Third ESCA/COCOSDA Workshop on Speech Synthesis

November 26-29, 1998
Jenolan Caves House, Blue Mountains, NSW, Australia

Where Is The Information In Speech? (and to What Extent can it be Modelled In Synthesis?)

Nick Campbell

ATR Interpreting Telecommunications Research Labs. Seika-cho, Soraku-gun, Kyoto, Japan

Di erent kinds of speech information are meaningfully used in human communication. This paper attempts to show how they can be modelled in speech synthesis and suggests that many conventional synthesis methods may fail to take into account the subtleties of human speech variation. It argues that modelling of voice quality should be the next main goal for speech synthesis technology, and proposes that evaluations of synthesis technology should aim to include a Turing component, which measures the ability of each system to perform on a range of human-speech features.


Full Paper

Bibliographic reference.  Campbell, Nick (1998): "Where Is The Information In Speech? (and to What Extent can it be Modelled In Synthesis?)", In SSW3-1998, 17-20.