In this contribution, we present the recent improvements of the Spanish version of the MULTIVOX text-to-speech system [1,2]. First the adaptation of the system to the Spanish version is considered in the framework of MULTIVOX. Then a new fast linguistic module for Spanish is probed to improve the naturalness of the synthetic speech. This result validates the methodology for prosodic modelling and encourage us for future improvements.
Keywords: Text-to-Speech Synthesis, Prosodic Processing