Data augmentation for audio

Specific Domains Machine Learning
,
1

Hey!

We are likely to embark on an adventure to create something very similar to @Evizero’s excellent package Augmentor.jl, but for environmental acoustic recordings. I think the overall idea of Augmentor.jl looks solid, and the support for audio would require us to simply implement a number of new Augmentor operations.

Supporting audio would likely require some new dependencies, while some of the existing dependencies of Augmentor are not required for audio. Would it make sense, and would you, Evizero, be supportive of, separating out core functionality of Augmentor.jl into an “AugmentorBase.jl” or similar? Augmentor.jl, as well as a potential new AugmentorAudio.jl, would then make use of the common functionality in AugmentorBase.

1 Like
2

I have a sense that someone will also create a TextAugmentor.jl eventually :smiley:. Though the methods would differ a lot.

3

I like the idea (and I appreciate being asked). The issue is that I have not had the time to spend any serious amount of work in julia related activities. I don’t think Augmentor currently works on 1.X for example.

If you do want to reuse things I encourage you to copy paste the code excerpts you deem useful into some “augmentorbase.jl” file in your new package. There is always the option to pool resources at a later point and then refactor into a dedicated AugmentorBase later

1 Like
4

Thanks for getting back to me. I’ll see what we can make of it and will keep you in the loop in case we move forward!

1 Like
5

Did you get anywhere with this new (potentially) project. I think this would be a useful package.

Robert

6

I started here
https://github.com/baggepinnen/AugmentorAudio.jl
but did not get very far, and have not touched it in a few years.