I'd imagine that accuracy is a key driver — e.g. for something like a courtroom setting, I think I'd more trust a professional stenographer to record proceedings accurately that a programme that's doing automated speech-to-text and then someone going through to correct errors afterwards while listening to a recording.

As an aside, I'd consider a different name if you need to say "rhymes with" (especially as without that I'd naturally read it as rhyming with clover).

Good explanation on the value of human captioning, https://www.youtube.com/watch?v=XKe1O7ppyqQ

voice recognition might be ubiquitious, but good voice recognition is not, and until that changes it's relevant. E.g. professional event live-capturing is going to be steno most of the time, for people doing subtitling on videos it's a useful skill for speed reasons, ...

Wherever the GDPR applies (and possibly similar regulations elsewhere), you cannot record someone's voice without prior consent. So, taking notes in meetings or courts is usually not possible/practical via voice recordings.