JP3018865B2 - Emotion expression device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an emotion expression device which has artificial emotions and changes and outputs facial expressions and voices according to the emotional state.

[0002]

2. Description of the Related Art In a book entitled Machine with Heart,
As Masahide Toda pointed out (1987), if machines such as computers and robots have the same emotions as human beings, and can be happy, sad, and caring for humans, humans and machines The communication of
It will be much warmer, more fun, and more human friendly.

[0003] There is such a practical expectation behind attempts to provide artificial emotions to machines, especially dialog systems. To date, various emotion expression devices have been proposed, and most of them have basically the following configuration. Hereinafter, an example of a conventional emotion expression device will be briefly described with reference to FIG.

[0004] Although this type of emotion expression device is used by being incorporated into a dialogue system such as a voice dialogue device, the dialogue system will not be described here to avoid complication. FIG. 14 is a basic configuration diagram illustrating an example of a conventional emotion expression device, where 01 is an input, 02 is a recognition unit, 03
Is an emotion eliciting unit, 04 is an expression / speech synthesis unit, and 05 is an output.

[0005] In FIG. 1, the meaning of input information 01 to the emotion expression device is recognized by a recognition unit 02. If the recognized meaning corresponds to a pre-specified emotion-evoking input, the emotion-evoking unit 03 may increase the intensity of the emotion component Y by Z if the emotion-evoking input X is recognized. According to the rules, the intensity value of the emotional state register included in the emotional arousal unit 03 is rewritten.

Here, the emotion state register is, for example,
This register stores the intensity value of the emotion component such as “surprise, fear, disgust, anger, happiness, sadness”. The intensity value of the emotional state register is changed according to the emotional arousal rule each time the emotional arousal input is recognized as described above,
In some cases, it is set to fluctuate depending on factors such as the state of the preceding emotion and the interaction between the basic emotions.

[0007] In addition, in order to return the evoked emotions to a steady state, there are many cases in which the intensity value of the emotion state register is attenuated over time. Now, when the intensity value of any emotion component in the emotion state register exceeds the expression threshold defined in advance, the expression / speech synthesis unit 04 synthesizes an expression or voice corresponding to the emotion. Otherwise, it synthesizes an emotionless expression and a voice without emotional inflection. The synthesized expression and voice are output (05) to the interlocutor via the display device or voice output device.

[0008] For details of the technology related to such an emotion expression device, see, for example, “Information Processing Society of Japan 45th (Heisei 4)
"Letters of National Conventions 2", "Emotion Model of Anthropomorphic Agents" (Hiroshi Kawamoto, Kengo Omura), and "Japanese Cognitive Science Association 1992 Regular Symposium, Cognitive Science of Emotions". "Artificial Emotions by Harmony Theory"
(Written by Fumio Hara and Shusuke Mogi).

[0009]

However, the above-mentioned conventional emotion expression device has the following problems. (1) The conventional emotion expression device is provided with a means by which a user can freely set and change the emotion expression style of the emotion expression device according to his / her taste or change in taste. Absent.

[0010] For example, the emotion of joy often appears,
I don't want to change my style of disgust, anger, and sadness so that it rarely appears. Therefore, a first object of the present invention is to allow a user to freely set and change the emotion expression style of the emotion expression device. (2) A conventional emotion expression device is a device for communicating with a conversation partner, a situation, a topic,
The tendency of emotional expression cannot be automatically changed to an appropriate one based on a difference in the past conversation amount or the like. That is, it does not have a function of controlling emotional expression based on so-called “display rules”.

[0011] Here, the "display rules" are originally defined as social and cultural norms on how to express emotions in various social situations and roles. We say custom. It is preferable that the emotion expression device also complies with the display rules as long as it interacts with humans, but the existing device cannot do so. As described above, the display rules are originally defined socially. However, in the case of an emotion expression device, it is convenient if the user can define the display rules.

Therefore, a second object of the present invention is to automatically change the tendency of emotional expression to an appropriate one based on a difference in a conversation partner, a situation, a past conversation amount, and the like. On the point.

