In this paper, we present linear transformation algorithms for many-to-one voice conversion (VC). Many-to-one VC is a technique for converting an arbitrary source speaker’s voice into the target speaker’s voice. A conversion model previously developed between many prestored source speakers and the target speaker is adapted into a new source speaker in an unsupervised manner. In this study, we implement several well-known model adaptation techniques based on linear transformation for many-to-one VC and evaluate their effectiveness.
Index Terms: many-to-one voice conversion, Gaussian mixture model, unsupervised adaptation, linear transformation