Singing Voice Beautification with Pitch Correction and Expressiveness Enhancement

CONTUNER, a diffusion-based model, can efficiently beautify amateur singing voices by correcting pitch and enhancing expressiveness without requiring paired professional-amateur data.