emotion information generation part 131 response system 132 operant system 133 basic emotion parameter generation part 134 reaction pattern search part 135 learning processing part 136 multivariate analysis part 137 spontaneous emotion control part 141 Emotion reaction pattern DB 142 Temporary storage DB 143 Personality information 150 Sensibility thinking recognition unit 151 Weighting processing unit 152 Collation processing unit 153 Multivariate analysis unit 154 Overall intuition determination unit 155 Personality information 156 Update processing unit 157 Learning processing unit 161 Knowledge DB 162 Kansei DB 163 Personal DB 164 Moral hazard DB 200 Emotion detection device 300 Environmental information input device 410 Conversation control unit 420 Conversation DB 430 Audio output unit 440 Audio DB 450 Speaker

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI theme coat ゛ (Reference) G10L 15/24 G10L 3/00 571Q 15/22 571T 15/02 7/08 A

Claims (10)

[Claims] 1. At least personality information for determining at least a reason, a predetermined characteristic, and a will of a sensibility generating object is stored in advance, and situation information representing an emotion or environment state of a partner is input, and based on the situation information, Generating instinct motivational information including at least a first instinct parameter indicating a degree of pleasure, a second instinct parameter indicating a degree of risk, and a third instinct parameter indicating a degree of achievement and a degree of change. And generating emotion information including at least basic emotion parameters of joy, anger, and sadness based on the generated instinctive motivation information, and controlling the generated emotion information based on the personality information. A sensibility generating method characterized by the following. 2. An episode situation information indicating a state of an emotion, an environment, and a will of a partner is input. Based on the episode situation information, a first instinct parameter indicating at least a degree of pleasure, and a second instinct parameter indicating a degree of risk Instinct determination means for generating instinct motivation information including a second instinct parameter and a third instinct parameter indicating a degree of achievement or a degree of change; and based on instinct motivation information output from the instinct determination means. And emotion generating means for generating emotion information including at least basic emotion parameters of joy, anger, and sadness; personality information providing means for providing personality information for determining at least the reason and emotional intention of the sensibility generating object; And an emotion control unit for controlling the emotion information output from the emotion generation unit based on the personality information provided from the information provision unit. A sensibility generator characterized by the following. 3. The sensibility generating device according to claim 2, wherein the emotion generating means includes: a life rhythm generating means for generating information representing a periodically changing environment or a living rhythm of a living body; A spontaneous emotion control unit for controlling a spontaneous emotion in the emotion generation unit in accordance with a probability corresponding to information on a living rhythm to be performed. 4. The emotion generating apparatus according to claim 2, wherein the emotion generating means includes: an instinct-emotion correspondence information holding means for holding pattern information for associating the basic emotion parameter with the instinct motivation information; A matching probability learning means for outputting information indicating a probability of matching / mismatch between the instinct motivational information output from the determining means and the pattern information of the instinct-emotion correspondence information holding means; Kansei generator. 5. The emotion generating device according to claim 2, wherein said emotion generating means includes at least emotion feedback control means for inputting at least the last generated own emotion information and reflecting the information in the next generated self emotion information. A sensibility generator characterized by being provided. 6. The emotion generating device according to claim 2, wherein said emotion control means reflects information of a life rhythm, which is a personality of the emotion generating target, to the input emotion information. . 7. The sensibility generating device according to claim 2, wherein a knowledge database for storing situation information indicating past situations, episodes and results thereof, and the past situation information similar to newly input situation information is stored in the knowledge database. A knowledge matching means for searching and extracting from the database and providing past situation information to the emotion control means, and updating the content of the knowledge database with situation information indicating a newly inputted situation and its result,
Data update control means for automatically deleting status information having a low priority among the status information accumulated in the knowledge database in chronological order according to the weight of the content is further provided. Kansei generator. 8. The sensibility generating device according to claim 2, wherein: a voice input unit that inputs a voice signal; an intensity detection unit that detects a voice intensity from the voice signal input by the voice input unit; Tempo detection means for detecting the appearance speed of voice as a tempo from the input voice signal; intonation detection means for detecting intonation representing an intensity change pattern in a word of voice from the voice signal input by the voice input means; A change amount detecting means for obtaining a change amount for each of the intensity of the sound detected by the detecting means, the tempo of the sound detected by the tempo detecting means, and the intonation of the sound detected by the intonation detecting means; And emotion detecting means for outputting a signal representing at least the respective emotional states of anger, sadness, and joy based on the amount of change. A feeling generating device characterized by the following. 9. The sensibility generating device according to claim 8, wherein a voice recognition unit that recognizes voice input from the voice input unit and outputs character information, and outputs the voice information recognized by the voice recognition unit in a natural language. And a natural language processing means for generating semantic information representing the meaning of the input speech after processing. 10. A software comprising a computer-executable program and data used for emotion generation control, wherein the program inputs situation information indicating a feeling of an opponent or a state of an environment. Based on the information, an instinct motivation including at least a first instinct parameter indicating a degree of pleasure, a second instinct parameter indicating a degree of risk, and a third instinct parameter indicating a degree of achievement or a change. A procedure for generating information, a procedure for generating emotion information including at least basic emotion parameters of joy, anger, and sadness based on the generated instinctual motivation information; and determining at least a reason and a will of a sensibility generating object. A step of providing personality information to be attached; and a step of controlling the generated emotion information based on the personality information. Software which is characterized in.