The height of the peak makes no difference. Other than the white noise is much louder than the music and
the NR fails to produce total silence. Look at the screen shots.
Yes - it wants the 'image' of the noise. So when doing NR properly you find a section of just that and no other signal.
I've now used Mozart, Beethoven, Merzbow, Vomir. Each when you use a sample of the 'music' as the reduction envelope gives silence, do the same with white noise you get artefacts. This is (for me at least) interesting. (and no I didn't select the maximum db peak area as reference!)
Yes i'm aware of how Dolby works, well the early tape versions, but that's again not the point. The point is what was expected doesn't occur, why? We observe phenomena, and some react with a 'so what?' or others get annoyed at the observer.
Maybe the important insights here then are more psychological? But i'm not interested in psychology, - here its sound. Funny how some use the term 'experimental' when there is none.... and when there is get abusive- by which I refer to poor Adrian, not yourself.