[0013]

In order to achieve the first object, an emotion expression device of the present invention, as shown in FIG. 1, has an emotion-evoking information recognition device for recognizing predetermined emotion-evoking information. Means 2 and each emotion component such as "surprise, fear, disgust, anger, happiness, and sadness" are given a quantity reflecting the emotion intensity to express the overall emotional state, and the emotional arousal information is recognized. Includes an emotion evoking means 3 for transiting the emotional state in accordance with a predetermined emotional state transition rule;
Expression / speech synthesis means 4 for changing and outputting facial expressions and voice prosody in correspondence with the emotional state specified by the emotion evoking means, and emotion expression of the apparatus by designating each expression threshold for each emotion component A presentation style setting unit 6-1 for setting or changing a style, and a presentation style storage unit 6-21 for holding a presentation threshold set in the presentation style setting unit 6-1 are included. An expression content determining means 6-2 for comparing the expression threshold held therein with the emotional state specified by the emotion evoking means 3 to determine an expression or voice to be expressed, and an expression style setting means Is provided with an expression style group holding unit 6-3 capable of storing and storing a plurality of expression thresholds set in the above.

Further, in order to achieve the second object,
As shown in FIG. 1, the emotion expression device of the present invention includes a speaker identification unit 6-41 for identifying and identifying a conversation partner and a past history of the speaker identified by the speaker identification unit 6-41. A dialogue amount storage means 6-42 for measuring and recording the number of dialogues or a total dialogue time; a situation recognition means 6-43 for identifying a current situation from a plurality of situations categorized in advance; A specific expression threshold is read out from the expression threshold set stored in the expression style group holding means 6-3 based on the difference in the situation, and the expression style storage means 6-3 is read out.
21. An expression style changing means 6-5 set to 21. Here, the expression style changing means 6-5 selects a specific conversion equation from some predetermined conversion equations, and stores it in the expression style storage means 6-21. It is also possible to adopt a method of directly rewriting by adapting to the present expression threshold.

[0015]

In the emotion expression device for achieving the first object, the emotion-recognition-information recognizing means 2 recognizes predetermined emotion-reduction information from the input information 1. The emotion-evoking means 3 is provided to each emotion component such as "surprise, fear, disgust, anger, happiness, sadness" based on the recognition result of the emotion-evoking information recognizing means 2 and a predetermined emotion state transition rule. The intensity value pattern is changed to define the current emotional state.

On the other hand, the expression style setting means 6-1 allows the user 7 to set an expression threshold value corresponding to each emotion component,
It defines the overall expression style. The expression style storage means 6-21 stores the expression style setting means 6-
The expression threshold set in 1 is held. The expression content deciding means 6-2 includes an intensity value of each emotion component defined by the emotion eliciting means 3 and the expression style storage means 6-6.
The expression threshold and the type of voice to be expressed are determined according to a predetermined rule with reference to the expression threshold of each emotion component held at 21.

The expression / speech synthesizing means 4 synthesizes and outputs the expression and the speech based on the determination of the expression content determining means 6-2. The expression style group holding unit 6-3 holds an expression threshold pattern that reflects some representative expression styles, and sets the table set by the expression style setting unit 6-1. The output threshold pattern is also held.

The user 7 sets the expression style setting means 6
A specific expression threshold pattern can be read out from the expression style group holding means 6-3 via -1 and set in the expression style storage means 6-21. Further, in the emotion expression device for achieving the second object, the speaker identification means 6-41 identifies and identifies a conversation partner.

The conversation amount storage means 6-42 measures and records the number of past conversations or the total conversation time with the speaker identified by the speaker identification means 6-41. Further, the situation recognition means 6-43 recognizes the current situation from a plurality of situations that are categorized in advance. The expression style changing means 6-5 includes:
In accordance with these recognition results and a predetermined rule (display rule), a specific expression threshold is read out from the expression threshold set stored in the expression style group holding means, and is read out from the expression threshold. The expression style storage means 6-21 is set in the style storage means 6-21, or a specific conversion equation is applied from some predetermined conversion equations.
, The display style can be automatically changed based on the difference between the conversation partner, the conversation amount, and the situation.

