Investigation on Shallow Diffusion Mechanism in Singing Voice Synthesis: Introducing Adaptive K Scheduler and SVS Evaluation Metrics

DiffSinger: Singing Voice Synthesis via Shallow Diffusion Mechanism

493G Final Project

DiffSinger, an acoustic model for SVS based on the diffusion probabilistic model.

TODO:

  • Document milestone
  • Deploy on VM
  • Know more about standard Diffusion model
  • move project proposal to main folder