What you need to understand is
WHY most studios use 24 bits (or even 32 bits) when recording, and how that relates to you - as a consumer -
PLAYING a recording.
What the number of bits (depth) affects is how much dynamic range you can have in a recording; the "distance" - in terms of loudness - between the quietest sound you can record and the loudest sound you can record. (The loudest sound will be determined by the point at which the signal clips; the quietest will be determined by the noise floor.) 16 bits (which is what CDs are) offers a dynamic range of about 96 dB; 24 bits offers a dynamic range of about 144 dB; those numbers also equate to the best possible theoretical signal-to-noise ratio you can have for audio recorded in that format. (Note that this has nothing whatsoever to do with sample rate - it is
ONLY dependent on bit depth.)
Now, when you're mastering a CD in a studio, you get to sit there and carefully adjust everything to take the maximum advantage of whatever format you're using. You carefully adjust things so the loudest sound is close to the maximum allowed for your chosen medium, which gives you the widest difference between that and the noise floor; you use compressors to allow you to turn up the average level while carefully avoiding overloads; you boost the things that are "lost in the noise"; in short, you get to
optimize your levels. In this situation, 96 dB is probably plenty of dynamic range; if the average listener sets his (or her) system so the loudest sound on a recording is at the threshold of pain, a sound 96 dB below that almost certainly won't be audible above the background noise of even a very quiet residence. For comparison, good vinyl has a dynamic range of 60-75 dB; and that is
VERY optimistic. This is why 16 bits was chosen for the CD standard.
Find the quietest and best sounding CD you have; turn it up until the loud parts are as loud as you can stand them; can you hear the background hiss on the quiet parts? (It's very doubtful that you can. If you hear hiss, odds are it's coming from one of your other components and not that CD.... or from the master tape from which it was made.)
However, have you ever recorded a live concert?
Things are sort of different then; you don't get to optimize anything! You have one chance to get things right; if you set your gain too high, something clips. Clipping is bad! It sounds nasty, and is pretty much impossible to fix after the fact. Clipping will ruin your recording if you let it happen.
There are two ways to make absolutely positively sure that this isn't going to happen in your live recording:
1) Use some sort of really
AGGRESSIVE limiter (compressor). A compressor reduces the dynamic range of everything you put through it - it makes loud things quieter and quiet things louder - so the loud things don't clip and you can hear the quiet ones above the noise floor; a limiter ostensibly only acts on sounds that exceed a certain preset "threshold". Most "dynamic processors" can do either, or a combination of both. A limiter acts quickly to even out the level of whatever you put through it. A good limiter can make it almost impossible to clip things. Unfortunately, it will also produce that "homogenized" "everything at the same level" sound that we all hate in modern recordings. A limiter is cool for recording speeches and sermons, and is probably much better than actual clipping, but it's not at all good if you want to record music and have it still sound "dynamic" and "alive". (If you go to "pro" groups, you'll find discussions about which limiter sounds best, and who prefers which one. Do remember that, amongst music
PRODUCERS, the fact that the limiter alters the sound isn't necessarily a bad thing - which is why you'll find endless discussions about which one sounds best.) The biggest "downside" of a limiter or compressor is that it is more-or-less irreversible; if you decide later that your recording is sort of squashed, you're sort of screwed. (There are dynamic processors that can "uncompress" things, but they tend to have nasty side effects, and are nowhere near a perfect solution.)
2) The other way to absolutely avoid clipping is simply to turn the gain down far enough that the loudest thing you will be recording cannot possibly overload anything. Unfortunately, owing to the nature of live music, this may mean turning the gain down 20 dB, or even 30 dB, below the level at which you would really like to record the "normal" stuff. (You make some loud noises, set your levels so the loudest just clips, then reduce the level 10 dB or 20 dB - "just in case".) Unfortunately, by turning the gain down far enough that "you can't possibly clip anything" you are "throwing away" a lot of your dynamic range. In fact, if you're recording using only 16 bits (with a dynamic range of 96 dB), but then you're backing off 20 dB to give yourself plenty of safety margin, suddenly your "usable" dynamic range is only 76 dB (and you
CAN hear a noise floor at -76 dB) .
So, assuming you take option #2, using 16 bits of depth (96 dB of dynamic range) really isn't enough.... Now, if you record at 24 bits instead (with its 144 dB of dynamic range), even after "throwing away" 20 dB for spare headroom, you
STILL have 124 dB of dynamic range... which is a whole lot better. Now, the reality is that no other equipment (like microphone preamps) really has quite that much dynamic range anyway, but, by using 24 bits instead of 16 bits, you do gain a huge safety margin. (You record things at a lower level, and you have plenty of "room" to turn them up later - in the mix - without raising the noise floor up to where you can hear it.) In practice, depending on the recorder, this may gain you more or less, but using 24 bits is always a significant improvement over 16 bits.
There's also another reason to use 24 bits....
When you edit things afterwards, you often need to raise the level on one or more tracks (or for short periods of time). You
REALLY need to make sure that your noise floor is totally inaudible to avoid hearing the noise "breathe" or jump up and down when you change levels or make cuts or changes. Even beyond that, most special effects and other types of digital audio processing involve lots of math, and that math introduces the possibility of rounding errors - which can add up after performing multiple operations. These rounding errors occur in the "lowest few bits".... In other words (to oversimplify) they might mess up the lowest one or two bits of your audio. If you only have 16 bits to begin with, this can be a problem; if you use 24 bits, then throw away the lowest 8 bits when you master your CD, then the problem is gone. Editing also often involves altering frequency response; if you turn the treble up 10 dB to add a little more sparkle, that's also going to boost your noise floor by 10 dB; and it sure would be nice to be able to do that without the hiss suddenly becoming obtrusive. Most serious editing programs actually use 32 bits (or even 64 bits) for internal processing to avoid all of this.
So, from the point of view of
RECORDING or
EDITING audio, there are lots of good reasons to use 24 bits. Since the ONLY real negative is the slightly larger storage space involved, it's really a no-brainer.
NOTE: If you do recording with portable recorders, the analog sections of most of them aren't nearly as good as the digital circuitry - so most of them don't give you the full advantage of 24 bit recording when you select a 24 bit format (very few, if any, have a real S/N of 144 dB). However, you almost always do get at least a significant improvement, and so it is virtually always worthwhile to use 24 bits.
Back to the topic, though.....
In a perfect world, 16 bits is probably plenty of bit depth for a recording... but, by making things easier, and making it easier to avoid issues and complications, using 24 bits makes it more likely that a given recording, produced by real fallible human beings, will avoid problems - and so will sound better. (It's sort of how a road with wider lanes will usually reduce accidents; 24 bits gives you a lot more leeway to avoid problems.) Also, by increasing that safety margin, it makes it less likely that the engineer will feel the need to ride that compressor quite as hard...
You should also note, though, that making a 16 bit file into a 24 bit file makes very good sense
IF YOU PLAN TO EDIT IT, but simply taking a 16 bit recording and "raising" it to 24 bits to listen to it (like converting a regular CD to 24 bits after it's already mastered or when you rip it) gains you
NOTHING.
Here in Nashville, I think most studios run 24 bits. Heck, our church choir recording was done with 24 bits 18 months ago. Maybe it is just local, but there is more and more 24 bit material being recorded.
Trey