SynPaFlex is an ANR project targeted at the improvement of Text-To-Speech synthesis engines through two main research axes:
- Pronunciation variants modelling and generation
- Context-adapted prosody modelling and generation
The main targeted applications are in the domains of entertainment (audiobook reading, video games), serious games (virtual environments), language learning (dictation, elocution style) or even for vocal aids designed for handicapped people.
This project is mainly supported by IRISA and involves members from LLF (Laboratoire de Linguistique Formelle) and from ATILF.