I made a device once to allow me to talk to an individual in a crowd from a
distance. I used an array of small speakers and DSPs to produce delays for
each one so that it acts like a phase array RADAR and the central beam can be
directed electronically with no motion of the array. The most difficult part is
that the signal has to be spectrally broken down and delays calculted for each
speaker AND each frequency.

This is aproblem in reinforcement. The same applies to candellation. Look up
"reciprocity callibration" for an idea of how complicated it is.

As for summing, there is nothing mysterious. If you put two engines near each
other you will sum in some places and cancel in others, but the location varies
with frequency so finding silence is probematic though there may be a "sweat
spot" if everything works out.

There is also likely to be sum and difference frequuncies and (and more if the
rsponse is nonlinear). The difference or beat frequency is the difference
between two frequencies, so if the engines are not perfectly synchonized you
will get a beat. If one is at 101 and one at 100, you get a nice loud 1 Hz beat
as any of yo uknow from synchonizing a twin in a pane or boat.

-- Charlie Springer