Fig. 2From: Reducing over-smoothness in HMM-based speech synthesis using exemplar-based voice conversionConversion stageBack to article page