https://hgpu.org/?p=24045
FastSVC: Fast Cross-Domain Singing Voice Conversion with Feature-wise Linear Modulation