[0020]

Embodiments of the present invention will be described below in detail with reference to the drawings. In this embodiment, the emotion expression system according to the present invention is applied to a voice interactive electronic secretary. FIG. 2 is a block diagram illustrating a system of a voice interactive electronic secretary, wherein 7 is a user, 8 is a speaker identification unit, 9 is an utterance understanding unit, 10 is an intention estimation unit, 11 is a task execution unit,
12 is a secretary task management information database, 13 is an emotion management unit, 14 is a voice synthesis / image display unit, 15 is a voice data storage unit, and 16 is a facial expression image data storage unit.

In FIG. 1, a speaker specifying unit 8 is a subsystem for specifying a speaker, an utterance understanding unit 9 is a subsystem for recognizing the symbol content of the utterance, and an intention estimating unit 10 is a subsystem for estimating the meaning of the utterance. The task execution unit 11 is a subsystem for planning and executing an action related to a task based on the intention of the user 7, the secretary task management information database 12 is a database storing various information related to the task, and the emotion management unit 13 is an artificial emotion. The voice synthesis / image display unit 14 performs a voice response / image display based on task execution, and the voice data storage unit 15 stores voice response information configured for each emotion type. And a facial expression image data storage unit 16 is a database for storing secretary moving image information configured for each emotion type.

First, the overall mechanism of the electronic secretary will be outlined based on the above block diagram, and then each subsystem constituting the system will be described. [Overall Configuration] The voice input from the user 7 to the electronic secretary is passed to the speaker specifying unit 8 and the utterance understanding unit 9 to specify the speaker and, at the same time, recognize the symbolic content of the utterance.

From the symbolic content of the utterance, the intention estimating unit 1
At 0, the task request intention of the user is further estimated and passed to the task execution unit 11. On the other hand, the speaker identification information and the utterance content are input from the speaker identification unit 8 and the utterance understanding unit 9 to the emotion management unit 13. The emotion management unit 13 manages generation and expression of emotions based on these two pieces of information and the event occurrence state in the task execution unit 11. The expression management here is a mechanism for controlling how much the generated emotional state is expressed in consideration of the interlocutor and the dialogue situation.

In the task execution section 11, a secretary task management information database 12 is provided in accordance with the estimated user's intention.
And perform tasks such as schedule and document management in a plan. At the same time, a request for emotion change based on the user's intention is sent to the emotion management unit 13. Conversely, the emotion state in the emotion management unit 13 is determined by the task execution unit 1.
1, which affects plan selection and execution.

That is, when a series of planned actions are performed (or failed), the result is transmitted to the emotion management unit 13 to cause a change in the emotional state. The emotion content determined to be expressed by the emotion management unit 13 is generated by the voice synthesis / image display unit 14 based on the voice data and the facial expression image data stored in the voice data storage unit 15 and the facial expression image data storage unit 16. Is output as a moving image or uttered voice.

The emotion expression contents determined by the emotion management unit 13 are concretely output by the speech synthesis / image display unit 14 and presented to the user 7. Hereinafter, the emotion management unit which is a main component of the present invention will be described in detail. [Emotion Manager] FIG. 3 is a block diagram for explaining the structure of the emotion manager together with related system modules.
Reference numeral 13 denotes an emotion management unit, 13-1 denotes an emotion generation unit, 13-2 denotes an emotion expression control unit, and the same reference numerals as those in FIG.

In FIG. 1, the emotion management unit 13 includes an emotion generation unit 13-1 and an emotion expression control unit 13-2. The emotion generation unit 13-1 is a system that generates and holds artificial emotions by stimulation from the environment in which the electronic secretary is active. For example, Japanese Patent Application No. 4-1695 filed by the present inventors.
It is possible to adopt a configuration as disclosed in No. 74.

The emotion expression control section 13-2 includes the emotion generation section 1
This is a system for automatically / forcibly controlling how to express the emotional state held in 3-1. The automatic control is a control for automatically adapting and changing the expression style in accordance with the conversation partner of the electronic secretary, the activity time, the conversation history, and the like. This adaptive change can lead the electronic secretary's dialogue attitude to a natural style appropriate to the situation.

