See, that's still a problem for me; as a person with hearing impairments, I find it very difficult to just pick out one sound (as well as the source of noise, but for a video, that's irrelevant). I love my videos with subtitles, but failing that, they need to have one channel of sound to work for me - either the original audio, or the commentary. Both just makes a video frustrating for me to watch.
I would love Jafar's suggestion, on the other hand, of customisable levels. Shame that'll probably never happen on YouTube!