search

yl4579 / StarGANv2-VC

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

Support needed for Teacher Forcing HiFiGAN Vocoder Finetuning #69

Closed MuruganR96 closed 1 year ago

MuruganR96 commented 1 year ago

@yl4579 Thank you so much for this wonderful work

Support needed for Teacher Forcing HiFiGAN Vocoder Finetuning. Couldn't training converted Mel spectrogram as input to HiFiGAN fine-tuning.

Thanks

yl4579 commented 1 year ago

For vocoder fine-tuning, you may want to fine-tune the vocoder to match the second-time converted samples (cycle-consistent samples) with the ground truth. The converted samples do not have ground truth so it cannot be fine-tuned.