On the other hand, the compulsory control is a mechanism by which a person who interacts with the electronic secretary, that is, a user, adjusts the tendency of the electronic secretary to express emotions according to his / her own preference. With this control mechanism, a user who does not express emotion can set a completely cool electronic secretary, or a user who does not allow the electronic secretary to get angry can set not to express only anger .

The emotion expression control section 13-2 transmits the emotion state in the emotion generation section 13-1 to the speech synthesis / image display section 14 under such two kinds of control. [Emotion Expression Control Unit] FIG. 4 is a block diagram for explaining the configuration of the emotion expression control unit.
13-22 is a scene composing unit, 13-23 is an artificial emotion value register (Re), and 13-24 is an expression value register (R).
r) and 13-25 are expression characteristic threshold value registers (Rt), 1
3-26 is a display mode flag (Fm) register, 13
-27 is an expression characteristic editor, 13-271 is an expression characteristic threshold setting changer, 13-272 is a display mode / flag setting changer, 13-273 is an expression characteristic rule editor, 13-27.
4 is an expression characteristic threshold pattern editor, 13-28 is an expression characteristic rule interpreter, 13-29 is an expression characteristic rule database, and 13-30 is an expression characteristic threshold pattern database.

In the figure, the subsystems constituting the emotion expression control section include (1) a register group (artificial emotion) for holding an emotion state, a display mode, an expression characteristic, and a final expression value. Value register (Re) 13-23, expression value register (Rr) 13-24, expression characteristic threshold value register (Rt) 13-25), (2) speaker identification unit 8, utterance understanding unit 9, intention estimation Expression characteristic rule interpreter 13-28, expression characteristic rule database 13-29, and expression characteristic threshold pattern for automatically determining and changing the expression characteristic threshold from the status information from the section 10 and the task identification section 11. A database 13-30; (3) an expression value determination unit 13-21 for determining a final expression value; and (4) an expression characteristic editor for receiving user settings relating to expression characteristics and display modes. 13-27 and (5) a scene components 13-22 constituting the scene to be output from the determined expression values, and a.

Hereinafter, the process on this subsystem will be described. 1. Process on Expression Characteristic Rule Interpreter 13-28 Information obtained from the speaker identification unit 8, the utterance understanding unit 9, the intention estimation unit 10, and the task execution unit 11 is sent to the expression characteristic rule interpreter 13-28. . The expression characteristic rule is a group of rules for automatically changing the expression characteristic according to the above-mentioned situation information, and the expression characteristic rule interpreter 13-28
Is a mechanism for setting various conditions in the expression characteristic rules in the expression characteristic rule database 13-29, and applying rules satisfying the conditions at each time.

The following [Equation 1] to [Equation 6] are examples of expression characteristic rules (display rules) in the expression characteristic rule database 13-29.

[0034]

(Equation 1)

Meaning of the rule IF: The speaker is not included in the user list, that is,
This is the case where a speaker of knowledge is supported. THEN: Expression threshold Θ_1,2To the suppression coefficient DEPRESS
Multiply by RATE (> 1.0). Start suppressing expression
Control indicating that Turn on flg. )

[0036]

(Equation 2)

Meaning of Rule IF: The speaker is included in the user list. In other words, the partner currently responding is a known speaker. THEN: The threshold of the corresponding person (speaker) in the user list is set to the expression threshold Θ _1,2 . In addition, FRIEDLY Multiply RATE (<1.0) to increase the frequency of emotional expression. Number of conversations contact ti
One is added to me.

ANDIF: the speaker is MY MASTER
R'S FRIEND, which is included in the main user's friend list. THEN: The expression suppression flag is turned off to cancel the expression suppression.

[0039]

(Equation 3)

Meaning of rule IF: Time division setting (current time ty
pe) is out of definition. That is, the setting of the time type (on duty, off duty) is incorrect. THEN: Reset the time division setting to work (or off work). FORMAL for expression characteristic threshold RATE (or RELAX RATE) to change.

[0041]

(Equation 4)

Meaning of Rule IF: Elemental emotion in emotion model becomes angry or hatred
And the value is the expression threshold Θ_1,2In
Exceeds the strong threshold and the user utterance word is NEG
ATIVEWORD is included. THEN: ANG for anger and disgust components in the expression threshold
ER DEPRESS Multiply by RATE (> 1.0) and subtract
Lift up.

