I tend to take all these discussions as a bit pointless. I've been doing this kind of thing for a very long time now, and I'd like to think I can recognise a good recording and a bad one. As soon as it's in the good category, then the differences are just differences, which we like or not. I rather like the compression and change in 'tone' that the minidisc system introduces. I like the sound of it. My mixer can play wav files but only in certain formats, 'downgrading' from my recording format make very little difference. Sure - some of the data wrangling changes the sound, sometimes for the better, sometimes for the worst - but it stays in the good camp. A 320K, or even 256K mp3 can sound pretty good on some tracks. I can't subscribe to the fuss and bother around the merits of ultimate quality. Cosmodan has discovered his system renders differently sometimes. I cannot quite reconcile this fact, unless there is a fault somewhere. If the data is contiguous, and intact, then the rendering is a mathematical process and should be bit for bit identical. If it is not, then it is faulty. The notion that a supposedly identical file sounds different makes no sense at all. Certainly, all the output files I produce without changing any parameters have identical file sizes. If it sounds different, it will not be the same, and a file comparison program should find a difference?
On my system, which is built around Cubase, I have projects going back to the 90s. These are mainly 44.1K 16 bit using wavs, but later ones are 32 bit with even a few at double the sampling rate, and quality wise, I really don't notice. I've recently been working on a few which appear to have some tracks as mp3s? No idea why, but they don't sound bad.
I simply choose to believe that some of my own material sounds better than others, but in general this 'better' relates to performance and the track itself rather than any technical specification. I can load up a track and say good or bad quality very, very simply! I do NOT need to check the file specs.