Millie's correct; I'm not intentionally doing anything special beyond lowering the bit depth and sampling rate. Converting 10-second-or-less audio clips in this way and saving them from Decker is pretty easy; something like the following in the Listener or a buttons script:
write[read["sound"]]
If you use Native-Decker this will only support reading .WAV files, but for Web-Decker it ought to work for any audio format your browser can handle.
Dealing with arbitrarily long audio is possible, but would be more complicated from within Decker.