Abstract
A revised Holmes-Mattingly algorithm for speech synthesis from phonetic input makes it possible to produce speech when control parameter values have been generated only for the first syllable. The algorithm is fast enough to produce speeded speech for a reading machine. With its supporting program, it is being used as a teaching tool for the synthesis of varied languages in an undergraduate speech synthesis course. Pass 1 labels the syllable structure, the allophonic contexts, and the pitch contour of each breath group in the utterance. Pass 2 calculates the control parameter values, including F0, for each syllable in turn. Major effects of voiceless intervals on the F0 contour are provided for. Place of articulation is referenced proportionally in order to facilitate the changing of apparent head size. The algorithm is being used for computer control of the Haskins and Glace-Holmes parallel resonance synthesizers, and of the OVE III serial synthesizer.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
