What makes you think that the languages sounded more similar in a time when communication was much slower and regional dialects could develop easily without outside influence?
The net difference today is the net difference 2000 years ago plus all the new variations since then.
After the time of the source, shortly after diverging[1], they would still have been more similar than they are now because they hadn't had much time to differentiate.
[1] Unlike "the time of the source", you can rigorously defend the existence of the period shortly after divergence.