The LPC-MPE synthesis method is an alternative method used for obtaining a better quality of the generated vocal signal, that can be easily implemented in vocal signal coding-decoding systems. Using the method in text-to-speech systems is more difficult because of the modification that must be done on the synthesized vocal signal in order to superimpose prosodical effects. This paper presents our steps in this direction, some researches and experimental results obtained for adapting the system to the pitch-synchronous LPC-MPE method.