Referring to German, this paper deals with (1) automatic symbolic stress assignment at the word level (simple and compound, including automatic compound decomposition), and at the sentence level on the basis of syntactic analysis, (2) pitch patterns based on the sentence stress assignment and further syntactic and semantic categories, (3) durations depending on segment type, context and stress. The implementation of these prosodic model components in the INFOVOX TTS are discussed and improvements of its output demonstrated with reference to rule changes under the three section headings.
Cite as: Kohler, K.J. (1990) Improving the prosody in German text-to-speech output. Proc. First ESCA Workshop on Speech Synthesis (SSW 1), 189-192
@inproceedings{kohler90_ssw,
author={Klaus J. Kohler},
title={{Improving the prosody in German text-to-speech output}},
year=1990,
booktitle={Proc. First ESCA Workshop on Speech Synthesis (SSW 1)},
pages={189--192}
}