Hum, that is going to be a tough one. The problem is as you stated that the voices are near the same frequency as the noise in the room. Your traditional notch filtering isn't going to help you much here.
Without going into the long lecture about how you should have used correct micing techniques here is one long shot possible attempt. I know that this will not be a magic bullet but here it goes.
Hopefully at some
point you either recorded "room tone" or a long enough segment of ambient sound without anyone talking. This "room tone" is often used to fill in the gaps where editing has created acoustical "dead" spots in the soundtrack. Take this room tone and try to get it into some sort of seamless loop. Now duplicate it until you have enough to
cover your bad audio. Invert it 180 degrees out of
phase and see if the noise will cancel itself out. You may have to slide the room tone
track in and out of sync with the video to help with the
effect.
It will thin out the primary audio some but I have used this trick to some extent between an on-camera mic and a lav or
boom with mixed results in post. Now if someone starts preaching about the time delay between the
boom and the on-camera mic not being exactly 180 degrees I am going to smack them. This is a rough solution here people.
kw
---edit---
You may have to go back to the original camera footage because I don't know what STPro did to the waveform for it's noise reduction.