Speech Prosody 2004

Nara, Japan
March 23-26, 2004

Accounting for Voice-Quality Variation

Nick Campbell

ATR Human Information Science Laboratories, Keihanna Science City, Kyoto, Japan

This paper proposes a two-layer model of the information carried in the speech signal. It attempts to define the role of prosody with a wider scope than has previously been considered in speech synthesis or linguistic research, by taking into account affective information in addition to that of linguistic content. The work is based on analysis of a large corpus of spontaneous conversational speech, in which we found that voice quality was consistently varied according to paralinguistic factors. We argue that research in language evolution and cognitive neurology suport our interpretation that tone-of-voice should be considered as a distinct prosodic feature, which is deliberately controlled to express interpersonal relationships as an integral part of a spoken utterance.

Full Paper

Bibliographic reference.  Campbell, Nick (2004): "Accounting for voice-quality variation", In SP-2004, 217-220.