The only reason why compression is so confusing and complicated is because conventional compressors do their calculations based on peak or rms basis, they’re too dumb to distinguish where a note starts or ends.
Thats why when an engineer wants to compress a performance without effecting dynamic envelope, or color, or avoid artifacts, he needs to experiment with these complicated attack and release stuff.
But imagine you give the computer awareness where notes starts and ends, the computer would be able to apply compression on a note by note basis.
Now, i know thats exactly what Sound Radix did with DrumLeveler, and i understand why with percussive hits its much easier for the computer to recognize a new event.
But the technology of distinguishing less-transient material already exists, for example melodyne disguishes notes based on pitch change, and the user can even make corrections to the analysis before hitting Process.
So if we can implement such an algorithm, this could really be HUGE, and really Make America Great Again…
The only challenge i can see, is the need to implement a smoothness algorithm on the note transitions, since - unlike conventional compression - compressing a loud note next to a neighboring quiter note would make the transition too drastic, (as opposed to conventional compression where the envelope of the note is altered uniformly with everything else, so the transition stays smooth)
Thanks guys, keep me posted, and keep on the good work.