JP6201212B2 - Character generating apparatus and program - Google Patents

Description

  The present invention relates to a character generation device and a program for generating a character capable of expressing emotion.

  Conventionally, a technique for generating a facial expression of a character based on arbitrary input information such as a mouse click or text data is known. For example, in the technique disclosed in Patent Document 1, input text data is analyzed, emotion information is given for each segment, and a character corresponding to the given emotion is displayed with a facial expression. In other words, this technique is a facial expression generation method that generates a specific facial expression by combining a plurality of facial part images constituting a facial image, and is set for each facial part image corresponding to each facial expression. Are stored in association with an expression category space composed of predetermined determination criteria, and position information about each facial part image corresponding to various expressions in the expression category space is stored. Then, the operation data input from the outside is associated with each facial expression set in the facial expression category space.

  The technology disclosed in Patent Document 2 supports appropriate service / information selection through natural conversation with a user, and provides services / information useful to the user. That is, this technique can perform a dialogue with a user via an interface via a character. The dialogue input from the user is recognized by the dialogue recognition unit and stored as user profile information. A dialogue recognition result and user profile information are fed back, a dialogue sentence is generated based on the knowledge base, and the dialogue is output to the user.

JP 2002-342769 A JP 2006-323875 A

  However, with the technique disclosed in Patent Document 1, a facial expression of a character to which emotion information is given can be generated, but it cannot be generated until a gesture of a character. In addition, in the technique disclosed in Patent Document 2, the user profile information is updated using the recognition result of the dialog with the user, the feedback to the dialog control unit is performed, and reflected in the selection of the future dialog. However, the character to be generated has no emotion, and feedback to the emotion generation unit of the character is not considered.

  The present invention has been made in view of such circumstances, and a character generation device capable of feeding back an arbitrary input / output result with a user to the emotion generation unit of the character and reflecting it in future character generation. And to provide a program.

  (1) In order to achieve the above object, the present invention takes the following measures. That is, the character generation device of the present invention is a character generation device that generates a character capable of expressing emotion, and determines an action including the facial expression or gesture of the character based on input data, and indicates the action Based on information indicating emotion classification of the input data, an emotion imparting unit that provides information indicating emotion to information indicating the motion of the character, display of the motion or display of the emotion And an information reflecting unit that updates information indicating the action or information indicating the emotion based on feedback data corresponding to.

  As described above, based on the input data, the action including the facial expression or gesture of the character is determined, information indicating the action is output, and the information indicating the character classification of the input data is used as the information indicating the action of the character. The information indicating the motion or the information indicating the emotion is updated based on the feedback data corresponding to the motion display or the emotion display, so that the reaction of the user is reflected in future character generation. Can do. Further, not only the facial expression of the character to which the emotion is given, but also the gesture of the character to which the emotion is given can be generated.

  (2) The character generation device of the present invention is further characterized by further comprising a feedback input unit for inputting the feedback data.

  Thus, since the feedback input part which inputs feedback data is further provided, a user's reaction can be reflected in future character production | generation.

  (3) Further, the character generation device according to the present invention is configured to reproduce an action when reproducing the information indicating the action of the character to which the information indicating the emotion is given based on the information indicating the strength of the emotion classification of the input data. It further comprises a strength control section for controlling strength.

  As described above, the apparatus further includes a strength control unit that controls the strength of the motion when reproducing the information indicating the motion of the character to which the information indicating the emotion is given based on the information indicating the strength of the emotion classification of the input data. The character's expressive power can be improved.

  (4) Moreover, the character generation device of the present invention is further characterized by further comprising an emotion analysis unit for analyzing the emotion classification of arbitrary input data or the intensity of the emotion classification.

  In this way, since it is further provided with an emotion analysis unit that analyzes the emotion classification of arbitrary input data or the strength of the emotion classification, complex emotions that cannot be expressed in conversation, such as e-mail and human behavior, are continuously input. Even in such a case, it is possible to automatically generate a character provided with emotion.

  (5) In addition, the character generation device of the present invention further includes a data input unit for inputting arbitrary data.

