Then you probably also read that multi-band processing not only requires a lot of components and meticulous matching but that splitting bands and summing them again can result in phase issues.
Just as an idea, maybe use a parallel subtractive approach? Needs three channels on a mixer:
- (1) material to mixer (mute 'on' at first)
- (2) material into EQ into comp into mixer
- (3) output from EQ also straight into mixer, but polarity flipped and level-matched with (2)
Send 'material' into an EQ and 'isolate' (boost, HP, LP) the offensive frequency band. Take that and send it thru a compressor at unity gain. Then 'sum' the output of the compressor with a signal straight from the EQ output but flip the polarity of the EQ line in the summing. Thus, EQ output and comp output are in parallel. If levels are matched and everything else is perfect, the two signals should cancel out. Then start attenuation on the comp. You should start hearing the 'difference' between the two signals, namely the part that gets attenuated down by the compressor (if it was sibilance, then just the isolated 'sss' sound). However, that 'difference' signal is polarity flipped, which is the basic idea, because if you finally 'sum' that with the (now unmuted) 'material', it will subtract or cancel out in the final sum.
I guess there will be phase issues too, but should work somewhat nicely in a DAW with plug-in latency compensation.
But don't take my word for anything, the above might be wrong -- it's Tokyo and we had degrees of 35C indoors today, now down to 32C---
頑張ってね