Voice synthesis application based on syllable concatenation

This article presents a voice synthesis application based on syllable concatenation. The system is dedicated for Romanian language, so it was need to work on special rules to decompose Romanian text into syllables. Also for preserving initial prosody of text, accentuation of syllables inside word had to be determined. Then we have recorded a vocal database with the most frequent syllables of Romanian language. A unit matching algorithm matches linguistic units from the input text and acoustic units from database. Acoustic units are then concatenated and converted into sound by mean of a synthesizer.