  Thus, since the data input part which inputs arbitrary data is further provided, it becomes possible to produce | generate a character using arbitrary data.

  (6) In addition, the character generation device of the present invention further includes a character reproduction unit that reproduces information indicating the action of the character to which the information indicating the emotion is given.

  Thus, since the character reproducing part which reproduces | regenerates the information which shows the motion of the character to which the information which shows an emotion was provided is further provided, it becomes possible to make a user visually recognize the produced | generated character.

  (7) The program of the present invention is a program for generating a character capable of expressing emotions, and determines an action including the facial expression or gesture of the character based on input data, and includes information indicating the action. A process of outputting, a process of adding information indicating emotion to information indicating the motion of the character based on information indicating the emotion classification of the input data, and feedback data corresponding to the display of the motion or the display of the emotion And a process of updating the information indicating the action or the information indicating the emotion, and a series of processes to be executed by the computer.

  As described above, based on the input data, the action including the facial expression or gesture of the character is determined, information indicating the action is output, and the information indicating the character classification of the input data is used as the information indicating the action of the character. The information indicating the motion or the information indicating the emotion is updated based on the feedback data corresponding to the motion display or the emotion display, so that the reaction of the user is reflected in future character generation. Can do. Further, not only the facial expression of the character to which the emotion is given, but also the gesture of the character to which the emotion is given can be generated.

  According to the present invention, a user's reaction can be reflected in future character generation. Further, not only the facial expression of the character to which the emotion is given, but also the gesture of the character to which the emotion is given can be generated.

It is a figure which shows schematic structure of the character production | generation apparatus which concerns on this embodiment. It is a block diagram which shows the function of the character production | generation apparatus which concerns on embodiment. It is a flowchart which shows operation | movement of the character production | generation apparatus which concerns on embodiment.

  The character generation device according to this embodiment feeds back an arbitrary input / output result with the user to the emotion generation unit of the character and reflects it in future character generation. In this specification, “expression” refers to a face when a creature including a person or a character that simulates a creature takes action or expresses emotion, and “gesture” refers to a movement of the body. It means to do things.

  A character generation device according to an embodiment of the present invention determines a facial expression or gesture of a character based on input arbitrary text data, and determines the facial expression of the character according to information indicating emotion classification in the input text data. Or while giving an emotion to a gesture, based on the feedback data with respect to action determination or emotion provision, action determination or emotion provision is updated. Thereby, a user's reaction can be reflected in future character generation. Further, not only the facial expression of the character to which the emotion is given, but also the gesture of the character to which the emotion is given can be generated.

  In this specification, a character generation device using text is shown as an example, but the technical idea of the present invention is not limited to text information. For example, text data or feedback data may be acquired by speech recognition using “speech” as input information. In addition, “acoustic” may be input information, and emotion classification and intensity may be extracted based on fluctuations in average tempo, tone color, sound pressure level, and sound pressure level. Further, the emotion classification and its strength may be input by key operation using “key operation” as input information.

  In addition, “touch” may be used as input information, and emotion classification and intensity may be input by a touch operation. Further, the emotion classification and its intensity may be extracted using “image” as input information using facial expression recognition technology in the image. In this case, an emotion recognition camera can be used. In addition, with “video” as input information, the examples of “voice” and “image” may be used together to extract emotion classification and its intensity. Thereby, even when complicated emotions that cannot be expressed in conversation, such as e-mails and human actions, are continuously input, it is possible to automatically generate a character with emotions.

  FIG. 1 is a diagram illustrating a schematic configuration of a character generation device according to the present embodiment. As shown in FIG. 1, the character generation device includes a display 10, a keyboard 20, and a PC (Personal Computer) 30. Then, text data 40 and feedback data 50 are input to the PC 30 by the keyboard 20. In FIG. 1, the keyboard 20 is connected to the PC 30, but the technical idea of the present invention is not limited to this, and it is sufficient if text data is input to the PC by any means. It is.

