Python in MIR, round 3?

[brm...@gmail.com]

Since 2012, we've had unconference sessions on using python (and not, say, other MAThematical software designed for LABoratory environments) for MIR.

To recap, 2012 saw a lot of interest, but not much in the way of practically useful tools.

In 2013, we ran a miniature bootcamp on getting python running, and using IPython notebook.

Much has happened in the world of python since then!  So let's do it again!

A rough outline would go something like:

- Discussion of existing tools for signal processing, scientific computation, symbolic music analysis, and evaluation

- Best practices (perhaps some overlap with the reproducibility session here), including the latest and greatest developments in IPython

- Discussion of "what's missing" and planning for the next year of development.

Who's in?

[Eric Humphrey]

Clearly I'm in, so long as we can avoid scheduling conflicts. I agree that this may have some overlap with the reproducibility / best practices discussion, but it'll be nice to have that conversation in as language-agnostic terms as possible, 

As far as a Python discussion goes, I definitely want to up-vote the discussion about missing tools and what can be accomplished in the next year. This might make for good fodder for possible software projects for next year's conference (in the vein of mir_eval). A couple ideas are...

• A deep network library: there's plenty of tutorial code floating around, but there's been talk (@colin, @sander) of producing something "importable". I plan on releasing my deep net framework in the very near future, but it's almost certainly a bit overkill as a starter library and fills a different need (serialization and cross-language support).

Ah, I've others but need to get on a plane to Taiwan now...

ejh

[张旭龙]

did you finish it ? what your deep network python library?