Also, I've learned that reinvention often takes less time than learning someone else's service. If the API is really clean and useful, maybe it's worthwhile, but in most cases, it's not. If only because the vast majority of the time, I'm expected to use what's effectively a beta. I can write beta software too!
For this service specifically, just because I can self-host won't make it any easier to move off of it. Moving from iTunes to Google Play was pretty easy once I grabbed a library off of RubyGems to hook into iTunes. I'd have a coding task to do to move off this too. My own software can have everything I need, so I won't ever need to move off, just write translation layers. It's more work up front, but buys me a lot of flexibility down the road.
Then I want to grab all the files and put them on S3 or something and store the links in the db. Then it's a front-end problem, I'll need a command line interface and then one day I'll make a GUI. I could also find tracks that exist only, say, on YouTube, and store links to those too.
Then I'll be able to use it as a music player and store no music on my local machine.
Then the biggest pain in the ass becomes syncing to my portable devices, which is the only reason I haven't done this yet. I would hate to have to maintain two libraries at the same time. So I have to think my way through that. Though I'm sure I could write a syncer that would work much better than Google Play's, I haven't bothered to sit down to do it yet.