  The PC 30 is connected to a display 10 as a character reproduction device via a cable 30a. The PC 30 analyzes the emotion classification or the strength of the emotion classification in the input text data 40 and determines the facial expression or gesture of the character based on the text data 40. In addition, an emotion is given to the facial expression or gesture of the character determined based on the analyzed and classified emotion.

  Furthermore, the strength of the character gesture generated based on the analyzed emotion classification strength is controlled. Further, the PC 30 generates an image of the character's facial expression or gesture that is generated based on the analyzed emotion classification, and a character's facial expression or gesture that is controlled in strength based on the analyzed emotion classification intensity. A signal is transmitted to the display 10 as needed. At that time, the image generated in the PC 30 is projected on the display 10 as A1. In the character generating device according to the embodiment of the present invention, the synthesized facial expression or gesture image signal may not be transmitted to the display 10. Then, the PC 30 updates the action determination or emotion provision based on the feedback data 50.

  FIG. 2 is a block diagram illustrating functions of the character generation device according to the embodiment. The emotion analysis unit 30-1 of the PC 30 receives an input operation of the user and analyzes the emotion classification or the strength of the emotion classification in each data from the input arbitrary data. That is, the emotion analysis unit 30-1 recognizes the text data 40 input from the keyboard 20 and analyzes the emotion classification or the emotion classification strength in the text data 40. Further, the action determining unit 30-2 of the PC 30 determines the action of the character. That is, the action determination unit 30-2 recognizes the text data 40 input from the keyboard 20, and determines the facial expression or gesture of the character based on the text data 40.

  The emotion imparting unit 30-3 of the PC 30 imparts emotion to the character's facial expression or gesture. That is, the emotion imparting unit 30-3 converts the neutral facial expression or gesture based on the emotion classification analyzed by the emotion analyzing unit 30-1, and imparts the emotion to the character facial expression or gesture. Further, the strength control unit 30-4 of the PC 30 controls the strength of the character's facial expression or gesture. In other words, the strength control unit 30-4 controls the strength of the facial expression or gesture of the character to which the emotion giving unit 30-3 has given the emotion based on the strength of the emotion classification analyzed by the emotion analyzing unit 30-1.

  The character reproduction unit 30-5 of the PC 30 reproduces the facial expression or gesture image of the character. That is, the character reproduction unit 30-5 reproduces the facial expression or gesture of the character to which the emotion imparting unit 30-3 imparts the emotion, and the facial expression or gesture image of the character whose strength is controlled by the strength control unit 30-4.

  The information reflecting unit 30-6 of the PC 30 reflects the result of the feedback data in future character generation. That is, the information reflection unit 30-6 recognizes the feedback data 50 input from the keyboard 20, and feeds back the result to the action determination unit 30-2 or the emotion imparting unit 30-3.

  FIG. 3 is a flowchart illustrating the operation of the character generation device according to the embodiment. First, text data 40 is input to the PC 30 using the keyboard 20 (step S1). Next, the PC 30 recognizes the input text data 40 and analyzes the emotion classification and strength in the text data 40 (step S2). Here, a case where the input text data 40 is a word will be described. However, the technical idea of the present invention is not limited to a word, and may be a phrase or a sentence. In this embodiment, an emotion word database composed of emotion classifications and respective intensities is used.

The emotion word database is the emotion classification "joy", "sadness", "anger", "disgust", "fear" used for human expression formation for all words in a large number of word databases such as Japanese dictionary Specify the ratio of “Sin”, “Shame”, “Interest”, and “Surprise”, and specify the emotional intensity in the range of 0 to 1 in 10 steps in increments of 0.1. Is formed. In addition, when the input text data 40 is a phrase or sentence, as in the case of a word, an emotion phrase database or emotion sentence database composed of the emotion classification and the strength of each is used for the phrase or sentence. It ’s fine. Here, the intensity of “joy” is S ₁ , the intensity of “sadness” is S ₂ , the intensity of “anger” is S ₃ , the intensity of “hate” is S ₄ , the intensity of “fear” is S ₅ , “sin” Is expressed as S ₆ , “Shame” as S ₇ , “Interest” as S ₈ , and “Surprise” as S ₉ . However, S _i satisfies the following equation.

