I recall a recent thread where folks discussed computers reading books aloud, or doing character impressions. It's wasn't useful for LibriVox, but it was interesting. I hope this is at least a little of both:
https://arstechnica.com/information-technology/2022/12/make-a-bathroom-sound-like-a-pro-audio-booth-with-adobes-free-ai-tool/
We'd obviously prefer to get a good recording in the first place, but for those cases where Audacity switched to recording from your laptop's built-in microphone, or there's a problem with the gain, or some other issue that you don't discover until the recording is done... well, this kind of thing could make a nice rescue tool.
While the news here is that Adobe is offering this as a service, free of charge (for now), I was more interested to see there's an open-source program that anyone can install themselves. I might have to see if it can remove the chorus of neighborhood dogs, which would mean I don't have to stop recording during their impromptu performances.
Another thing that's interesting on the technical side is what they called "hallucinated results" or "phantom voices".
This kind of program is typically called "AI" because, after being "trained" by trial and error, they now produce results that seem way too smart for a basic "if this, then that" program. Their "ears" have similar limitations to ours, but their "brains" process things very differently... which means they make a few of the same mistakes we would make, but also a few that we would find surprising, and which always fascinate me.