hparams for libritts training

NVIDIA / mellotron

Mellotron: a multispeaker voice synthesis model based on Tacotron 2 GST that can make a voice emote and sing without emotive or singing training data

BSD 3-Clause "New" or "Revised" License

855 stars 183 forks source link

hparams for libritts training #92

Closed krishnashankar closed 3 years ago

krishnashankar commented 3 years ago

Hi! I'm training on Libritts, and the rythm/gate outputs don't look quite right. Other than training_files, validation_files and sampling rate, are there hparams that need to be modified to train on libritts? Thanks!