14th International Congress of Phonetic Sciences (ICPhS-14) San Francisco, CA, USA August 1-7, 1999

Language-Universal and Language-Specific Components in the Multi-Language ETI-Eloquence Text-to-Speech System

Susan R. Hertz (1,2), Rebecca J. Younes (1), Nina Zinovieva (1)

(1) Eloquent Technology, Inc., Ithaca, NY, USA
(2) Department of Linguistics, Cornell University, Ithaca, NY, USA

ETI-Eloquence is a multi-voice, multi-language, rule-based text-to-speech system based on the innovative Delta synthesis technology. In the last three years, the system has been extended from General American English to nine new languages and dialects, including British English, Mexican and Castilian Spanish, Canadian and Parisian French, Mandarin Chinese, German, Italian, and Brazilian Portuguese. One of the goals of this work has been to develop a universal strategy for text-tospeech rule development and as large a common rule base as possible. This paper describes the result of this work, providing an overview of the language-universal and language-specific components underlying our current rule system.

Full Paper

Bibliographic reference.  Hertz, Susan R. / Younes, Rebecca J. / Zinovieva, Nina (1999): "Language-universal and language-specific components in the multi-language ETI-eloquence text-to-speech system", In ICPhS-14, 2283-2286.