Paper: Learning Intonation Rules for Concept to Speech Generation

ACL ID P98-2165
Title Learning Intonation Rules for Concept to Speech Generation
Venue Annual Meeting of the Association of Computational Linguistics
Session Main Conference
Year 1998

In this paper, we report on an effort to pro- vide a general-purpose spoken language gener- ation tool for Concept-to-Speech (CTS) appli- cations by extending a widely used text gener- ation package, FUF/SURGE, with an intona- tion generation component. As a first step, we applied machine learning and statistical models to learn intonation rules based on the semantic and syntactic information typically represented in FUF/SURGE at the sentence level. The re- sults of this study are a set of intonation rules learned automatically which can be directly im- plemented in our intonation generation compo- nent. Through 5-fold cross-validation, we show that the learned rules achieve around 90% accu- racy for break index, boundary tone and phrase accent and 80% accuracy for pitch accent. Our study ...