Fig. 4From: Reducing over-smoothness in HMM-based speech synthesis using exemplar-based voice conversionOver-smoothness reduction for spectral envelopeBack to article page