basically all we need is access to the equivalent of ue3’s GeneratePCMData() or somewhere in that region.
so that before the sound is sent to the buffers and actually played we have chance to do things with it. but the way sound cues are set up and work atm i cant really see how.
so, if going down that far is unfeasable (obviously i dont know whats going on under the hood), a very usefull alternative that springs to mind would be the ability to write our own nodes for the sound cue editor.
so we can take the raw signal coming in from the soundwave, even if its a silent wave, mess around with it (effects or whatever) and send it on its way, or for example generate waveforms in pure c++ inside the node and mix it in.
i guess custom nodes would be the best way because it would make it totaly optional, if you dont need it dont make one, and it wouldnt interfere with facefx or anything.
many many many questions have been asked on a regular basis relating to advanced (and not so advanced) audio on the udk forum, and always the answer is no its not possible. hoping with all my heart its not the same for rocket.