音声に含まれる感情を考慮した自然言語対話システム

URI http://harp.lib.hiroshima-u.ac.jp/hiroshima-cu/metadata/12255
ファイル
タイトル
音声に含まれる感情を考慮した自然言語対話システム
別タイトル
Natural Language Dialogue System considering Emotion Guessed from Acoustic Features
著者
氏名 TANG Ba Nhat
ヨミ
別名
氏名 目良 和也
ヨミ メラ カズヤ
別名 MARA Kazuya
氏名 黒澤 義明
ヨミ クロサワ ヨシアキ
別名 KUROSAWA Yoshiaki
氏名 竹澤 寿幸
ヨミ タケザワ トシユキ
別名 TAKEZAWA Toshiyuki
抄録

With the previous IVR (Interactive Voice Response) systems, we can tell the computer to do our simply tasks only by giving a short voice command. But in human interaction, we do not rely on only the content of utterance, like the person who said "I'm fine" with sad voice maybe not really fine, for instance. In this paper, we propose an intelligent IVR system which uses not only the content but also the acoustic features of the utterance to make better response. Firstly, the method estimates user’s emotion (Acceptance, Anger, Hope, Disgust, Fear, Pleasure, Sad, Surprise and Neutral) by SVM using acoustic features extracted from user’s voice. Then the method applies AIML-based matching rules to the user’s utterance based on acoustic and linguistic features. As a result of experiment, the proposed system was preferred, was felt as more flexible and got closer to human-like communication.

掲載雑誌名
HAIシンポジウム2014 = Human-Agent Interaction Symposium 2014
開始ページ
87
終了ページ
91
出版年月日
2014-12
出版者
HAIシンポジウム2014運営委員会
本文言語
日本語
資料タイプ
会議発表論文
著者版フラグ
出版社版
権利情報
Copyright (c) by Author
区分
hiroshima-cu