Its a good question perhaps something that only few can answer or want to answer. I am also thinking about a way how this could work out. But as long as someone not uses headphones it could always end up in a feedback loop. And also something like krisp implemented in Unreal would be good too. Filtering out everything out except the players voice is something I would like to hear in the future.
I am bumping this thread therefore too or come back to it when I found something that might solve or help it.
Have a good one!