innnky/glow-svc

singing voice conversion based on glow-tts

Glow-SVC

another implement by rcell based on official glow-tts repo

• content-vec + fcpe(f0) -> glowtts -> nsf-hifigan -> wav
• 坑多多多多多
• fp16会炸炸炸

pretrain:

• content-vec-best
• fcpe
• nsf-hifigan

preprocess：

• resample.py -> preprocess_flist_config.py -> extract_vec.py -> extract_f0_mel.py

train:

• python train.py -c configs/config.json -m model_name

infer:

• 暂无