The generation of waveform fluctuation for the enhancement of the quality of sustained vowels

It is indicated that the pitch-synchronous subtle waveform fluctuations of sustained vowels are the acoustic cue for the naturalness in case the sustained vowels do not include fundamental frequency fluctuations. From the viewpoint of speech synthesis, it is expected that the more naturally sounding sustained vowels can be synthesized by incorporating the appropriately modeled waveform fluctuations. In this study, the waveform fluctuations of the inverse-filtered normal sustained vowels were analyzed with the aim of high quality speech synthesis. The series of the analyses suggested that the waveform fluctuations could be modeled by the same rule for all speech samples obtained from ten male subjects. In addition, the psychoacoustic experiments confirmed the validity of the method which generates the completely artificial waveform fluctuations for the remarkable enhancement of the quality of sustained vowels.