I am somewhat less convinced of the "320 kb/s is indistinguishable from a CD" assumption today than I once was.
If I am intimately familiar with a piece of music, such as one of my own mixes, I actually can hear the difference between MP3 and FLAC, and prefer the latter. In a blind A/B/X test I got a 100% hit rate - but only with my own mixes. I cannot, however, distinguish between 320 kb/s MP3 and 256 kb/s AAC under any circumstance.
When I listen to the radio in my car I am listening to AAC-encoded files, as that is how content is distributed to radio stations. Do I notice? No. After technically-necessary bandwidth reduction and compression applied by broadcasters, the wildly varying frequency response and resonances of my car's sound system - all competing with road noise - any limitations of lossy encoding are microscopically trivial. If the broadcaster had instead used an MP3 or a CD, I'd be none the wiser.
The other time I listen to file-compressed audio is during my nightly routine: in the dark, in a quiet room, with high-quality headphones. Sometimes with herbal assistance. It's a mixture of ripped CDs (WMA) and purchased MP3/AAC files that are typically 256 kb/s or less. This should be very close to an ideal listening situation, but do I care which format I'm listening to? Nope.