Alleviating the one-to-many mapping problem in voice conversion with context-dependent modeling
Elizabeth Godoy, Olivier Rosec, Thierry Chonavel · 2009
This paper addresses the one-to-many mapping problem in Voice Conversion (VC) by exploring source-to-target mappings in GMM-based spectral transformation. Specifically, we examine differences using source-o nly versus joint source/target information in the classificati on stage of transformation, effectively illustrating a one-to- many in the traditional acoustically-based GMM. We propose combating this effect by using phonetic information in the GMM learning and classification. We then show the success of our proposed context-dependent modeling with transformation results using an objective error cri terion. Finally, we discuss implications of our work in ada pting current approaches to VC. Index Terms : Voice conversion, GMM, Spectral mapping.