not so much thrown over as both the video and the original recording are synched in a editing program such as vegas or premiere or final cut
there is probably very good synch information recorded to the video before they show us the whats up intros that is used to synchronise everything during editing; i.e. a clapper board is used running time code or just recording the sound of the clapper, as well as count in beeps for the backing track which are then edited out during editing/synchronisation
the other clue that this not a live to video recording is that there is no ambient noise in the room - we do not hear the cameraman breathe for instance nor do we hear her moving about