Using phase to minimize two frequency signals' combined amplitude...without harming audibility?

Question

I am working on an open-source audio compression program and am trying to find a way to efficiently minimize the amplitude of a wave file, both for compression purposes and to make sounds modified with it put less stress on speakers.

Here's my mathematical dillemma:

Supposed you have two frequencies that last for about 1/10th of a second (IE a 2048 sample FFT window)...and you want to modify every other frequency to be rotated a certain amount of phase from the last frequency.
You are given first frequency is f1, second (f2) and magnitudes m1 and m2 to form an optimization equation that defines what phase relative phase angle (-180 to 180 degrees) would each phase START at relative to the other to minimize the peak amplitude of the combined signals.
As a precondition, though, the phasing should not reduce audibility of each frequency (IE no significant phase cancellation...at least to the point the human ear can hear both frequencies clearly)

Answer 1

I think that "trial and error" will get you there faster.

ie make multiple equation models and TEST them with your ear.

If you are dealing with "real-time" sound samples, "phase" becomes "delay" at the wavelength level. I don't think that mathematical treatment will satisfy these requirements more
effictively and certainly not any easier.

"a priori .. a posteriori" each has its place

Answer 2

any barrier in front of sound will and has always melded with such barrierto forever alter such wave .the key is to have a complete knowledge as to the composit of all players concerning the creation anew.to completely know the origin of your sound would be improbable,if you can successfully label one sound you can labelthem all r u willing to put forth action.keep thinking thanxPET