Conversor texto a voz en el dialecto venezolano por medio de la concatenación de difonos

Abstract

We give a detailed description of a software module which automatically converts an input text into a file appropriate for speech synthesis using the mbrola speech synthesis program. The system uses the diphone data base for Venezuelan Spanish consisting of 794 diphones (Rodriguez et al 2003). The module is written in Perl. In this way all the necessary ingredients for text to speech are completed. Complementing the software, we include contributions regarding the generation of symbolic and numeric entonation patterns, segment duration, context dependent phonetic changes and the transformation of the phonetic sequence in the SAMPA representation.