I spent about a year or so reading and reimplementing interesting papers around machine learning on audio - probably six months was actually getting to know sig

I spent about a year or so reading and reimplementing interesting papers around machine learning on audio - probably six months was actually getting to know signal processing in a real way and learning tensorflow, and then some months doing seq2seq for ASR slightly more seriously. And now I've been doing things that you can find in the lifeclef or dcase competitions, with a group that focuses on one of the sub tasks.