Can you please share how you pre-processed crema-d dataset, and the dataloader for it. Due to the differences in audio sampling rates and the number of frames between the videos, I am getting a synchronization error. Maybe it is not as straightforward in the case of crema-d dataset, as it for grid and lrw dataset which are already pre-processed.
Can you please share how you pre-processed crema-d dataset, and the dataloader for it. Due to the differences in audio sampling rates and the number of frames between the videos, I am getting a synchronization error. Maybe it is not as straightforward in the case of crema-d dataset, as it for grid and lrw dataset which are already pre-processed.