VoiceLens: Controllable Speaker Generation and Editing with Flow

Yao Shi, Ming Li

Currently, many multi-speaker speech synthesis and voice conversion systems address speaker variations with an embedding vector. Modeling it directly allows new voices outside of t…