I experimented with the wait time field and it made things much worse as far as syncing.
What
happened is that the first frame had 120 seconds of audio but when I
switched screens it continued recording another 20 seconds of audio in
the first frame (for a total of 140 seconds, should have only been 120) instead of putting that 20 seconds in the next frame.
From then on, the audio looked like it was 20 seconds ahead of sync with
the video screen. i.e. I would say 'click file' and you wouldn't see the screen change to click file until 20 seconds later when the audio was commenting on the next step.