Magdalena Fuentes

What do they use in Task 3 to evaluate the models do you know? @iranroman? I think that ideally we want `soundata` to load things in memory in a way...

Hey @iranroman thanks a lot for taking a look at that! The discussion about `SpatialEvents` is making me think of adding a few extensions to the annotation types that were...

Hi both, I'm in sync with the suggestion of a child class `SpatialEvent` that inherits from `Event`. However, I think we should implement it in `annotations.py` because it would be...

Hi @iranroman Thanks for reviving this and bringing up the standardization issue with spatial annotations. I think we could take this as an attempt to standardize this annotations, and draft...

Also it should be easy to navigate dataset-level annotations in case the user wants to e.g. get some stats on the entire dataset

After a conversation offline with @pzinemanas we think that it would be great to have a couple things at the dataset level: 1. Access to splits if there are any....

1. I'm a bit more inclined to have a unique function and change the dictionary keys depending if it's `split` or `fold`. Maybe we can call it something more general...

@justinsalamon should we include this in the `Release 0.1.0` milestone?

Before I go further in the implementation of the tests let me know what do you think @justinsalamon @nkundiushuti @iranroman

Good catch @sripathisridhar @justinsalamon may wanna take a look?