Trying to Keep Myself Honest

Posted on Sat 28 May 2016 in Data Science

This post is mostly to keep myself honest. Figured if I make it public, it'll help me keep motivated to work on the project.

I wrote a scraper and gathered many 1000s of lyrics to heavy metal songs. I have a few tweaks left to extracting some of the content, and need to settle on storage (postgres or something) and I'll put the code up on GitHub.

Next steps are to see if we can discover sub-genres from lyrical content alone. Like black metal, gore metal, death metal, grindcore, doom, etc. First, we'll see if unsupervised techniques work, and then maybe try supervised. (Need labels from somewhere else though. That doesn't come with the lyrics).

And then I want to explore neural nets. I know nothing about them and inspired by this post, I want to see if we can teach a neural net to write Cannibal Corpse lyrics or something. Just cuz.