This assumes you know the method they are using. As the post mentioned, a few milliseconds + or - at the end of a scene would not be noticeable and the overall timestamp would still be constant if "self-compensating offsets" were used.