24 bit effects, 16 bit Recording

  • Thread starter Thread starter New Fool
  • Start date Start date
N

New Fool

New member
I more or less understand the dithering thing about transferring a 24 bit recording to a 16 bit CD recorder or DAT - my question is:

Since your recording e.g. on the new Fostex VF16, is going to be 16 bit, is it such a big deal that the effects are 24 bit? Presumably the problem will still remain where when you transfer your 16 bit recording to a 16 bit CD/DAT recorder, you will still lose a few bits (which is why 24 bits are better so that even if you lose a few bits you're still going to get the full 16 bits on the CD/DAT recorder.

Or am I missing the point completely?

Thanks, y'all.

A New Fool.
 
Put the link that Buck gave in your favorites immediately. There is a ton of great information there... and it probably won't make perfect sense the first couple of times through. Visit repeatedly, and re-read stuff from time to time. It all starts to come together.

I'll try to give you an extremely simplified explanation as a primer. Everything in the digital realm is numbers (thus "digital"). When you process a signal with digital effects, it does a lot of complex and fancy math on your original signal to come up with your "processed" result.

Now think of math with integers only (no decimal point). You can add 2 to 2 and easily come up with 4. As long as you're working only with other integers, no rounding problems occur. But try to add 2.5 to 2.5 in the integer world. We'd have to round the 2.5 to 3 first, add them together, and come up with 6. The 6 is an approximation, since the exact input values cannot be represented with our limitation of using only integers. The more numbers you work with, and the more rounding that has to happen, the greater the possible variance in the result. You can improve things by using using decimals during the math, and then rounding down the final result only. In the example above, you'd even get the correct result of 5.

Even though your source is 16 bits and your final product is going to be 16 bits, all the processing math that goes on in-between will benefit from more internal bits. In fact, 24-bits is less than I'd want for projects with heavy DSP. The more level changes, panning, EQ, reverb, and sonic ejaculators used on the signal, the greater the chance for rounding errors in the processing. It doesn't even take golden ears or the highest end monitoring system to hear this difference, either. A few tracks with minimal DSP won't be as big a deal, though.
 
Thanks ya'll - yes I read digidoo before posting the question - digidoo explained dithering very well, but I was kinda wondering about the processing bit - the last paragraph of pglewis' reply answers my question completely. What a wonderful messgae board! Its my first time on.

Thanks again.
 
So pglewis, how many bits is your sonic ejaculator?

-jhe
 
You must log in or register to reply here.
Back
Top