It seems that music is a sound, and sound is constituted by vibrations of a surrounding medium, i.e., typically air. But you can listen to a musical piece through a bone conduction headset. In that case, you're not listening to vibrations of a surrounding medium. Moreover, we could suppose that the performer is producing electronic music live, which is directly piped to the audience's bone conduction headsets without any speakers anywhere, so it's never in the air (except accidentally). Assuming that the bone conduction headsets produce the same experienced quality as listening to the music the normal way, it seems that the audience wouldn't be losing out on anything musically relevant. Yet, if music is a kind of sound, and sound is vibrations of a surrounding medium, then we have a paradox:
- There is no music there.
- But the audience isn't missing out on anything musically relevant.
- So, music need not be musically relevant!
Perhaps, though, music doesn't require vibrations of a surrounding medium. The vibrations of bones might be sufficient to qualify as music. I am not sure, however, whether in the concert that I have imagined the audience counts as hearing vibrations of their bones. Yes, their bones vibrate, but the content of the experience isn't the vibration of bones. Rather, it sounds like sound coming from outside them, so the content is external sound, but in my story that's absent, replaced by a mere illusion of sound.
In any case, we can modify the story. Suppose the piece is performed electronically, and never generates the relevant vibrations. Instead, it is directly piped to the performer's and audience's brains' auditory centers. It seems that musically nothing is lost, even though now there is definitely no sound at all.
The conclusion that music needn't be musically relevant is absurd. So we have to deny the claim that music is a sound. What is it then? A sequence of experiences? Then there is no music when the performer and audience are deaf. Maybe that's a bullet we should bite?
Maybe music--both the music composed by a composer and the music performed by a performer (who may be in part or whole a composer, as in cases of improvisation)--should be seen as an abstract sound type. The composer and performer discover music but don't create it. In order to grasp an abstract sound type, it is not needed that one hear an instance of it, but only that one have an experience as of hearing an instance of it. The performer, thus, causes the audience to have experiences as of hearing instances of it. Those experiences are neither sound nor music. We can then, by extension, call an instance of the abstract sound type--i.e., a concrete sound--"music". But music in this sense is not musically relevant except as a vehicle for music in the Platonic sense.
(In case it's relevant, I should note that I'm largely tone deaf, and I do not speak from experience.)
(One can mount another argument against the thesis that music is a sound on the basis of Cage's 4'33". But if 4'33" is music, we could still say it normally involves sound, in that one could have a more nuanced theory on which sound isn't the vibrations, but a token pattern of vibrations. And no-vibrations counts a pattern of vibrations.)