More audio file-type support about textract

[achembarpu]

Possible file-conversion API - CloudConvert
It has a Python Wrapper, as well - python-cloudconvert

However, the py-wrapper is licensed under GPLv2, while textract is under the MIT license.
They are partially incompatible, so that might be an issue.

[deanmalmgren]

I'm open to any of those options. Whatever is the most effective and functionally correct is great as far as I'm concerned.

I'm not as familiar with audio, but the convert unix command is great for converting images between various formats. Is there an equivalent command for audio so we don't have to hit a cloud API for conversion?

As far as having a dependency on a GPLv2 library, I don't think that's a problem. Just include it in requirements/python and we should be good to go. Things get messy when you actively distribute the source of the GPL-ed code with your package, which we aren't doing.

[achembarpu]

Not exactly equivalent, but I think a combo of ffmpeg, flac, etc..., could do the trick.
Won't be as portable as a cloud-solution, but definitely faster.

I'll test both solutions out, and see how it goes.

Exactly, that's the problem - it isn't hosted on PyPi/other, so we'd have to include the source.

[deanmalmgren]

Exactly, that's the problem - it isn't hosted on PyPi/other, so we'd have to include the source.

Ahh, I see. I'd recommend taking a similar approach as in #17 then and starting a PR that includes a setup.py to install the library as a python package. It doesn't have to be hosted on PyPi for us to include it so long as it has a setup.py. For example, we can add something like this to our requirements/python as soon as you get something working:

-e git+http://github.com/arvindch/python-cloudconvert@setup-branch#egg=CloudConvert

[achembarpu]

Ah, alright then. Sounds good.

I'll explore our options, and finalize something after some testing.

[achembarpu]

I found a cool cmdline tool - sox.

sox audio.x temp.wav
textract temp.wav

Looks straightforward, and easy to implement.
Works really well on my end.
I'll have some initial code up in a while.

edit - Code is up - PR #62.