A fuzzy approach to text-to-speech synthesis

In this paper we describe a fuzzy approach for the synthesis of a speech waveform from a phonetic description in Italian language. The system is based upon a set of fuzzy rules which linguistically describe the transitions between phonemes. The fuzzy system has several interesting properties, such as an easy management of the set of rules and the possibility to continuously improve the system by adding more knowledge to it using linguistic descriptions. Some experimental results are included, and future possible developments are outlined.