Standardize loading for auditory data

[adwaykanhere]

No description provided.

[PSSF23]

@adwaykanhere could you explain more on how to standardize? Is it different from #31 ?

[adwaykanhere]

By standardizing the code I mean refactoring the code such that it follows PyTorch's Dataset and Dataloader primitives. This can help add transforms/augmentations more easily.

[PSSF23]

I see, is this only for auditory? If so I would specify in the title.

[adwaykanhere]

I will do it for Auditory first, but technically this should extend to images as well.

[PSSF23]

Vision already uses Dataloader. We used different loading functions for cifar and svhn because their attributes are named differently in pytorch.

[adwaykanhere]

I see, thanks for the clarification.

[PSSF23]

I don't think we will use the package directly but this one might be useful: https://github.com/eonu/torch-fsdd

[adwaykanhere]

Thanks, will check it out!