Designing, Implementing and Testing the Acoustic Component of a Text to Speech System for the Romanian Language

The large majority of the available Text to Speech (TTS) synthesis systems intended for commercial applications were developed for international languages. However, there are also a few precedents for Romanian language. The present paper resumes the authors’ experience in defining a generic architecture as well as in implementing and testing the Acoustic Component of the concatenative TTS system for the Romanian language that they have developed. The Acoustic Component is one of the most important parts of a TTS system being responsible for loading all the acoustic units, changing them in conformity with a prosody model and concatenating them in order to generate a single acoustic unit.