Audio is contained in USound objects, it’s imported as wav and converted to ogg to aid in cross platform deployment.
Functionality like this would likely require significant modification of the engine source at the moment from what I have seen.
You might be able to avoid USound all together and integrate something to just work on one specific platform for instance with a lot less headache, but I don’t have the knowledge to know what’d really be involved for that.
There has been talk of FMOD integration into the engine coming soon. http://www.fmod.org/
I’m not aware if the middleware will add any features like this though.