An advanced singing voice synthesis system with high fidelity, expressiveness, controllability and flexibility based on DiffSinger: Singing Voice Synthesis via Shallow Diffusion Mechanism
Apache License 2.0
2.62k
stars
275
forks
source link
Split note sequence data in variance_binarizer.py #110
The note data from the csv needs splitting before sending to librosa and building Tensors, otherwise it runs character by character and breaks