I do run multiQ XT on my marantz Processor.
I've spent a lot of time learning about Audyssey, calibrating, measuring, etc. I will say this.
The only acceptable version of Audyssey is XT32. All lesser versions (including the XT I have on my Marantz) attempt to correct the "hair" in the high frequencies. This is counterproductive, because if you move your head even an inch, the high FR changes drastically. This is how high frequencies behave, and it just doesn't matter because your brain bins these frequencies together.
But XT tries to "fix" this problem, and it results in really unnatural higher frequencies (not to mention, it doesn't have the proper resolution to correct the bass frequencies either). XT32 transitions more to a parametric EQ on the higher frequencies, ignoring the "hair" as it should. XT32 also does a bang up job on the bass, as I've measured on my subwoofer.
My SS8s sound thin, brittle, terrible with Audyssey XT engaged. There is simply no way around it except to run them on Bypass with the option of crossing them over to a sub. The first comment I get from the Audyssey crowd is that I'm not running it right. I know how to run Audyssey properly, using all positions, I've re-done it several times, with/without toe in, using a boom mic stand, etc... the result is always the same.
So the artifacts in the male vocals you're hearing are likely more due to Audyssey XT than anything else. My GR Research N3S center channel sounds freaking terrible with XT engaged. I'd bypass XT at least on the mains, cross over to a sub and let the XT32 filters there manage the bass.
Just my $.02 after much research and hours upon hours of toiling trying to get XT to work properly with the SS8s. The answer is, in my opinion, XT and below are not properly designed correction software. XT32 is not just an incremental upgrade from XT, it operates fundamentally differently (and better) on the higher frequencies. -Jon