そして、入力されるテキストデータ４０と一致または類似する単語を感情語データベースの中から検索し、その単語における感情分類と強度を抽出する。すなわち、テキストデータ４０における感情分類は、次の数式に表わされるように、９次元のベクトルで示される。

Then, a word that matches or is similar to the input text data 40 is searched from the emotion word database, and the emotion classification and strength in the word are extracted. That is, the emotion classification in the text data 40 is represented by a 9-dimensional vector as represented by the following mathematical formula.

また、入力されるテキストデータ４０における全体強度Ｗは、次の数式で表わされる。

Further, the overall intensity W in the input text data 40 is expressed by the following equation.

例えば、テキストデータ４０として、「泣き笑い」を入力すると、抽出される感情分類と強度Ｗは、次のようになる。

For example, when “crying laughter” is input as the text data 40, the emotion classification and intensity W to be extracted are as follows.

次に、ＰＣ３０において、入力されるテキストデータ４０を認識し、テキストデータ４０に基づいて、「目を閉じる」「お辞儀する」「手を挙げる」等のキャラクタの表情または仕草を決定する。（ステップＳ３）。本実施形態では、無感情の表情または仕草における動きデータと、日本語辞書などの代表的な単語データベースに存在する各単語をパラメータとして算出した各動きデータにおける類似度パラメータのペアで構成される動作データベースを用意する。

Next, the PC 30 recognizes the input text data 40, and based on the text data 40, determines the facial expression or gesture of the character such as “close eyes”, “bow”, “raise hands”. (Step S3). In the present embodiment, the motion is composed of a pair of motion data in an emotionless expression or gesture and a similarity parameter in each motion data calculated using each word existing in a typical word database such as a Japanese dictionary as a parameter. Prepare a database.

  Note that the motion data described here describes in which spatial position coordinates all skeletons of the character exist at an arbitrary time. When the text data 40 is input, the similarity parameter in the text data 40 is calculated, the cosine similarity with the similarity parameter in all motion data in the motion database is obtained, and the facial expression or gesture having the largest value is obtained. While determining as the character's motion h, the character motion data in the motion database motion h is generated as motion data in the character generation device.

  Next, in the PC 30, an emotion is given to the facial expression or gesture of the character determined in step S3. (Step S4).

  First, a method for imparting emotion to the facial expression of the character determined in step S3 based on the emotion analyzed in step S2 will be described. In the present embodiment, the motion h determined as the character's facial expression in step S3 is synthesized and converted according to the emotion analyzed in step S2, thereby giving the emotion to the character's facial expression. First of all, the emotional expressions used in the human emotional expression “joy”, “sadness”, “anger”, “disgust”, “fear”, “sin”, “shame”, “interest” and “surprise”. The amount of change from is defined for each polygon in the character's expression.

Here, for a certain polygon P in the character's facial expression, the difference in coordinates from the emotional expression in the basic expression of “joy” is (X ₁ , Y ₁ , Z ₁ ), and the basic expression of “sadness” (X ₂ , Y ₂ , Z ₂ ), the difference in coordinates from the emotional expression in the basic expression of “anger” (X ₃ , Y ₃ , Z ₃ ), The coordinate difference from the emotional expression in the basic expression of “disgust” is (X ₄ , Y ₄ , Z ₄ ), and the coordinate difference from the emotional expression in the basic expression of “fear” is (X ₅ , Y ₅ , Z ₅ ), the difference in coordinates with the emotional expression in the basic expression of “Sin” (X ₆ , Y ₆ , Z ₆ ), the difference in coordinates with the emotional expression in the basic expression of “shame” the _{(X 7, Y 7, Z} 7), the coordinate difference between the expression of apathy in the basic facial expressions of "interest" _{(X 8, Y 8, Z} 8), "surprise _{(X 9, Y 9, Z} 9) the difference between the coordinates of the facial expression of apathy in the basic facial expression, and it represents.

