After playing around with the original mic (that failed) I just couldn’t make it reliable. I wound up finding an alternate option, and it yields some pretty good results. Shake test not show here, due to sheer ridiculousness.
Just an example compilation of the B Roll footage as collected.
This is the original video, with no audio processing. It looks good… but sounds like it’s in a cave.
This is with metric shitload of audio processing done. It doesn’t sounds much better, but that’s the nature of broad spectrum audio pollution from machinery.