in other news, if you want to help me create an even better singing/speech classifier, you can now donate audio!
warbler.wasthatzero.net/crowd.html
donating is COMPLETELY optional and clips ran through the standard inference page are, as before, not kept and not used