Ok i'm sure this topic has been talked about a lot, but being new to digital music, i would still like to get some more views on it.
(Bit depth sample rates) if recording audio at 24bit/96k, but the final burn is at 16bit/44.1 (due to red book standard media), would not the quality in the 24bit/96k recording be lost??
Something ive often wondered. I would have thought upping the sample quality to 24bit would introduce the possibility of artifacts.....
Problem for me is that my deliverables all need to be 24bit/48Hz....anyone know what sample rate and bit depth the new sample libraries for ARIA will be? and if the GPO upgrade will be upgraded to 24/48 too?
If you up to a 24-bit sequencer file, you're just adding a bunch of zeros. No need to worry there. As for sample rate conversion, I guess you might from interpolation, but any good DAW will make them negligible.
I've done a lot of experiments in my studio concerning this question when it eventually goes down to 16/44.1 anyway.
A higher bit depth and sample rate DEFINITELY sounds better, even after you go down to CD quality. I'd say 24 bit makes a larger difference than a higher sample rate, to my ears. Even so, I record at 24/96. Some engineers record at 24/88.2 because of some mathematical reason which I don't quite understand. Explanation for this, anyone?
Higher bit depth means lower noise floor. So all of that gain noise that you can get if you turn up speakers and mics and sources really high (or buzz/hiss from equipment) gets lower in the signal. If you start summing a whole bunch of tracks the mix gets tighter. The noise floor also rises. With higher bit depths there's more space for them to all fit.
Higher sample rate means that (originally and through processing at least) you don't get the rolloff in the high frequencies as a result of analog to digital conversion. Sample rates can carry information about frequencies up to 1/2 their value. Around that point (termed the Nyquist frequency), the information drops off on a slope. As the range of the human ear goes up to about 20Khz (and deteriorates as we age), it was decided that 44.1Khz would be able to accommodate the range of hearing, so that's where we see CDs. DVDs use 48kHz, and DVD-Audio supports 96kHz. The logic is that tracking at 96Khz preserves the full spectrum of hearing unaltered, and that then the higher audible frequencies don't suffer further deterioration in processing.
Note: Your mic in the first place is going to affect the frequency response of the sound. It's never going to be even in the first place!
When converting sample rates, algorithms will use interpolation between points to construct the new sample rate. As 88.2 is an integral multiple of 44.1, I would imagine converting downwards does not require interpolation but rather just dropping every other value, whereas going from 96 requires the computer to make guesses as to where middle points would be.
Granted with samples, they've already been recorded, so you won't gain anything from recording higher, and SRC will be performed on the fly. For sample-based recording, I recommend using the dominant sample rate of the virtual instruments and running SRC (to, say 48KHz from 44.1) as a last step manually. And definitely running at 24-bit if you have the capability. I'll let DPDan take the dithering argument.
Studies have shown that people, even audiophiles, can't reliably tell the difference between 96 and 192KHz (which you'd expect since people don't hear at 40Khz). And of course, in playback, there's a digital-analog conversion happening which further 'distorts' your sound.