Exploration of affect sensing from speech and metaphorical text

Abstract

We report new developments on affect detection from textual metaphorical affective expression and affect sensing from speech. The textual affect detection component has been embedded in an intelligent conversational AI agent interacting with human users under loose scenarios. The detected affective states from text also play an important role in producing emotional animation for users' avatars. Evaluation of the affect detection from speech and text is provided. Our work contributes to the conference themes on engagement and emotion, virtual AI agents, narrative storytelling in education and evaluation of affective social interaction. © 2009 Springer Berlin Heidelberg.