44, 48, 96, 192 Kz sample rates...

[nst7]

What might be a good happy medium is to track simple stuff with only a few tracks (like singer-songwriters, basic acoustics stuff, etc.) at the higher rates, and at 44k for stuff with lots of tracks.

[obiwan177]

Is it possible to record everything at a lower sample rate, like 44 or so, and then record the vocals at 96 (because they are the most noticeable and usually are tuned) ? It doesn't seem like it should work, but if it did that would be amazing.

[necjamc]

I'm not sure that would be the best idea, but one way to track at lower sample rates for instruments than higher for vocals would be to save your session at 44.1, then start a new session, at 48, 96 or what ever and then use session import to upconvert the audio from the other tracks into a new session. But of course then you would have a high sample rate tracks and the originals taking up space. Just a thought

[obiwan177]

Or you could record the instrumentation at 44k, than import them into a 96k session, record the vocals in that session at 96k. Then put the vocals through melodyne, and after they're tuned import them into the first (44k) session. A little bit more work.

[Majorek]

Thanks folks for your inputs. Very interesting to read!

[O.G. Killa]

Originally Posted by necjamc

bouncing to 96KHz will also give you better headroom for mastering effects if you are mastering it. If you are just bouncing to play back then think of it this way, DVD is 24bit 48KHz, cd's are 44.1 so you don't really need anything over 48k IMO for just playback.

Bouncing at 96KHz doesn't give you any more headroom. BIT DEPTH controls "headroom"/dynamic range. Recording at 24bit gives you more headroom than 16bit. Recording at 48bit gives you more headroom than 24bit, and so on.

[O.G. Killa]

Originally Posted by obiwan177

Is it possible to record everything at a lower sample rate, like 44 or so, and then record the vocals at 96 (because they are the most noticeable and usually are tuned) ? It doesn't seem like it should work, but if it did that would be amazing.

Why bother? Just record at 96KHz if you feel you need to. It's not going to "hurt" the other tracks. And the tracks aren't going to "sound" better at 44.1KHz anyway... so there is no benefit to jumping back and forth. You are better off just recording at 96KHz to begin with.

[necjamc]

Originally Posted by O.G. Killa

Bouncing at 96KHz doesn't give you any more headroom. BIT DEPTH controls "headroom"/dynamic range. Recording at 24bit gives you more headroom than 16bit. Recording at 48bit gives you more headroom than 24bit, and so on.

Ya your right. But I've read articles pertaining to using 96KHz to gain ?? something. But now I can't find the article. Something along the lines that 96KHz giving more room for effects etc before clipping. Don't quote me on that I'll have to find the article. I believe it was in Sound on Sound, but I'm unsure. Now I have to find it, It was very informative.

[necjamc]

This was interesting I thought,

Originally Posted by tbkbt

This is something we have spent the last 15 weeks or so learning about in college in our Digital audio theory class.

*here comes the science lesson*

I personally use 44.1 an 24 bit, only because I was once going a session someone else had set up and I just thought it sound better than anything I had ever done before (44.1 16 bit) and I don't think it was the performer, had been recording different versions of the song for 5 weeks before this happened.

I can understand why you should in theory record at 96K it’s basically the nyquist theory dictates that the highest frequency that can be captured is exactly half the sampling rate (in this case 48K). Since all signals are attenuated from 20 KHz in AD conversion the cut off slope is much more gradual than at 44.1 or 48K, the reason this is good because the more of a brick-wall the filter is then the more resonance around the filter occurs. Therefore there is less resonance around this cut off frequency. also when you are recording at higher sample rates such as 96K then quantise noise created when a sample doesn’t land directly on a step level is spread right over 48KHz(half of a 96K sample rate) so we only hear a small percentage of the noise so it seems less to us.

We have also have learned about oversampling and I can throw that into debate if you like but I’ll leave that for another day.

if you want me to explain anything more about this, or I’ve probably just rattled this off due to my class in it I’ve probably missed something out just ask me to explain whatever gets you confused (I still am after this long of learning it) also I’ve typed this at 11 o’clock my time so I might have a few things wrong.

So as I said I personally use 44.1 and 24 bit just because it sound good to me an let’s face it, who these days listens to their music in a perfect listening environment, it’s all MP3 through crappy iPod headphones unfortunately, after its been overly compressed to hell to get it loud.
Hope this helps the debate slightly, sorry for the long post btw, well done if you read this far.

tbkbt

P.S. I can scan in notes for anyone who wants them that we learn from

[O.G. Killa]

one problem I have with that post is it is assuming that quantization noise only happens at the nyquist frequency. but it doesn't. It is broadband as far as I know. If it were only at nyquist, a simple filter would be enough to stop quantization noise. But that is not the case. Dither is needed to stop quantization noise.