In "The Nature of Noise", John Kulvicki defends a 'stable property' view of sounds, on which sounds are "dispositions of objects to vibrate in response to being stimulated." They aren't the compression waves that pass through the air -- they're the dispositional properties of objects that make them vibrate and produce those waves in response to thwacking. (I give him points for using the word 'thwacking' liberally in the paper.) As Kulvicki says, this would make sounds a lot like colors, at least on some fairly intuitive views of color.
Two things. The minor point is that we're going to have to have a really complex view of thwacking in order to make this all work out. Intuitively, the sound of Roy Sorenson is the sound he makes when he talks, not the sound that he makes when you thwack him, unless you interpret his speech as some kind of internal self-thwacking. Kulvicki's distinction between having sounds and making sounds, which he uses to deal with cases like audio equipment, doesn't seem to do the necessary work here. While I suppose I could get into saying that my stereo makes sounds it doesn't really have, it sounds weirder to say that Sorenson is making sounds he doesn't really have when he talks.
But my bigger objection has to do with silence. It seems pretty straightforward that when it is silent, there are no sounds. This goes along perfectly well with the view that sounds are compression waves. But it's big trouble for the view that sound is a stable dispositional property of objects. On the stable property view, silence is compatible with there being lots and lots of sounds! Objects retain their dispositions even if those dispositions aren't being activated, and the presence of lots of drums or other objects with wave-making dispositions will make it the case that there are lots of sounds, even if nobody is beating the drums and all is totally silent.