Agreed. That sounds like a recipe for "we don't know how 'the algorithm' came up with what it did" kinds of excuses when, inevitably, inaccuracies are found. It also seems, conveniently, to make the processing system practically unimpeachable.
You're not the first person to focus on the transcript, but you're forgetting that the person checking the note, the doctor, was also in the session and remembers what happens. This isn't an issue.
That's the job of the provider. There's no other way to actually verify the accuracy of the note. You can't actually engineer humans out of the loop, the loop revolves around humans.
They read the note. They were in the session, if they can't remember what happened minutes before then we have bigger problems than a lack of transcript.