I would go the other way on that -- overwhelmingly, your customers want to get their music streamed, NOT get a pretty wrapper around a separate service that is way the heck over their technical skill level which they nonetheless have to understand. If I were you, I'd see if you could possibly structure it such that a) they paid you in advance and b) you stored all the music on YOUR S3 account, which is then an implementation detail that no one has to care about.

Now, of course, the obvious worry with this is the major advantage for your service is "If you use us, you can listen to music without having to pay for it", which is pretty much the selling point of every Internet music based startup I've ever heard of, and you'd probably prefer to structure things in such a way that you don't know that 90% of your customers are using your app as a convenient pirating service, because that will result in less legal hassles.

This is why hell will freeze over before Apple integrates any sort of cloud offering with their iPods. As long as they can't see the contents of their hard drive, they can continue with the polite fiction that their hipster twenty-somethings are filling those eight gigabytes with music they've either purchased legitimately or which is licensed appropriately.

More broadly I think the challenge with startups in the music space is coming up with a value proposition which is not "Using this service is a lot cheaper than paying for your music".