ewrcode

Cool. I think there's huge disruption to be done in the tech stack and listening model for broadcast radio.

I wrote up a few ideas back when KPCC was applying for a Knight Foundation News Challenge grant:

http://ewr.is/2012/03/1788-taking-radio-beyond-the-play-butt...

StreamMachine supports all the cool bits underneath, but we've never gotten a chance to do the client integration to really take it to the next level.

I just pushed an updated example config and a chef cookbook that includes a Vagrantfile for installing a basic config in a VM. Both give you the most basic of setups, but they should at least point you in the right direction.

Just streaming audio isn't a hard task; old hardware can handle it pretty well. What I've found more taxing in the KPCC deployment is all the junk connections that cause us to go through setup and teardown just for a script that's looking to pull metadata or who knows what else... I haven't seen KPCC's historic stats in a while, but I'm guessing they're probably doing roughly a million hours a month off three 5-year-old Dell 1U's.

While working at KPCC, a Los Angeles-area NPR affiliate, I wrote a Shoutcast replacement in Node.js that the station has been running in production for the past 18 months. It is open-sourced and lives at:

https://github.com/StreamMachine/StreamMachine

I'd love to get others involved in the project. It's been doing great work on the bread-and-butter live listeners, but there's so much more potential out there for building smarter clients and listening models.