There is an interesting twist to the DTMF decoding puzzle that makes my
original idea unsound. Basically I hadn't included the phase of the waves in
my analysis.

In a nutshell, my idea was to sample the zero crossings of the combined
sine waves to extract a relatively unique bit pattern sequence for each tone.
This appeared clever at first but in doing the analysis it became clear that
if the phases of the two waves were allowed to move relative to one another,
there are infact quite a few bit pattern combinations that appear legally.

To make the zero crossing detector solution somewhat easier, the system
needs to first split the tones using a highpass and a lowpass filter, then the
low tones and high tones are sampled independently through a zero crossing
detector, and then *those* bit patterns are unique, caveat noise in the
system.

--Chuck