yl4579 / StarGANv2-VC

StarGANv2-VC: A Diverse, Unsupervised, Non-parallel Framework for Natural-Sounding Voice Conversion
MIT License
466 stars 110 forks source link

Training questions: samples and resuming #31

Closed skol101 closed 2 years ago

skol101 commented 2 years ago

I've started training and have questions:

  1. I can see train/eval phases, but no wav samples in the log_dir
  2. Can training be resumed from ? I assume it can if I put "epoch_xxx.pth" in the config file ?
yl4579 commented 2 years ago
  1. There is no waveform generated in the training and evaluation process, so you will need to do it yourself with a vocoder.
  2. Yes, just edit the pretrained_model field in the config file.