Oleg Goldshmidt wrote:
I would not be so sure that it will be random at all. Especiallyclose
to fan.
I'll clarify my suggestion.
Place the microphone. Whether a noisy environment (where there are a lot
of contributers) or a semi-quiet environment (where even small
contributers make a difference) is to be debated.
In any case, after placing the mic, start sampling. Keep playing with
the gain control so that the actual noise is at about 80% of the maximal
sampling range (or, at the very least, no more than 80% of the sampling
range). From each sample, take the least significant BIT.
Rational:
While an environment has, obviously, non-random noises, the affect of a
loud noise over the LSb is no more than that of a quiet noise. This
reasoning needs to be tested against two cases:
1. A VERY loud noise would cause us to lower the gain, thus reducing the
affect of the non-loud noise. This is not a problem for us, so long as
the non-loud noise still has ANY affect.
2. We must never reach the microphone's saturation point. This is the
only scenario I can think of where the LSb will lose randomness. I
suggested playing with the gain control to achieve that, but it may just
as well be ok to discard long sequences (say, greater than two) where
the sample remains the same (the entire sample, not just the LSb). This
means that placing the microphone in a noisy environment produces less
random bits per unit of time, but does not alter the actual quality of
the random numbers produced.
Something to test - how much "noise" do you get from the sound card's
sampler when no microphone is connected at all? Can that be used as a
random source?
Shachar
--
Shachar Shemesh
Lingnu Open Source Consulting ltd.
Have you backed up today's work? http://www.lingnu.com/backup.html
=================================================================
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]
