Thoughts on the following technique:
1. Gate the source so that only the signals above threshold remain (the opposite of what we want to add or compress).
2. Reverse the polarity of this signal and sum with the source track. This should cancel out the signal above the threshold leaving what was removed during gating (low level signal).
3. Compress (or not) the signal from step 2 and sum with the source track. Without compression the original attack is maintained and just the level is raised by the amount of the signal added.
All of this assumes that everything lines up properly so latency compensation in Pro Tools needs to be turned on (and working correctly). Also gating can be a bit tricky, it's not going to be sample accurate like DSP can perform, but that might be a benefit in a way if it isn't chattering since you can control the envelope of the low level signal. Replacing the gate with a downward expander might also help to smooth things out and give a more natural sound.
This is very similar to what I suggested in my last post
As for latency compensation, if you have Pro Tools HD, you can use the
delay compensation feature. However, for LE users like myself, there's no such
thing, so the latency issue could screw it up. This is why I suggested strip
silence. It doesn't work in real time, so it has no latency issues. However, if
you wanted to use the gate route, you could solo the track, apply a gate
(real time), fine tune the parameters, and then use Audiosuite (non-real time
processing) to render the track with the gating effects. This should eliminate
the latency issue.
The only problem I'd see with using phase reverse to remove the peaks would
be if the gate altered the sound in any way. It won't completely cancel. I've
done tests where I've loaded 2 different compressors onto the same track,
and used the exact same parameters, and phase reversed them. They barely
cancelled (and they shouldn't, otherwise we wouldn't need different
compressor plug-ins).
This is why I suggested strip silence. It can be used to remove audio above
a threshold, and audio below a threshold, enabling you to separate the two,
and apply processing to each individually. But because it isn't using any
"processing" to separate the signals, and it's purely editing, there should be
no sonic alteration to the audio. I'm going to give this a shot
), I believe you can. Just set up the dynamic range zone you want to affect, set the ratio to 1:1, and then apply the L & R output gains as desired. This will provide a gain control without compression/expansion for just that slice of the dynamic pie. This is one additional option for those times when something like the piano decay envelope doesn't want to be messed with.
