A study on an emotion model for prosody control and dialog management in speech communication
| Project/Area Number |
17500079
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Media informatics/Database
|
|---|
| Research Institution | Tokyo University of Technology |
|---|
Principal Investigator |
OHNO Sumio Tokyo University of Technology, School of Computer Science, Associate Professor, コンピュータサイエンス学部, 助教授 (80256677)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
KAMEDA Hiroyuki Tokyo University of Technology, School of Computer Science, Associate Professor, コンピュータサイエンス学部, 助教授 (00194994)
|
|---|
| Project Period (FY) |
2005 – 2006
|
| Project Status |
Completed (Fiscal Year 2006)
|
| Budget Amount *help |
¥3,100,000 (Direct Cost: ¥3,100,000)
Fiscal Year 2006: ¥1,400,000 (Direct Cost: ¥1,400,000)
Fiscal Year 2005: ¥1,700,000 (Direct Cost: ¥1,700,000)
|
|---|
| Keywords | spoken utterance / emotion / spoken language model / prosody / mind model / 音声情報処理 / 感情表現 / 基本周波数パターン |
|---|
| Research Abstract |
In this scientific research project, it was achieved that a conversational system can process the emotion, and the object was 1) concerning the spoken expression of the emotional utterances and 2) concerning the modeling emotions in a dialog management. The following results were obtained in this project.
1) Emotional expression of synthetic speech
Not only the kind of emotions but also the levels of their strength were our research targets. Four emotions, i.e., "Joy", "Sadness", "Anger", and "Fear" were taken up as a kind of feelings, and prosodic features of the emotional utterances were examined by collecting and analyzing it for synthesizing expressive speech with controlling the level of the emotion. The controlling rule war formulated on the basis of a linear modeling of obtained prosodic parameters. The evaluation experiment was conducted by a synthetic speech. As a result, emotional strength control can be realized with a limitation of the kinds of emotions. A relatively good performance was obtained for "sadness" and "fear" in case of controlling the pitch features.
2) Modeling of emotions in dialog management
When the processing of the expression of emotion is taken to the conversation between human and machine, advanced communications function is achieved by guessing other's emotions. It proposed an engineering model of the mind. The knowledge representation in the model of the mind was defined from the viewpoint of "Own inner environment", "Own inner environment from others" and "Own external environment from others", and emotions located it with one parameter of an inner environment. Moreover, the guess method by the inference-by-synthesis technique was designed as a processing technique for the others recognition in the model of the mind.
|
Report
(3 results)
Research Products
(26 results)
