At first I thought this was a DSP problem. It wasn't. The solution was much simpler.
To start, I listened to the entirety of the song. It's not quite everyone's favorite genre, but I rather enjoyed it. I noticed a short burst of static at the beginning. Static typically means data, but I figured I'd look for other patterns.
Some GNU Radio later showed that there were no obvious patters in the spectra of the signal or in the relation of the right and left channels.
Initially, I incorrectly converted the file to a raw type. In the converted file, there was obviously some data at the very beginning in the burst of static, but it wasn't readable as anything.
I re-did the conversion to raw and looked again. The static was an ELF for linux x86_64. Thinking it couldn't be that easy, I threw it into IDA. It's a binary that grabs samples from the raw audio and prints the samples out as characters.