Speech Prosody 2006

Dresden, Germany
May 2-5, 2006

Employing Intonational Events Parameterization for Emotion Recognition

Panagiotis Zervas, Iosif Mporas, Nikolaos Fakotakis

Electrical and Computer Engineering Dept., University of Patras, Greece

In this work we introduce the utilization of Fujisakiís modeling of pitch contour for the task of emotion recognition. For the evaluation of the proposed features we have employed a decision tree as well as an instance based learning algorithm. The datasets utilized for training the classification models, were extracted from two emotional speech databases. Results showed that knowledge extracted from Fujisakiís modeling of intonation benefited all resulted emotion recognition models. Thus, an average raise of 9,52% in the total accuracy of all approaches was achieved.

Full Paper

Bibliographic reference.  Zervas, Panagiotis / Mporas, Iosif / Fakotakis, Nikolaos (2006): "Employing intonational events parameterization for emotion recognition", In SP-2006, paper 192.