Text-to-speech conversion system for Brazilian Portuguese using a formant-based synthesis technique

This paper contains a general description of the text-to-speech conversion system for Brazilian Portuguese developed at the Signal Processing Laboratory of FEEC-Unicamp. The system is fully operational and performs the whole process of text-to-speech conversion, including components for text processing (preprocessing, grammatical classification and phonetic transcription), prosodic processing (generation of intonation patterns and segmental durations) and signal processing (using the Klatt formant synthesizer, with a synthesis-by-rule technique for generating synthesizer control parameters).