Fig. 1From: Reducing over-smoothness in HMM-based speech synthesis using exemplar-based voice conversionConstruction of source and target dictionaries for each utteranceBack to article page