It also doesn't try to distinguish the speakers, which is weird. (It could just be "speaker #1", "speaker #2", etc.) I'd have thought it would be a lot easier to distinguish voices than to understand the English words they are saying.