I've been transcribing more vocal lines lately in songs I like, to go along with the usual instrument transcription I do, and I've noticed something interesting - a lot of my favorite vocal parts clash very very noticeably with other instruments in the song when that melody's played back by an instrument, but they sound perfectly fine, consonant, even, when performed by a vocalist. There are multiple dimensions to this clashing - usually there's some rhythmic clashing (vocal parts are often polyrhythmic when compared to instrumentals) and harmonic clashing (voice-crossing, dissonant intervals, microtonal shifts, to name a few, often in the same range as other instruments).
At first I thought that maybe I was transcribing the vocal melody in the wrong octave, maybe one octave too low, but after more careful listening that wasn't it.
Why / how could this be the case? What is it about the human voice that makes it possible to get away with clashing vocal melodies that'd just sound like a hot mess or noise if played on any instruments instead? Or is there something in the way the melodies themselves are constructed that makes them work with the human voice in this way?