I can't prove it, but I think the problem is that the soundfile played during some of these dialogues is shorter than what is actually written. So when the game finishes the soundfile it automatically goes to the next part.
It's like they created these voiceovers during development and some guy decided to add some more text to the existing dialogues but never thought of redoing the voiceovers.
In some occasions it's annoying, since dialogue often contains hints on what to do next.
All three games (Menzoberranzan, Ravenloft 1 and 2), although awesome, are very sloppy products in my opinion.
The music in your video sounds off.
Here's a great example of what the music used to sound like:
https://www.youtube.com/watch?v=qz8RjL6YbjA&list=PLXkzOfiaQyuIemA_WjLJVDcvCRPA-gqOz (Unfortunately this guy hasn't got any tracks from Menzoberranzan.)
If only Dosbox could emulate those old OPL2/OPL3 soundcards...