The expression suppression flag is turned on to start expression suppression. C to the threshold of the current interlocutor in the user list
LOSE Multiply by RATE (> 1.0) and raise.

[0044]

(Equation 5)

Meaning of Rule IF: A signal indicating that "task failure continues" is received by the task execution unit. THEN: DEPRESS as expression threshold RATE (> 1.
0) to raise the overall threshold. The expression suppression flag is turned on to start the expression suppression.

[0046]

(Equation 6)

Meaning of Rule IF: POSITIVEWORD is included in the utterance of the user. THEN: OPEN for expression threshold DEPRESS RA
Multiply by TE (<1.0) and lower. The expression suppression flag is turned off to cancel the expression suppression.

Note that "USER LI" in [Equation 1]
ST] has a format as shown in the following [Table 1].

[0049]

[Table 1]

Also, Θ _1,2 is the expression characteristic threshold, as shown in [Table 2] below.

[0051]

[Table 2]

All of these rules that satisfy the condition part are sequentially executed, and a final change of the expression characteristic threshold is proposed to the expression value determination part. 2. Process on Expressed Value Determination Unit The following signals are sent to the expressed value determination unit 13-21. (1) Expression characteristic change from the expression characteristic rule interpreter 13-28. (2) Emotion change notification from the emotion generation unit 13-1. (3) Task execution event occurrence from the task execution unit 11.

The expression value determination unit 13-21 determines the expression based on these requests and the setting of the display mode flag. Figure 5 is an explanatory view of the mode flag Fm, emotional expression permits f _e, manual <-> consisting of two automatic priority f _p. Also, the expression value determination unit 13-21 has three registers and one flag register shown in FIG.

FIG. 6 is an explanatory diagram of an example of a register group and a stored value in the emotion expression control section. The artificial emotion value register (Re), the expression characteristic threshold register (Rt), and the expression value register (Rr). Of three registers. The artificial emotion value register (Re) stores the emotion state value sent from the emotion generation unit 13-1. The expression characteristic threshold value register (Rt) holds a threshold value for determining whether or not emotion expression should be performed. And
The expression value register (Rr) is mainly composed of an expression value determination unit (Re) and an expression characteristic threshold value register (R).
The stored expression value of the emotion determined by comparing the expression with t) is stored.

The display mode flag in FIG.
The register (Fm) sets various control information related to the display as shown in FIG. FIG. 7 is a flow chart illustrating a process on the expression characteristic rule interpreter, and includes a display mode flag register (Rm).
When the display mode flag _fe for instructing start / stop of the process is output from the CPU, the expression characteristic control process is started (S-1), and the speaker specifying unit, the utterance understanding unit, the intention estimating unit, and the task execution Wait for an event signal from the section (S-
2).

A rule condition section is set in accordance with the input of the event signal from each of the above sections (S-3), and the set rule is expressed by the expression characteristic rule database and the expression characteristic threshold pattern database. Execute the rule matching with the data of (S-
4). The result of executing the rule matching is passed to the expression value determination unit (S-5). Next, a description will be given of a process based on an expression characteristic change signal which is the basis in the expression value determination unit.

[0057] exposed value determining process 8 is a flowchart for explaining the outline of processes on exposed value determination section, the process of the exposed value determining unit S-1
0 is an expression value determination process S-102 for rewriting an expression characteristic value based on an emotion change occurrence signal from the emotion generation unit 13-1 and an expression characteristic change request from the expression characteristic rule interpreter 13-28; And an emotion giving process S-103 for giving a feeling to the task performing scene based on the task information from the task performing unit 11.

The process execution result of the process S-10 on the expression value determining unit is given to the scene composing unit 13-22. Hereinafter, the details of the expression request process, the expression value determination process, and the process of giving an emotion to the task execution, which constitute the processes on the expression value determination unit, will be sequentially described. Expression value determination process Figure 9 and Figure 10 is a flowchart for explaining the exposed characteristic value rewriting process, when Problem emotion change from the emotion generating unit 13-1 (S-36), and expression characteristics It starts when the rule interpreter 13-28 requests a change in expression characteristics (S-30).

