Yeah, I think it was the voiced lines with multiple sentences. I've never thought about this stuff before, but ideally it would be nice if the text matched what you're hearing timing wise. But I'm not a programmer so I don't know how you'd do that exactly.