ISCA & IEEE Workshop on Spontaneous Speech Processing and Recognition April 13-16, 2003 Tokyo Institute of Technology, Tokyo, Japan

Using a Non-Spontaneous Speech Synthesiser as a Driver for a Spontaneous Speech Synthesiser

Nick Campbell, Parham Mokhiari

JST/CRFST Expressive Speech processing Project, ATR Human Information Science Laboratories, Kyoto, Japan

This paper describes a method by which the synthesis of spontaneous-sounding speech can be generated using a current speech synthesizer as a driver for unit selection from a spontaneous speech database. The driver is used to generate acoustic targets, which are then used for pre-selection of synthesis units, with the final candidates selected according to further acoustic constraints to ensure the selection of units having appropriate voice and speaking style characteristics.

Full Paper

Bibliographic reference.  Campbell, Nick / Mokhiari, Parham (2003): "Using a non-spontaneous speech synthesiser as a driver for a spontaneous speech synthesiser", in SSPR-2003, paper TAP19.