Doesn't sound to me like the issue is frequency related, but levels of the music relative to levels of the crowd. This is strictly an amplitude - not frequency - issue. At least as I understand it.
Not sure how to do it in SF, but in CEP compression lives in the main menu Amplitude > Dynamic Processing. Play around with different settings. I apply compression either to each individual audience peak between songs, or just across the entire recording in one fell swoop.
The compression I've applied to NC shows in the past has been to each individual audience peak. Takes a bit more time and effort, but I prefer the results. I usually compress the audience at a high ratio, something like 10:1 at X dB, where X is some level quieter than the music. But sometimes a high ratio will squish the applause to the point it sounds like horseshit. S sometimes I set the dB level lower and use a lower ratio - this won't compress the applause as far, generally, but it sounds softer on the ears. Either way, I'm not especially concerned with the dynamic range of the applause.
And I try to use a value for X that I can apply to *all* audience peaks, instead of compressing different peaks at different levels. Just makes the listening experience smoother, IMO. I always make sure there's no music playing where I'm applying the compression, though, or if the music *is* playing, it's below X, my compression threshold.
Dunno if that makes sense, but...play around with the compression ratio and the dB threshold and you'll get the hang of it. FWIW, if I end up compressing a show (pretty rare), I usually compress and then apply gain to the whole recording instead of doing it in 1 step. CEP and likely SF will let you do both in the same step but for some reason I don't.
So, for example, if my applause peaks around -2dB and my music around -10dB, I would:
[1] apply compression across the whole show, say 10:1 at -10dB. This would bring the crowd levels down close to -10dB but would not actually compress the music at all. Then I'd amplify the whole thing and the crowd wouldn't be as loud relative to the music.
or
[2] apply compression to each individual audience peak. If the crowd is attentive and doesn't bust out into loud applause while the music is still playing, I'd set the ratio to 10:1 or so, but set my threshold at -16 dB. And again, finally amplifying the whole thing. This would reduce the applause so they're actually *quieter* than the music, but takes more effort and can't always be done smoothly.
My personal philosophy is if I can't apply the compression smoothly so it isn't grossly noticable to the listener - no drastic drops or noticable fluctuations in levels that sound unnatural - I won't bother, I'll just leave it alone. Of course, best option is always to have the levels peaking well during the recording so it doesn't require tweaking afterwards.
Play around with the settings a lot, and remember - far more recordings have been screwed up by poor editing than fixed. So take your time and listen critically each time you make a change.
Oh! And don't forget to keep a backup of the unedited master!