This paper focuses on a partial emphasis process in sentences known as "prominence". Prominence is considered important in achieving rule-based synthetic speech that is easily understood. In general, prominence in Japanese sentences may be divided into two types, default and intended. Intended prominence can be further classified depending on where focus is placed. Features of Japanese sentences which include both types of prominence are analyzed in terms of prosody. The analysis results show that the most dominant form of prominence is enhancement of fundamental frequency and increasing of power. Specifically, this analysis concerning F0 contours confirms that a syntactic unit or a part of it, as an object of placed focus, is mapped into one of several categories of prosodic words and variation types. Finally, listening test results are briefly presented to show the effectiveness of the rules proposed based on the above analysis results.
Cite as: Takeda, S., Ichikawa, A. (1990) Analysis of prosodic features of prominence in spoken Japanese sentences. Proc. First International Conference on Spoken Language Processing (ICSLP 1990), 493-496, doi: 10.21437/ICSLP.1990-147
@inproceedings{takeda90b_icslp,
author={Shoichi Takeda and Akira Ichikawa},
title={{Analysis of prosodic features of prominence in spoken Japanese sentences}},
year=1990,
booktitle={Proc. First International Conference on Spoken Language Processing (ICSLP 1990)},
pages={493--496},
doi={10.21437/ICSLP.1990-147}
}