Then, the motion h determined as the facial expression of the character in step S3 is synthesized and converted according to the emotion analyzed in step S2, and the emotion is added to the facial expression of the character. That is, if the coordinates of P in a frame with motion h are (X ₀ , Y ₀ , Z ₀ ), the coordinates of P (X _E , Y _E , Z _E ) in the facial expression of the character to which emotion is given are It is expressed by a formula.

同様の操作をキャラクタの表情における全ポリゴンに対して実行し、さらに動作ｈにおける全フレームに対して実行し、キャラクタの表情に感情を付与する。

A similar operation is performed on all the polygons in the character's facial expression, and is further performed on all the frames in the motion h to give emotion to the character's facial expression.

  Next, a method for imparting emotion to the character's gesture determined in step S3 based on the emotion analyzed in step S2 will be described. In this embodiment, the emotionless motion data and motion data of each emotion classification in all pre-recorded gestures are learned, the conversion from the emotionless motion data to the motion data of each emotion classification is defined in advance, Emotion is given to the character's gesture by synthesizing and converting the emotionless motion data in motion h determined in S3 according to the emotion analyzed in step S2.

In this specification, as an example, emotion is imparted to the character's gesture using principal component analysis, but the technical idea of the present invention is not limited to principal component analysis, but nonlinear state space mapping or machine learning. Emotion may be imparted to the character's gesture using another method. First of all, there are several gestures for nine emotions: emotion, joy, sadness, anger, disgust, fear, sin, shame, interest, and surprise. Prepare and register in the learning database. The emotionless motion data and the motion data of each emotion classification in all registered gestures are learned, and parameters for converting the emotionless motion data into motion data of each emotion classification are calculated by a linear regression method. That is, using the differential value k ^j (m) of the j-th principal component coordinate of emotionless motion data in the recorded gesture m (m = 1, 2,...), Linear regression calculation is performed using the following formula. , Conversion parameters a ^j _i and b ^j _i are calculated.

ただし、ｑ^j _i（ｍ）は仕草ｍにおける各感情の動きデータの第ｊ主成分座標の微分値を示しており、ｉ＝１は感情分類が「喜び」の場合を、ｉ＝２は感情分類が「悲しみ」の場合を、ｉ＝３は感情分類が「怒り」の場合を、ｉ＝４は感情分類が「嫌悪」の場合を、ｉ＝５は感情分類が「恐怖」の場合を、ｉ＝６は感情分類が「罪」の場合を、ｉ＝７は感情分類が「恥」の場合を、ｉ＝８は感情分類が「興味」の場合を、ｉ＝９は感情分類が「驚き」の場合をそれぞれ表す。次に、テキストデータ４０が入力されると、テキストデータ４０に基づいて決定した動作ｈにおける無感情の動きデータを、テキストデータ４０から抽出された感情分類に対応する各変換パラメータａ^j _i、ｂ^j _iを用いて合成変換し、キャラクタの仕草に感情を付与する。すなわち、感情が付与されたキャラクタの動作ｈにおける動きデータの第ｊ主成分座標の微分値ｐ^j（ｈ）は、次の数式で表わされる。

However, q ^j _i (m) indicates the differential value of the j-th principal component coordinate of the motion data of each emotion in the gesture m, i = 1 indicates that the emotion classification is “joy”, and i = 2 indicates the emotion. When the classification is “sadness”, i = 3 is when the emotion classification is “anger”, i = 4 is when the emotion classification is “disgust”, and i = 5 is when the emotion classification is “fear” , I = 6 is when the emotion classification is “Sin”, i = 7 is when the emotion classification is “Shame”, i = 8 is when the emotion classification is “Interest”, and i = 9 is when the emotion classification is Each case represents “surprise”. Next, when the text data 40 is input, the conversion parameters a ^j _i and b corresponding to the emotion classification extracted from the text data 40 are converted from the emotionless motion data in the action h determined based on the text data 40. synthesizes and transforms using ^j _i to add emotion to the character's gesture. That is, the differential value p ^j (h) of the j-th principal component coordinate of the motion data in the motion h of the character to which the emotion is given is expressed by the following formula.

