8th European Conference on Speech Communication and Technology

Geneva, Switzerland
September 1-4, 2003


Towards Synthesising Expressive Speech; Designing and Collecting Expressive Speech Data

Nick Campbell

ATR-HIS, Japan

Corpus-based speech synthesis needs representative corpora of human speech if it is to meet the needs of everyday spoken interaction. This paper describes methods for recording such corpora, and details some difficulties (with their solutions) found in the use of spontaneous speech data for synthesis.

Full Paper

Bibliographic reference.  Campbell, Nick (2003): "Towards synthesising expressive speech; designing and collecting expressive speech data", In EUROSPEECH-2003, 1637-1640.