TRAINING INTONATIONAL PHRASING RULES AUTOMATICALLY FOR ENGLISH AND SPANISH TEXT-TO-SPEECH

We describe a procedure for acquiring intonational phrasing rules for text-to-speech synthesis automatically, from annotated text, and some evaluation of this procedure for English and Spanish. The procedure em ploys decision trees generated automatically, using Classification and Regression Toe techniques, from text corpora which have been hand-lab eled by native speakers with likely locations of intonational boundari es, in conjunction with information available about the text via simpl e text analysis techniques. Rules generated by this method have been i mplemented in the English version of the Bell Laboratories Text-to-Spe ech System and have been developed for the Mexican Spanish version of that system. These rules currently achieve better than 95% accuracy fo r English and better than 94% for Spanish.