Side note, I couldn't help notice that both your comment and the article at one point spell the word "meat" with a double E (as in meeting) - we can tell it's a speech-to-text mishap given the context here, but it makes me wonder how often this happens, and how much of an issue this is when the tool is used on a daily basis.

It's not a voice recognition issue as presumably both pronounciations are identical - so it's up to the software to infer the right word from context cues or grammar rules.

I've noticed quite a leap in the quality of dictation/STT over the past 10 years or so, and I really hope this and other features can continue to improve at this pace.