search

EvelynFan / FaceFormer

[CVPR 2022] FaceFormer: Speech-Driven 3D Facial Animation with Transformers
MIT License
761 stars 135 forks source link

How to train at 60fps on VOCASET #62

Closed Parikshat1998 closed 1 year ago

Parikshat1998 commented 1 year ago

The original vocaset dataset is in 60 fps. But I checked in the wav2vec.py file that fps are then reduced to 30 fps. If I want to train at 60 fps what needs to be changed? @EvelynFan

Parikshat1998 commented 1 year ago

The model only supports 30 fps