同様の操作をキャラクタの動作ｈにおける全フレームに対して実行し、キャラクタの仕草に感情を付与する。

A similar operation is performed on all frames in the character's motion h to add emotion to the character's gesture.

Next, the PC 30 controls the emotional strength of the assigned character's facial expression or gesture based on the analyzed emotion classification strength (step S5). In the present embodiment, the differential value p ^j of the j-th principal component coordinate of the motion data in the motion h of the character imparted with emotion in step S4 according to the overall strength W in the input text data 40 is expressed by the following equation: The differential value o ^j of the j-th principal component coordinate of the motion data in the new motion h of the character whose weight is controlled is calculated.

同様の操作をキャラクタの動作ｈにおける全フレームに対して実行し、強弱を制御したキャラクタの新たな表情または仕草を生成する。

A similar operation is performed on all frames in the character motion h to generate a new facial expression or gesture of the character whose strength is controlled.

  Next, the PC 30 generates an image of the expression or gesture of the character to which the emotion is given based on the analyzed emotion classification, and the expression or gesture of the character whose strength is controlled based on the intensity of the analyzed emotion classification. (Step S6). Here, it is determined whether or not the text data 40 input from the keyboard 20 is to be continued (step S6-2). When the text data 40 is to be continued, after generating an image of the character's facial expression or gesture, the process returns to step S2 to The process for the text data 40 is started. On the other hand, if the text data 40 does not continue in step S6-2, the process proceeds to step S7.

  Finally, the generated facial expression or gesture image is displayed on the display 10 (step S7).

  On the other hand, in the PC 30, the action determination or emotion provision is updated based on the feedback data for the action determination or emotion provision (step S8). According to the technical idea of the present invention, a user different from the user who input the text data may input the feedback data. In this case, the processing from step S1 to step S7 and the processing of step S8 are performed in parallel. Done. At this time, if the action determination process in step S2 and the information update process in step S8, or the emotion imparting process in step S3 and the information update process in step S8 are executed at the same time, the information update in step S8 is performed in any case. Processing is prioritized and executed.

  First, a method for updating step S3 based on feedback data for the operation determination will be described. In the present embodiment, the feedback data 50 input from the keyboard 20 is recognized, and the degree of satisfaction with respect to the action determination is acquired. Here, the case where the input feedback data 50 is an arbitrary phrase such as “Thank you for replying” will be described. However, the technical idea of the present invention is not limited to the phrase, It may be a word as long as the feedback data 50 is limited to the feedback data for the operation determination. That is, it is possible to acquire the feedback data 50 in a natural exchange between the user and the character such as a dialogue.

Here, the facial expression or gesture in the character image generated immediately before is defined as an action h. If the feedback data 50 is input, as in step S3, and calculates the similarity parameter in the feedback data 50 calculates the cosine similarity between the similarity parameter in operation h, if the value is greater than the threshold theta _h The feedback data 50 is determined as feedback data for the operation h. In this embodiment, a word representing satisfaction is extracted from a large number of word databases such as a Japanese dictionary, and classified into two categories of “satisfied” and “unsatisfied”, and prepared as a satisfaction word database.

  Then, a word that matches or is similar to the satisfaction word database is detected from the feedback data 50 determined as the feedback data for the action h. If the detected word is a word classified as “satisfied”, when text data that is determined to be identical or similar to the text data 40 is input in the following step S1, the cosine similarity degree is determined. Regardless of the value, step S3 is updated so that the motion h is determined as the facial expression or motion of the character. If the detected word is a word classified as “dissatisfied”, when text data that is judged to be identical or similar to the text data 40 is input in the following step S1, the cosine is the second. Step S3 is updated so as to determine the facial expression or action of the character having a high degree of similarity.

