The use of a generative model of F/sub 0/ contours for multilingual speech synthesis

Abstract

This paper describes the command-response model for F/sub O/ contour generation originally developed for common Japanese, and demonstrates its capability of generating F/sub O/ contours of various other languages with minor language-specific modifications. The model is especially useful in multilingual speech synthesis, since the same mechanism can be driven by language-specific patterns of input commands to produce F/sub O/ contours of utterances of the respective languages.