In this paper we present an strategy to improve the naturalness of a text to speech system for Spanish based on the development of a new pitch generator. The new model works with text previously parsed with an heuristic breath group parser. The results show that the new pitch generator is preferred by listeners compared to our previous model. It is also shown that errors of the heuristic parser can have a very adverse influence on the naturalness of the speech output.
Bibliographic reference. Moreno, P. J. / Martinez, M. / Pardo, José M. / Vallejo, J. A. (1989): "Improving naturalness in a text-to-speech system with a new fundamental frequency algorithm", In EUROSPEECH-1989, 1360-1363.