Fig. 3From: Multimodal voice conversion based on non-negative matrix factorizationFlow chart of multimodal conversionBack to article page