quote:

Originally posted by DJ RANN Absolutely, but to clarify this further, even if your system runs at 32 bit, you will not be able to use really any benefit off that as your system is only as good as the weakest link in the chain. In nearly everyone's case this is the soundcard, which as diginut rightly points out is 24bit. So basically, to maintain the highest quality while producing your host needs to be at 24bit and so does your all your equipment. Any higher than that is quite pointless as you can't take advantage of it at any stage of the production process. Therfore going lower than that is not maximising your available quality.

Ehhh, I understand what you're trying to say but you're using an oversimplified model and therefore coming to the wrong conclusion.

There's no advantage to trying to push above 24 bits at either the input or output stage, but there is a very good reason for using higher precision in intermediate stages, specifically, intermediate processing and effects.

Think about it. Every line item on your credit card has exactly 2 decimal places (cents). But if they rounded everything to 2 sig figs, if they calculated your interest this way for example, then your interest would be zero, the system wouldn't work. Everything that comes into the system has 2 decimal digits, and everything that goes out of the system has 2 decimal digits, but that doesn't mean it makes sense to limit everything inside the system to 2 decimal digits.

Multiply 1.2 by 1.3 and you get 1.56. Look, we needed an extra digit of precision to store the result. Multiply again by 1.1 and you get 1.716, which rounds to just 1.7. But if you round the intermediate result, you have 1.6 * 1.1 = 1.76, which rounds to 1.8. By limiting your internal precision to the "weakest link", you've just created a clear error.

If you're just going to record a bunch of tracks, mix them together, and toss them back out without actually doing anything to them, then sure, the sequencer itself could just as well use 24 bits internally. But that's almost never what happens; every EQ, every reverb, every compressor would degrade the signal just a little bit at fixed-point precision, and after 50 of these you'd start to get real audible artifacts. Using 32-bit float for all internal computations, you get almost none of that. You lose a little bit of quality at the very end, converting back to 24-bit fixed point, but with a decent dither it's impossible to hear.

Your host does not need to be at 24-bit for the best quality, and in fact it isn't at 24-bit, it's at 32-bit. It is not pointless, and you couldn't change it even if it was!

___________________
My party schedule:
2009-02-21 - DJ Attention @ I'm So Popular
2009-06-18 - DJ Annoying @ People Need To Know Where I'll Be
2012-11-32 - DJ Insufferable ɸ Or At Least the Stalkers I Complain About
2048-06-66 - Spastic & Whocares ¶ Although I'm Actually Flattered
9999-45-81 - Tweaker Gimp ☼ I Probably Won't Even Go To This But I Have To Make Sure I Fill Up All The Available Space Here