Well, I'm no programmer but isn't it a matter of thresholds, envelopes and logic statements?
Say a source has an average loudness of a certain value above a set level. Also a certain stereo loudness balance over time, and one side drops by half the loudness but material resembles the loud side, stereo balance is off.
Logic at that point tries to raise up to desired level, and if no large peaks, it holds it there. When drop ends and all volume comes back the logic releases the volume back to before drop, and stereo pressure evens out.
I can see that would be a problem on a lot of music material, but standard mastered pop would be possible to predict to some extent.
There is also the easily detected s/n ratio to measure on.
I'm not aware of anyone ever attempting to program this.
|