The expression characteristics are changed when the mode flag emotion expression permission fe is on and the user-specified flag of the expression characteristic threshold register Rt is off, or when the user-specified flag is on, • Only when the flag fp is on, the expression characteristic threshold register Rt
Is rewritten (S-33).

When the expression characteristic threshold value is rewritten as described above and when a notification of an emotion change is received, the maximum value of the artificial emotion value register Re is detected (S-3).
5). The obtained maximum value and the expression characteristic threshold register Rt
Are compared with the strong / weak threshold values (S-37), and the expression value of the i-th emotion having the maximum value is determined as weak-strength (S-38) or strong-strength (S-39). Outgoing value register R
i, and immediately after the scene composition unit 13-2
2 is sent.

According to the notification from the task execution unit and the emotion generation unit
That processes 11 and 12 is a flowchart for describing an emotion imparting process to the task execution scene. When a task execution event occurs in the task execution unit, a notification signal to that effect arrives at the expression value determination unit. In order to determine the image display and the speech synthesis accompanying the task execution, emotion information is required. Therefore, the emotion expression value determination process in which the rewriting of the characteristic threshold is omitted in the above-described process is started.

Waiting for task execution occurrence signal reception (S-41)
Detects that task execution has occurred, the mode flag fe
Is checked (S-42). When fe is off, that is, when emotion expression is not performed, only the task number is transmitted to the scene configuration unit 13-22 without performing the subsequent processes. When fe is on 0, the maximum value is detected in the artificial emotion value register (S-43).

The obtained maximum value is compared with the strong-weak threshold value in the expression characteristic threshold value register Rt (S-44).
If the expression of the i-th emotion having the maximum value is less than or equal to the low-intensity, only the task number is transmitted to the scene composing unit. Otherwise, the low-intensity (S-45) or the high-intensity is used. (S-46) is determined and stored in the expression value register Ri.

Immediately, the contents of the expression value register Ri and the task number are output to the scene composing section (S-7).
And S-8). 3. Expression control designation by user's expression characteristic editor The expression characteristic editor 13-27 in the block diagram of the emotion expression controller of FIG. Is an editing device for forcibly controlling the editing.

The expression characteristic editor 13-27 includes an expression characteristic threshold setting changer 13-271, a display mode / flag setting changer 13-272, an expression characteristic rule editor 13-283,
It is composed of four sub-modules of an expression characteristic threshold pattern editor 13-274. Expression characteristic threshold setting changer 13
The function -271 fixes the value on the expression characteristic threshold register to a desired value. The fixed value is identified by a user-specified flag assigned to the expression characteristic threshold register.

Display mode / flag setting changer 13-27
The second function is for designating a display mode flag which is control information of the expression value determination unit 13-21. This flag specifies permission / non-permission of the expression of the fine expression and adaptation of the expression characteristic depending on the situation. Expression characteristic rule editor 13-
The function of H.273 is the expression characteristic rule database 13-29.
An editor for editing the rules inside. These rules are for the system to automatically rewrite the expression characteristic value.

The function of the expression characteristic threshold pattern editor 13-274 is an editor for editing and registering a group of patterns in the expression characteristic threshold pattern database 13-30. By editing the expression characteristic threshold pattern, the user can set a desired expression characteristic. 4. Process on Scene Constructing Unit When the expression value determining unit 13-21 determines the type and intensity of the element emotion to be expressed, the scene composing unit is required to have an output configuration.

FIG. 13 is a flowchart for explaining the process on the scene composing section. Each function of the expression value determination unit 13-21 in FIG. a. Task Execution Request When a task execution scene is requested, a scene with a specified emotion and intensity is searched from the scene database 15. b. Emotion change request When an emotion change expression is requested, it is first checked whether the task execution scene configuration (S-51) is busy (S-52), and the scene relating to the current display state is changed to a new emotion mode. Is retrieved from the scene database 15 and configured (S-53). The scene configured here is composed of a sequence of an image code and an utterance data code, and is immediately sent to the voice synthesis / image display unit 14.

