16 vs 24-bitI'm not 100% onboard with the notion that recording in 24-bit is, in absolute terms, any sort of real advantage to me. At least not when recording rock music.
Reading about it on the Interwebs there's a lot of BS. The place I'd go for low BS is Sound Devices. And they have a little article on the real audible differences between 16 and 24-bit recordings.
My issue with just doing everything at 24-bit is that it uses up a lot of power and data when I'm recording. 24-bit recordings are by their very nature less susceptible to precision when setting levels. In other words, if you're recording a bit on the low side you can boost up the levels in mixing later on and not have problems with the digital garbage that would get louder when boosting.
In the world I live in it's the analog stuff that's noisy. Let's face it, I record a guitar with P90 pickups. There's plenty of noise to go around.
For recording dialog in films: sounds like 24-bit is the way to go. Because who knows what level some
For classical music: 24-bit 96 (or 88.2)kHz. Because who knows how quiet that clarinet solo is going to be? And since you never know how freaking loud that part in the middle might be you're always going to set your gains 8 to 10dB lower than the loudest rehearsal. Because you know it's going to get louder than that.
For rock music: 16-bit. Why? Because I'm doing multi-track recording and any machine I drag around simply does not have the power to handle even six tracks of 24-bit 96kHz recording without ASIO buffers dropping.
AddendumHere's the thing. Back in the olden days when dinosaurs walked the earth [Editor's note: the early 1990's] some A/D converters claiming to be 16-bit were, in fact, only 13 (or even 12) bit. The big difference with Apogee converters is that they were actually 16-bit. So they sounded vastly better (they still do). Another thing is that converters used to be really terrible about only being linear at the upper part of their range. As things got quiet the converters got wonky.
So there's a possibility that the same converter at 24 bit sounds better in the upper 16 bits of its range than when it's just set to 16 bits. I'm not, however, hearing anybody on the Internet saying that. And if it's not on the Internet, it must not be true.
Also note that when we're in post-production and/or mixing we're working at 64-bit floating-point, dual-overhead cam, fuel-injected steppin-out over the line. Yes, we need all kinds of high-bit-depth for anything we do to those recordings after we've actually recorded them. And all mixes are delivered at 24-bit (except for film, but we do pre-mix at 24-bit in film).
48 vs 96kHz
The next big issue is what sampling rate to record at. Boy my music life would be easier if I could record at 48kHz. But I think, feel, and in part believe, that 96kHz sounds "better". We all know now that the reason oversampling rates sound better is likely due to the analog portion of the A/D conversion process (right?) We're not actually hearing stuff up at 48kHz but rather we're hearing the anti-aliasing filters which are phase-shifting down where we can hear the difference.
For recording dialog in films: 48kHz is such a standard across the board we're just going to go with 48k. I don't see this changing any time soon. I can't even deliver my movies in 24-bit, I'm certainly not going to change the sample rate.
For classical music: 96kHz* because more = better.
For rock music: I believe that all the converters I use, even the cheap ones, sound subtly better at 96kHz. By "subtly" I mean "enough to make a difference here".
AddendumNow it may very well be (from what I've read on the Internets) that 60kHz is the "ideal" sampling rate and that above 60kHz is just nonsense. But I can't set any of my converters to 60kHz so that's a moot point.
It also may be that cheaper converters start to sound much more expensive at high sample rates. That's just me making stuff up though, as I'm wont to do.
*For purposes of this discussion, 96kHz and 88.2kHz are the same. I'm not ready to go into the notion of which is better for which purposes yet.