[Don Pearce]

On Tue, 27 Oct 2009 12:35:04 -0700, (Dave Platt)
wrote:

In article ,
Don Pearce wrote:

So using my original construct ([S + N] - N = S), there's no way that
using XOR instead of subtraction is going to get you there. The
nym-shifter just seems to have XOR on the brain for some reason. (Quite
possibly XOR *does* work in a similar circumstance in video work: I
don't know.)

Wouldn't have thought so. XOR is not a subtraction process in any
discipline that I am aware of. Unless he has redefined XOR, I suppose.
There is a lot of that sort of thing going on round here right about
now.

XOR is equivalent to subtraction (and addition!) in a modulo-2 number
field.

Useful in many cryptographic operations.

Not at all useful in audio, at least not for the original poster's
stated purpose of removing noise. Even if the audio is being
represented in the commonest numerical form (e.g. linear PCM, two's
complement), even the slightest amount of error in measuring or
quantizing the noise will usually result in *multiple* bits being
different between the "real" noise and the "noise to be subtracted".
Trying to XOR the two values together will thus result in many bits
being different between the "original S" and the "computed S", and the
noise level in the signal will probably increase almost as often as it
decreases.

We aren't in a modulo 2 number field, and I fear you may have muddied
the water needlessly here.

XOR is a process independent of the order in which the numbers are
presented. 4 XOR 7 is identical to 7 XOR 4. The same is clearly not
true of subtraction.

d