Never underestimate the power of the desire to cheat. Making yourself seem superhuman is a big temptation for some personalities. Blind testing is great. But unsupervised and unmonitored tests aren't necessarily any more authentic than sighted subjective ones.
I did a test on another forum where I took two clips, one of choral music and one of orchestral music. I bounced it out to a half dozen different codec/bitrates including lossless and then edited it together and upped it to lossless. A bunch of people had a very hard time identifying anything over mp3 128. One guy who had made a big deal about how he had golden ears identified a specific time in the track that he said was lossless. He was correct, but he had missed the choral piece that came before which was also lossless. (They were paired in every sample with a big space between, so if he had listened, he should have known that the beginning of the choral piece was the spot, not the beginning of the orchestral.) I asked him if he had just looked at the frequency spectrum with an audio program and chosen the only place in the track with high frequency info at the edge of human hearing. Choral music doesn't have any sound up that high, so it was invisible searching for the lossless that way. He promptly shut up and disappeared and hasn't been seen since. Later, I found a post where he said that he had an Mbox setup and sent the output of his computer through a pair of severely band limited wireless guitar amps to a set of computer speakers. Good luck hearing anything with that.
Can anyone cite an independently verified listening result where someone is able to consistently ID lossless vs AAC 320. I don't believe anyone can do it.
And for me, "small enough" is anything I can't hear.
Edited by bigshot - 1/18/14 at 7:35pm