[PD] Box Muller Gaussian noise

[Charles Henry]

One thing I've been thinking of lately for pd is to synthesize a
impulse train with the poisson process.  Many natural phenomena, like
the timing of rain drops, can be simulated with the poisson process.
It's just one of those ubitquitous relations that pops up everywhere.

A poisson process models a chain of events that occur with a density
with respect to time.  Each of the inter-event times is an exponential
variable, which is easily calculatable, just as in the previous
example.
Take U1 a uniform variable on (0,1] , and calculate ln(U1).  You've
simulated a draw (observed value) from the exponential distribution,
just like that.
So, it's easy to make a message chain which does this calculation and
outputs a series of bangs.  What confuses me is whether it will be
accurate in timing, and how to turn it into a series of impulses.  I'm
not sure I know how to do the rest of this problem.

My interest in making an impulse train is to make the impulses, which
will have subtle deviations about a central frequency, and process
them through a filter (formant).

I wonder what the spectrum of the poisson process looks like...

Chuck