I’m automating vocal processing (EQ>Compressor>Deesser) using the Axe Fx that sends the audio to a Camera so I don’t have to edit the audio in post
The crossover frequency in the multiband compressor block limits you to having to use double the frequency value if you try to compress the 5khz to 7khz range
The mid crossover freq I choose is 5khz then the high crossover frequency range automatically changes to 10khz , if I lower the high crossover freq then the mid crossover freq lowers with it
TL: DR Basically just need to compress only a certain range of frequencies , in my case 5khz to 7khz
Sorry for the delay in replying. – super busy day.
What you want to consider doing is what us Old Sound Guys refer to as parallel (or "split") processing – i.e. set up two (or more) separate processing "pathways" for a single source (so you can deal with them as separate signals as required) and then mix them back together. In your case (wanting a de-esser that operates in the 5-7kHz range on a voice source) you need to take your vocal source and split it into two "rows" on your AFX – let's call them "Path A" and "Path B". Once you've set these up, you'll want to:
1) Use your EQ block of choice (mine would be PEQ parametric) to reduce the amount of the offending frequency band (as you've already identified, 5-7kHz) in Path A. I'm guessing about 12-15dB of cut should be sufficient, maybe even less. You don't need to completely get rid of it, just reduce the 5-7kHz down to a level where the peaks are well below being offensive. This will mean, of course, that the overall Path A signal is now too "dark" to be used by itself as-is.
2) Use another instance of PEQ block on Path B, but set it so that it reduces everything EXCEPT the 5-7kHz range you're reducing in Path A. This time I'd set up a lo-cut shelf at 5kHz and a hi-cut shelf at 7kHz – your results may vary.
3) Put a compressor/limiter processing block after the EQ on Path B and set if for fairly fast attack/release (since it's dealing only with HF) and adjust the threshold as needed to even out the response of the 5-7kHz range.
4) Mix Path A and Path B back together, adjusting their levels to get the amount of HF you wish – only now without the signal peaks that you want to get rid of. Always keep in mind that you can fiddle about with all of the settings while listening to either or both of the Paths, as well as monkeying around with the relative levels of same.
5) Obviously, using Snapshots so you can compare differently settings of all of this can be pretty useful to determine the best overall result. I'd also be inclined to play around with using the different types of filters in the PEQ block: "BLOCK" would be tempting, but the "brick-wall" characteristics might be a little harsh, while the SHELVING/PEAKING filters and their wider bandwidths might sound better. Do share and let us know what you find works best.
Voila! You've just created a (programmable, no less!) de-esser the same way we used to do it in the days before plug-ins when we didn't have a dedicated de-esser in the rack but did have an extra comp/limiter and the board's channel EQs. As an aside, we'd typically use something like an 1176 with a pretty stiff ratio (20:1) on the HF side of the chain and a dbx160 "vca" compressor (set with low ratio and a low threshold so it's pretty much "on" all the time) on the post-mix signal.
FWIW, this same concept of channel splits can be used to get a powerful tight drum sound (via strong-ish overall compression to create a solid "base" mix level) without losing the dynamics and "life" that more lightly-compressed parallel channel processing can provide.
Hope this helps! Now that I've said all this, I would strongly encourage you to solve the problem where it's starting – mic selection and placement are the usual culprits.