Next, a method of updating step S4 based on feedback data for giving emotion will be described. In the present embodiment, the feedback data 50 input from the keyboard 20 is recognized, and the satisfaction level for giving emotions is acquired. Here, the case where the input feedback data 50 is an arbitrary phrase such as “Please be more pleased!” Is explained. However, the technical idea of the present invention is not limited to the phrase but is a sentence. Alternatively, words may be used as long as the feedback data 50 is limited to the feedback data for giving emotions. When the feedback data 50 is input, the emotion classification and the intensity S ′ _i in the feedback data 50 are obtained as in step S2, and if S ′ _i is not 0, the feedback data 50 is determined as feedback data for giving emotions. . Then, a word that matches or is similar to the satisfaction word database is detected from the feedback data 50 determined to be feedback data for giving emotions.

If the detected word is a word classified as “satisfied”, text data that is determined to be identical or similar to the text data 40 in the next step S1 is input. Step S4 is updated so that S ′ _i is added to the emotion strength S _i to add the emotion of the character. If the detected word is a word classified as “dissatisfied”, text data that is determined to be identical or similar to the text data 40 in the next step S1 is input. Step S4 is updated so that the emotion intensity S _i is subtracted from S ′ _i to give the character's emotion. In both cases, in the emotion classification where S ′ _i = 0, the value of the emotion strength S _i in the text data 40 is maintained.

  As described above, according to the present embodiment, the emotion classification or the strength of emotion classification in the text data 40 input to the PC 30 is analyzed, and the character's gesture is determined based on the text data 40. In addition, an emotion is given to the character's gesture according to the analyzed emotion classification. Further, the control unit controls the strength of the character gesture generated based on the analyzed emotion classification strength. Then, the PC 30 generates an image of the character of the character to which the emotion is given or a gesture of the character of the character whose strength is controlled, and transmits an image signal to the display 10 as needed. At that time, the image generated in the PC 30 is projected on the display 10 as A1. In addition, the PC 30 updates the action determination or emotional assignment based on the feedback data 50. Thereby, in this embodiment, a user's reaction can be reflected in future character production | generation. Further, not only the facial expression of the character to which the emotion is given, but also the gesture of the character to which the emotion is given can be generated. Furthermore, even when complex emotions that cannot be expressed in conversation, such as e-mails, Twitter, and human actions, are continuously input, characters with emotions can be automatically generated.

10 Display 20 Keyboard 30 PC
30-1 emotion analysis unit 30-2 motion determination unit 30-3 emotion imparting unit 30-4 strength control unit 30-5 character reproduction unit 30-6 information reflection unit 30a cable 40 text data 50 feedback data

Claims (5)

A character generation device that generates a character capable of expressing emotions,
An input section for inputting data;
Based on the input data, to determine the operation including facial expressions or gestures of the character, and the operation determining unit that outputs information indicating the operation,
Based on the information indicating the input data emotion classification, and emotions imparting portion that imparts information indicating the emotion of the information indicating the operation of the character,
A feedback data determination unit that determines whether the input data is feedback data corresponding to the display of the action or the display of the emotion of the character generated immediately before ;
On the basis of the feedback data, and information reflecting unit for updating the information indicating the information or the feelings showing the operation, a character generator, characterized in that it comprises a.   And a strength control unit for controlling the strength of the motion when reproducing the information indicating the motion of the character to which the information indicating the emotion is given based on the information indicating the strength of the emotion classification of the input data. The character generation device according to claim 1. Input any data emotion classification or the character generating apparatus according to claim 1 or claim 2, wherein further comprising an emotion analysis unit for analyzing the intensity of the emotion classification. The character generation device according to any one of claims 1 to 3 , further comprising a character reproducing unit that reproduces information indicating a motion of the character to which the information indicating the emotion is given. A program that generates characters that can express emotions,
Based on the input data, a process of determining the operation including facial expressions or gestures of the character, and outputs information indicating the operation,
A process based on the information indicating the input data emotion classification, imparts information indicating the emotion of the information indicating the operation of the character,
A process for determining whether or not the input data is feedback data corresponding to the display of the action of the character or the display of the emotion generated immediately before ;
On the basis of the feedback data, the process of updating the information indicating the information or the feelings showing the operation, a series of processes, a program for causing a computer to execute.

Images