Fundamental frequency rules for the synthesis of simple declarative English sentences

Publisher Website
Google Scholar
Abstract
A recent experiment by Olive and Nakatani [J. P. Olive and L. H. Nakatani, ’’Rule−synthesis of speech by word concatenation: a first step,’’ J. Acoust. Soc. Am. 55, 660−666 (1974)] has shown that when telephone numbers were rule−synthesized by splicing together numbers which were spoken in isolation, the proper adjustment of fundamental frequency was more important than adjustments of amplitude, duration, and between word coarticulation, for making these spliced numbers sound as if they were spoken naturally. This experiment also showed that in the case of telephone numbers, fundamental frequency adjustments were sufficient for providing synthesis by rule of quality comparable to the quality of natural speech which was processed by an analysis−synthesis procedure. This paper describes the next step in our research of rule synthesis: the study of a restricted set of sentences of simple syntactic structure, instead of telephone numbers. Methods for studying fundamental frequency contours of naturally spoken sentences are discussed; some fundamental frequency rules for synthesizing short English sentences are derived from these natural utterances. More specifically, only rules which are dictated by the syntax were investigated; the appropriateness of these rules was determined by incorporating them into a computerized rule−synthesis scheme. Subject Classification: 70.50.

This publication has 0 references indexed in Scilit: