Julia Hirschberg, TMH/KTH & Columbia University, New York
Human Communication seminar series OBS! lokal: D2, Lindstedtsvägen 5
A speaker\'s emotional state is often conveyed by acoustic and prosodic factors, as well as the words they choose and the gestures they use. Much research has been done in recent years to detect emotional state in Spoken Dialogue Systems, primarily from acoustic and prosodic cues, so that angry or frustrated users can be directed to a human agent. Other work has sought to identify a wider variety of emotions and intentions in recorded meetings, again from acoustic and prosodic cues. From the perspective of speech generation, the problem of conveying emotional state is a critical barrier to making automated systems sound more like real human beings.
Computer game designers cite the limits of prosodic and emotional \'naturalness\' of speech synthesizers as a major obstacle to their use in games. In this talk she will describe ongoing research on the detection and production of emotional state in spoken language and a variety of applications of this research, including call centres, automated tutoring systems, speaking style assessment, deception detection, and medical diagnosis.