44.1 versus 48KHz sample rates

Question

Since manufactured CDs are 16-bit, 44.1 KHz, is it better to record tracks in that format? It looks like STP defaults to 24-bit. Is it better to use the higher sample rate as well? 48KHz?, or should it be even higher than that?

Answer 1

Hi Bob. Let me see if I can remember what I learned way back in my college physics class. Keep in mind there are almost as many ill-informed opinions as there are people who record digital audio. (Oh! Dig!)
Some of that stuff sounds silly to me. Generally it's wise to use the best quality that's available to you unless you have space or bandwidth constraints. Saying you should do all your recording at 16/44.1 because it's ultimately going to end up on a CD is like saying you should shoot all your digital camera pictures at 320x200 because it's ultimately going to end up on a web page. Shooting pictures at higher resolution allows you more flexibility when editing them down for the web site, right? And audio is no different.
Here's some basics, explained in pictures:
http://www.musiciansfriend.com/document?doc_id=88273&g=home&src=3SOSWXXA
You can find more stuff like that (and endless hand waving and opinons) if you google "Bit Depth Sample Rate Physics".
Bit Depth:
Imagine in some simple world your recording software could sample waves that had amplitudes between -1 and 1. With 16 bit samples, you can record 65,536 discrete levels. (You might define a sample value of 0 to be -1 and 65,535 to be 1.) At 24 bit, you can record 16,777,216 discrete levels. The resulting representation of the wave you record will be far more accurate.
Sample Rate:
This comes down to Nyquist Frequency. The idea is that you can only record frequencies up to exactly half of your sample rate. That doesn't mean you can get a good or accurate recording of those highest frequencies, though. The higher your sample rate, the more accurate picture you can get of those higher frequencies.
This is important for recording and playback, but it's even MORE important when you later go to combine this recording with other recordings. Mixing signals, the mathematical results of processing with effects, etc. All those things will give you better results if you give them higher resolution going in.
Combining 16 low resolution tracks will give you a much worse result that combining 16 high resolution tracks and then down-converting the result. The combination of the 16 high resolution tracks will be a much more accurate representation, right?
That said, there are certainly diminishing returns. 24 bit/96kHz can give you great results, but will take up more disk space and processing bandwidth than 24/48. If you're not using superb mics and preamps, the improvement might not justify the difference. You might consider trying some experiments to see if you can detect differences yourself.