[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: streams and groups
Regarding the question about the transition from claps to
applause: one limit is caused simply by the fact that the sum of many
transients converges to a gaussian process (white noise if the
transients are wideband). The signal-to-noise ratio of each individual
component becomes progressively worse as you become more famous, until
it is essentially masked by the background composed of all other claps.
Thus, beyond a certain limit, it is essentially impossible to infer the
properties of the individual transient from the summed response.
I once made an experiment of summing sequentially more and more
chirps in order to see when the statistics would become
indistinguishable from gaussian, and it turned out that for my criterion
of indistinguishability, about 5 simultaneously singing birds would be
Translated to claps, assuming that a clap is let's say about 50
long, assuming a rate of claps of 2/s, it means that you need about 50
people simultaneously clapping at uncorrelated times to go from
individual claps into an applause. If you take into account some kind of
cocktail party effect, you should multiply this number by the effective
angular resolution of the CPE, which I guess is about 60 degrees, so in
front of an audience occupying about 180 degrees you would need 150
people. Because of statistical variance, I guess these numbers should be
multiplied by 2-3 to give a margin for fluctuations. This could be
further elaborated, but I guess I'm already overdoing it :-).
Any comments from people experienced at talking in front of an
audience about the validity of these numbers?
Israel Nelken | Tel: Int-972-2-6757087
Dept. of Physiology, | Fax: Int-972-2-6439736
Hebrew University - |
Hadassah Medical School | Note change in email address!
P.O.Box 12272, Jerusalem 91120 | Email:email@example.com