This paper describes an approach that allows text-to-speech synthesisers to be produced for new languages for use with assistive applications. The approach uses a simple rule-based text-to-phoneme stage. The phonemes are passed to an existing phoneme-to-speech system for another language. We show that the match between the language to be synthesised and the language on which the phoneme-to-speech system is important for the perceived quality of the speech but not necessarily the understandability of speech.
展开▼