Support audio

[AlekseySh]

Let's add audio to the already supporting modalities like images and texts.

The plan:

• Add an analogue of get_mock_images_dataset
• Add the corresponding datasets inherited from the corresponding interfaces
• Add a sound processing NN model to Zoo, or add a wrapper to 3rd party sound models (like HFWrapper)
• Based on the objects above, create train / val example.
• Evaluate an example on a dataset having larger scale in order to check the example may be applied on real data.

[amanteur]

Hi, I'll do it!

[AlekseySh]

@amanteur What a coincidence!

You are welcome! :)