Dimensional Emotion Representation as a Basis for Speech Synthesis with Non-extreme Emotions

Past attempts to model emotions for speech synthesis have focused on extreme, “basic” emotion categories. The present paper suggests an alternative representation of emotional states, by means of emotion dimensions, and explains how this approach can contribute to making speech synthesis a useful component of affective dialogue systems.

• Type: Conference
• Year: 2004
• Where: ADS
• Authors: Marc Schröder
• Keywords: ADS 2004, Computer Graphics, Emotion Categories, Emotion Dimensions, Speech Synthesis