Actually, in this case the pre limiter is already functioning roughly as a 1-band filter: If the lows get dropped, I'm taking down the highs as well to avoid 'shrill' sounds.
If I read what this "gating" does I don't see how that would help in this case? Basically, there's a very loud tone at around 400-1000 Hz which takes down the rest of the sound. Since it's a very loud sound I need to do
something there.
I was just googling a bit and I found this:
if the audio falls below the threshold the compression won't try to increase the gain until it's above that point again.
So, if there's little sound, I should do nothing (not increase the volume at all)... Well, if the input is relatively constant already that might indeed improve things (after silence, the volume won't have increased to the maximum so I don't have to drop it very rapidly - and that should also help in songs with short bursts of sound with silence in between) - but I don't see how that solves the problem with the voice which causes a drop in lower frequencies?
Bojcha, if possible, could you try to describe what you expect to happen in this case? If I understand what you want I can probably implement it... But right now I don't see how this is going to help. (For the problem that you describe, that is. Gating seems to be very useful and I'll definitely put it on my todo list. Did I ever mention that I'm not an audio processing expert and that I only wrote Stereo Tool to get some experience in using Visual C++

?)
(Edit: Gating + AGC have been added to the todo list. Fortunately now the CPU load is so much lower I have room for extensions...)