This paper propose a Bayesian-based method of emotion detection from talking voice. Development of a entertainment robot and joyful communication between human and robot have given us the motivation for a computational method for robot to detect its dialogist's emotion from his talking voice. The method is based on the Bayesian networks which represent the dependence and its strength between dialogist's utterance and his emotion, by using a Bayesian modeling for prosodic feature quantities extracted from emotionally expressive voice data. In this paper, we propose a biphase inference method using the Bayesian networks. This inference method has two steps: to reduce the choice of emotion at the first step and to infer a certain emotion reliably from little choice at the second step. The paper also reports some empirical reasoning performance of this method.
展开▼
