People have demonstrated they can hear this reliably? Not to get all sound science forum here, but may I ask if you can hear this noise over top of music, and if so how have you been able to test that out? There is a difference between measureable/theoretical and audible. I conducted a 7 subject, multiple trial per subject, very well controlled experiment at a head-fi meet a number of years back and not one person could tell the difference between a 320mp3 and the lossless master it was made from. I just find it hard to believe that something as tiny as RFI/EMI are audible to you over top of the music. Anyway, I guess we shouldn't take this further as it will derail the thread which isn't my intention. Cheers and enjoy the music as we always say.
Some years back a client purchased some patented IP from me and I was under contract to design new devices using this technology, and to embed the IP into the company. It was a large semiconductor company, and they had some extremely clever and talented engineers. Unfortunately, when explaining details of the IP, it often came down to very lively discussions as to why it was necessary to have performance levels that were way beyond the ear's threshold of audibility. I on the other hand, have consistently heard errors that are much smaller than the ear's ability to detect from the threshold of audibility POV, and the reason these small errors are audible is down to the brain's processing of the data from the ear, and the threshold of audibility tests do not cover the brain's processing.
Anyway, I was getting fed up with the arguments, as the engineers would not accept that these levels were important. So I set up a single blind listening test with one of their non-audiophile engineers at my home. Basically I was comparing two different noise shapers, that both have acceptable performance from the threshold of audibility POV. The performance is covered in my Hugo technical masterclass slide:
What was interesting was the sound quality differences - the bottom noise shaper had better depth and detail resolution, and sounded significantly warmer and smoother. The change in warmth is down to noise floor modulation, as the noise floor at -190 db of the first noise shaper improves to -200 db with no signal present. This level of noise floor modulation would not be measurable on real test equipment, as the analogue noise would swamp the noise shaper noise floor modulation.
After the test, the engineer came up to me and said that what surprised him was not that he could actually hear a difference, but that the difference was so easy to hear - it was a night and day change to him. Anyway, the guy wrote a report to the company, and thereafter I got no arguments about the necessity of doing things to this level of performance.
As to your comments about MP3 - well on my BluDave I can very easily hear the difference from 320k AAC to CD quality - the AAC sounds soft and warm by comparison, and it's very apparent.