This paper describes an algorithm for rule-based prosody in Norwegian. Authentic prosodic patterns are essential in establishing natural sounding synthetic speech. Norwegian is predominantly an intonation language. There is a limited use of tones though. Most Norwegian dialects have two distinguishing tones for words. Many of those words are homonyms. Revealing and realizing of the correct tone and pronunciation among homonyms are especially addressed. The algorithm for rule-based prosody is based on a word lexicon, a grammatical and prosodic parser, and authentic prosodic patterns. The prosodic model is implemented as a preprocessor to a real-time synthesizer on a PC. Keywords: Speech synthesis, Intonation, Word tone, Prosody, Parsing
Cite as: Stensby, S. (1991) Prosody in a rule-based norwegian text-to-speech system. Proc. 2nd European Conference on Speech Communication and Technology (Eurospeech 1991), 1149-1152, doi: 10.21437/Eurospeech.1991-276
@inproceedings{stensby91_eurospeech,
author={Sverre Stensby},
title={{Prosody in a rule-based norwegian text-to-speech system}},
year=1991,
booktitle={Proc. 2nd European Conference on Speech Communication and Technology (Eurospeech 1991)},
pages={1149--1152},
doi={10.21437/Eurospeech.1991-276}
}