That's what I suspected ... the difference between adding a music soundtrack and adding a voiceover is pretty negligible. So if I am going to consider adding narration instead of just captions, it will increase the size of the video by the same amount anyway.

So it seems the most economical delivery would be just to have captions, and no sound at all.  If I add narration, it the download size won't be any different than it would with a music track.

Interesting.