Is the Fujisaki model a suitable (prosodic) model for the voice-conversion task?

摘要

This contribution focuses on the Fujisaki model and its use within the VC-task. The model can be used to modify the accentuation of any target speaker, if the model parameters of the source are mapped non-linear. Thus, the prosody of the target can be changed. However, the Fujisaki model was not tested in a VC-system so far. In a next step it will be tested in such a scenario.