A few days ago I read something about 1dB being the minimum difference in volume that can be noticed — or something along these lines. I know that when I am approaching the final balance for a track, I often end up doing much smaller changes (especially on vocals, or parts which I want to stand out without 'ungluing' them), easily of 0.1 or 0.2 dBFS. It does seem to make a difference to me — in the sense that I end up with a defined and repeatable preference between two options which are a 10th of a dB apart. Obviously in these situations my brain is highly attuned to the details of the mix in question (something that happens after listening to it some 20 times in a row!), so I can perceive and appreciate a lot of details that I normally wouldn't. I will keep on doing it regardless, but I wonder if I am fooling myself into believing there are differences or actually other people experience that too?
SOS Technical Editor Hugh Robjohns replies: I don't know the context of that, but I believe the 1dB thing comes from testing people's ability to detect absolute level changes. So you play a tone or a track or something, and then play it again with a small level change, and see whether the listener can identify whether the level is the same or not. A skilled listener is supposed to be able to hear changes of 1dB or more. Less skilled listeners need more like 2 or 3 dB before they are sure there is a difference. (This is measuring source signal amplitude, by the way, not acoustic power.)
But listening to sound levels in isolation is a very different thing to listening to the relative internal balance of two or more sounds. As you say, a skilled listener can easily tell the difference in relative mix levels to a much finer degree — 0.25dB being a figure I've come across often.