Consistency Model

The paper introduces CoMoSVC, a consistency model-based Singing Voice Conversion (SVC) method aimed at achieving high-quality generation and high-speed sampling. The authors first design a diffusion-based teacher model specifically for…