Lately I've been trying to find a way to do exactly that - initially using audio in - but so far without any luck.
Any caches? Seemed interesting
If you want the technical details behind it, a good starting place is http://www.icrobotics.co.uk/wiki/index.php/Turning_the_Raspb...
Or the blog of the creator at http://www.omattos.com/node/19
Ideally I'd like for this project to run Android, and drive my car stereo too, and do a bunch of other crazy things like GPS.
There's surprisingly little information out there on how to approach such a project. I've looked for components I could use on Sparkfun, and there's not really anything that jumps out at me, for either building 2m/70cm transceivers, or even for doing the stereo amplifier part.
I did find this fascinating article, though: http://hackaday.com/2013/02/28/hacking-a-ham-radio/
I don't know the current drive capabilities of the Broadcom SoC, but let's assume it can perfectly drive a dipole antenna presenting around 72 ohms. The supply is 3.3V, I think. 3.3^2/72 = 0.15125W. In reality I doubt it's anywhere close to that, let's assume 50mW instead (it could be lower, I just don't know). According to a legit-looking field strength calculator I found online this works out to just over 0.5V/m at 3m! That sounds a bit high. Various factors would conspire to making even this much field strength unlikely, but it'd still exceed the FCC limits easily.
The Pi's output is also a reasonably decent square wave, as pointed out elsewhere. So there are some harmonics that are, themselves, above the part 15 limits.
In other words, don't do this, unless you have a spectrum analyzer, attenuator (don't forget to block DC). Actually, someone needs to do this with the right equipment just to find out how well it really performs.
one of the fastest ways to tell that someone didn't know what they're doing when they designed a board is that they placed the oscillator anywhere other than right next to the component it drives. the traces that connect the oscillator to the component are basically an antenna - the longer they are, the stronger they broadcast. This becomes an issue when there are multiple oscillators in close proximity - their broadcasts can interfere with each other and cause them to synchronize. I forget what this effect is called but here's a neat video that kinda illustrates it: https://www.youtube.com/watch?v=kqFc4wriBvE&feature=kp
