Athanassios Katsamanis1, Pirros Tsiakoulis1, Petros Maragos1 & Alexandros Potamianos2
1National Technical University of Athens; 2Technical University of Crete

ID 1580
[full paper]

Modern articulatory speech synthesizers simulate the human speech production system in an increasingly accurate manner. In this direction, we relax the simplifying assumption of zero mean flow velocity during speech production and we investigate potential effects. Further, we introduce a reduced parameter set for our 3D articulatory model which simplifies its control and does not allow humanly infeasible articulations. Vowel-Fricative-Vowel synthesis experiments using our twofold augmented synthesizer are reported.