actual side-by-side content samples, one in voice and one out of voice, so there is a real reference point.