It'll be an indoor concert (A Cappella... approx 15-20 various singers). I'll be taping the show with 3 V1U cameras, and I'll have two mics on the sides but a Sony D50 in the middle. I'll be doing a test during one of their practices, so I'm not sure exactly where I'll set it up, but it'll actually be pretty close to the stage (maybe 6-10 feet or so, on a stand that I can elevate pretty close to voice level). I know it's a Newbie question... I was deciding between the Zoom H2 and the Sony D50, and while the D50 is a bit beyond me at the time, I figured I might as well go for it. Everything I'm reading seems to indicate that there really wouldn't be a benefit to set the D50 to 24/96.

Upping the sampling rate to 96KHz does indeed probably gain you nothing. Upping the bitdepth to 24bit can gain you quite a bit of head room, at least in theory. So we always record 24/48KHz. The added cost in file size imposes no penalty worth worrying about and any headroom we may or may not gain is certainly usefull. It means we can set the gain so our peaks are only hitting -12dB and bring it back up in post i.e. we know we'll never clip during recording so setting up the audio recorder is a set and forget affair.

We "might' be fooling ourselves though. In theory 16bit gives you quite a high dynamic range and the extra 8 bits adds a lot more. However if your noise floor is being determined by the preamps and not the sampling them all you're really doing is recorded really clean noise!
However most of what we record is coming in at line level, usually from decent consoles so the mic pres in our recorders aren't in that equation.

Look at another way. Going from 48KHz to 96KHz doubles your file size and makes your NLE have to resample (you should set the Resample Quality in your project to Best if doing this) gobbling up CPU cycles. On the other hand going to 24bit only adds 50% to the file size and your NLE doesn't have to resample.

Bob.

Unless you're really doing a lot of temporal editing, such as timestretching, using 96 doesn't give you enough benefit outside a proper studio environment, and even inside it's a "more theoretical than evidential" argument.

24bit - definitely a good step. 48Khz is plenty for 99.99% of the time. (unless you really need to be able to recreate frequencies way outside the human hearing range?)

According to the SEP's (super ear people on the Klipsch forum) 88-96 is about optimum and only SEP's can tell the difference and only on very high end systems.
JJK

I use 24 bit audio regularly when I'm working with music: mainly so that I can normalize later and dither back down to 16 bits for a true 16 bit dynamic range.

There are occasional uses for higher frequency responses while recording. One of these would be when you are recording a sound effect that you are going to slow down later. In this case it is kind of like overcranking video. You want to slow something down and so you capture really fast. This is great for creating monster sound effects or making things sound really massive.

Another use for high frequency responses is if you are working with instrument samples and want to transpose the pitch upward without having the high frequencies alias back down into the audible range.

These days I always master music at 24bits with little or no compression, then add compression as I need it for a given project. CD Architect is absolutely wonderful for printing 16 bit 44.1k CDs from 24 bit 48k masters.

In all these cases, I would still deliver the final product at 16bits and either 44.1k or 48k. There is simply no reason to waste bits on things that the final consumer can't hear anyway.

There's not too many options to deliver at anything other than 16bits at 44.1 or 48Khz is there?
SACD and DVD A is out of the reach of most of us.
One of the things I've recently been made aware of is the quality of the A>D converters we use. The top shelf ones cost more than my EX1.

Bob.

For the live recordings I've done at 48, the background noise level has been high enough to obviate the need for a greater bit depth, or so I've always assumed. It would be nice to record at 24 with a default level that guaranteed never having to worry about clipping.

Oh, see... I was all set until that last comment. Clipped audio from a concert is one of my biggest fears. I always do a trial, catch as many variables as I can, but I've had 1 or 2 horror stories of taping concerts with clipped audio when I get back to edit. How would recording at 24/96 change the clipping issue vs recording at 24/48?

>> How would recording at 24/96 change the clipping issue vs recording at 24/48?

No difference... it's the 24bit part which gives you the dynamic range.
When you do the trial, set levels so the highest peaks on your meters are at least 10dB off maximum - then you should be safe.

Ahah. Thanks for the clarification.