A speech prosody conversion system using a newly developed high quality speech analysis-synthesis method is presented. This system can convert prosodic features such as average pitch, pitch contour, speech rate and so on with small impairments in original voice personality. Users of this system can convert prosodic features of desired part of original speech waveforms to what they want by themselves with interactive operation. The speech analysis-synthesis method used here is characterized by correcting spectral envelopes distorted by pitch manipulation to keep original spectral envelopes.
Keywords: Pitch contour, Speech rate, Prosody conversion