search

Rongjiehuang / ProDiff

PyTorch Implementation of ProDiff (ACM-MM'22) with a Extremely-Fast diffusion speech synthesis pipeline
MIT License
431 stars 55 forks source link

Fine-Tuning/Transfer Learning? #5

Closed AnujSaharan closed 2 years ago

AnujSaharan commented 2 years ago

Do you plan to release instructions on how we could use the pre-trained checkpoints to fine-tune on a new dataset formatted like LJSpeech, but for a new single speaker?

Rongjiehuang commented 2 years ago

Hi, you can load the pretrained weights and train the model using your custom dataset following here