Neural network control for a cascade/parallel formant synthesizer

Neural network control of a cascade/parallel formant text-to-speech synthesizer model is investigated. The task under investigation is the synthesis of isolated words spoken in isolation and with neutral intonation. The goals are to have the network successfully learn to produce the temporal parameters that drive the synthesizer and to evaluate its performance on novice utterances. The synthesis strategy and the network architecture and training are discussed, and results for the generation of the fundamental frequency contour using feedforward and sequential networks are shown.<>