I know this thread is two years old...
There's the math and the mathematicians. There's the music and the musicians.
There's the audiophile and his expensive stuff, and there's the music lover who knows what to listen for.
Here's a practical approach to the whole thing by someone who's been mastering in analogue and digital.
When we were recording on tape, we knew right away what the difference was between the live and the recorded. Even in the best studios there was a clear difference between 15ips and 30ips. Your album would sound entirely differently should you choose an option over the other. No question about it.
The digital domain made us look for the differences in the same places, but we were looking at the wrong place. For example, it became clear that the closest you'd work to the limit of the 16bits, you would get loud clipping, with no warning. So we'd give ourselves some "space" not to go over, therefore loosing a little of the 16bit dynamic range. From this standpoint, here's my first rule: 24bit allows you to work further from its full dynamic range without loosing quality (I'll get to the "quality" down-to-earth definition later), but stricly from a "practical" point of view.
- One of the first CD's I heard was "Sting - Nothing like the sun". My first edition CD was lousy at best and I prefered the vinyl version. Listening to "Be still my beating heart" intro with a fade-in, it was clear in my head what we had to look for to detect "quality" differences in digital masters. In this very example, it becomes clear that we hear the floor of the track much higher than it should be, letting us hear what poor dithering sound like. Without that dithering, we would have hear the actual 16bit steps of amplitude one by one until a pivot point where the amplitude of the mix overshadows the artifact itself.
- Testing with recordings made at 16bits VS 24bits, I've discovered that the break points of fade-ins and the dying point of fade-outs are where 24bits recordings really stand out. There are only a few recordings in which I've noticed that 24bits would have been better and only 1 is of analogue source, i.e. Supertramp - Crime of the century, between "School" and "Bloody Well right". The combination if a fade-out and a note glissendo created that artificial second "note" that a 24bit would not have, because of the "smaller steps" between the specific amplitude levels of 24bits, much smaller than those of 16bits. This is subtle, but it is there. On the vinyl, I cannot hear it, simply because it's not there. So rule #2: 24bits is better than 16bits at LEAST when we expect to record and reproduce notes, chords, strings, that will either make a up or down glissendo AND an up od down crescendo.
- Today's albums are mastered and compressed so they can be heard loud and clear on the car radio, in your iPod, table top system... Most sound LOUZY when played on a HI-END system, capable of better nuances and subtleties, because the CD is recorded that way. 20 years later though, when they'll release the "remastered" versions of these recordings, they'll be happy to have chosen (or unhappy if they haven't) to record in 24 bits when they'll decide NOT to make it sound as "punchy", but the 24bit original recordings will let them master the whole thing without any audible compromise on definition. Rule #3: Use it now if you have it, you're making choice upwards.
- I've heard re-mastered albums from the 70's when digital wasn't around yet, put on SACD... They sound good. They sound JUST AS GOOD when digitally burned on 16bits because the tape resolution and noise floors are way ABOVE those of 16bits/44kHz. If they "sound" better, it's because they've added EQ, filters and whatnot. They would have come out just as good on 16bits with the same corrections. Your original vinyl in perfect shape on a hi-end turntable is even better than those remasters, but such conditions are rare, so very few people can compare. Rule #4: Old masters won't sound any better whether they're put on 16bits OR 24bits OR Sacd. The original quality is just not there yet. It would be like expecting colors to appear from a B/W movie when copying it on DVD, or Blu-Ray. When they sound "better" than, say, the MFSL version of it, they've just made different EQ's and filtering decisions. All of which will sound just as good on any 16-24-Sacd format.
Very few know what to listen for when attempting to compare bit depths. For instance, a know-it-all seen-il-all put up a site where you could listen to 4 wav's files, recorded at different bit depths and the author asks to guess which one is which. The problem is, the author revealed his lack of knowledge when he chose the musical passage he did, in which it was impossible to determine that with such little mucical information. I'm not going to mention his name but for those of you who have search for answers like me, you've heard of the whiner I'm sure.. My point is, you're only going to hear the difference IF you know what you are searching for AND if you KNOW what MUSIC should sound like in the first place.
*I've attempted to be as little technical as possible to reach as many music lovers as possible.
