This all happens on the Cut step, where the first pass lands. The right of the window carries two tabs, Tune for the cut itself and Notes for what you flagged, so checking and fixing are the same place rather than a trip to another screen.
Press play and watch. Space pauses and asks what is wrong, in a sentence. M marks the moment and keeps playing, for when you would rather not stop.
A note remembers the instant in the recording and the word being said, so it still points at the right moment after the cut changes around it.