Processamento linguístico para um conversor texto-fala para a língua portuguesa.

Abstract

Due to its characteristics and applications Voice Response Systems (VRS) have grown in importance in the past. SRV can be classified as Stored Voice Systems (SVS) and Text-to-Speech Conversion Systems (TSCS). SVS and TSCS pose a major difference: the former originates speech from words previously recorded and stored. The latter employs a technique to produce voice by concatenation of acoustic units, allowing the implementation of the entire set of words of a language. TSCS present two main modules: linguistic processing that characterizes an idiom and a signal processing unit responsible for sound synthesis. The main objective of this work is to formulate a TSCS structure, considering the Portuguese language. This structure must be able to reproduce a natural and comprehensible speech for the Portuguese language spoke in Brazil.