[0069]

As described above, since the emotion expression device according to the present invention can set a threshold value for expression for each emotion component, the user can change the emotion expression style of the device to his / her preference. It can be adjusted freely to suit. Also,
In the emotion expression device according to the present invention, the speaker, the amount of past dialogue,
Recognizing the expression control information such as the situation, and changing the expression threshold in association with those recognition results, the speaker,
The emotion display style can be spontaneously and appropriately switched according to differences in intimacy and situations.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a basic configuration of an emotion expression device according to the present invention.

FIG. 2 is a block diagram illustrating a configuration of an embodiment of an emotion expression device according to the present invention.

FIG. 3 is a block diagram illustrating a configuration of an emotion management unit in the embodiment shown in FIG.

4 is a block diagram illustrating a configuration of an emotion expression control unit in the emotion management unit illustrated in FIG.

FIG. 5 is an explanatory diagram of a mode flag in the emotion expression control section shown in FIG.

FIG. 6 is an explanatory diagram of a register group in the emotion expression control section shown in FIG. 4;

FIG. 7 is a flowchart illustrating a process on the expression characteristic rule interpreter.

FIG. 8 is a flowchart illustrating an outline of a process on an expression value determination unit.

FIG. 9 is a partial flow chart for explaining an expression characteristic value rewriting process.

FIG. 10 is a partial flowchart following FIG. 9 for explaining an expression characteristic value rewriting process.

FIG. 11 is a partial flowchart illustrating a process of giving an emotion to a task execution scene.

FIG. 12 is a partial flowchart following FIG. 11 for explaining a process of giving an emotion to a task execution scene.

FIG. 13 is a flowchart illustrating a process on a scene composition unit.

FIG. 14 is a basic configuration diagram illustrating an example of a conventional emotion expression device.

[Explanation of symbols]

1 ... input, 2 ... emotional information recognition means, 3
··· Emotional arousal means, 4 ··· Expression / voice synthesis means, 5 ··· Output, 6 ··· Emotion expression control unit, 7
... User, 6-1 ... Expression style setting means,
6-2 ... Expression content determination means, 6-21 ... Expression style storage means, 6-3 ... Expression style group holding means, 6-4 ... Expression control Information recognition means, 6-4
1 ... speaker identification means, 6-42 ... dialogue amount recording means, 6-43 ... situation recognition means.

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-60-144799 (JP, A) JP-A-2-183371 (JP, A) JP-A-4-199098 (JP, A) JP-A-5-1990 100692 (JP, A) JP-A-5-216618 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G10L 3/00 G06F 3/16 G06T 13/00

Claims (2)

(57) [Claims] 1. An emotion-evoking information recognizing means for recognizing predetermined emotion-evoking information, comprising: “surprise, fear, disgust, anger,
The emotional state is expressed by giving each emotional component such as “happiness and sadness” a quantity reflecting the emotional intensity, and when the emotional arousal information is recognized, the emotional state transition rules defined in advance are used immediately. And an emotion expression device comprising: an emotion evoking means for causing the emotion state to transit; and an expression / speech synthesis means for changing the expression or voice prosody in accordance with the emotion state defined by the emotion evoking means and outputting the result. An expression style setting means for setting or changing an emotion expression style of the device by designating each expression threshold value for each emotion component; and holding an expression threshold value set by the expression style setting means. An expression style storage means is provided, and the expression threshold and the emotional state defined by the emotion eliciting means are compared with each other to determine an expression or voice to be expressed. A content determining means out, the exposed a set exposed threshold stores plural in style setting means, emotional expression system characterized by comprising the expression style unit holding means can be stored. 2. The method according to claim 1, wherein a speaker identification unit for identifying and identifying a conversation partner, and the number of past conversations or the total conversation time with the speaker identified by the speaker identification unit are measured and recorded. Dialogue amount storage means, a situation recognizing means for specifying the current situation from a plurality of situations categorized in advance, and pre-stored in the expression style group holding means based on the difference in the conversation partner, the conversation amount, and the situation. A specific expression threshold is read out from the set of expression thresholds that have been set, and is set in the expression style storage means, or a specific conversion expression is applied from some predetermined conversion expressions. And an expression style changing means for directly rewriting the expression threshold stored in the expression style storage means.