For me, I think it depends on how closely I care about the words or (for fiction at least) whether the scene is people talking or a description of the area or some action.

With dialog, or a narrator's voice, I often imagine them speaking it or provide a voice for them in my own head. With action scenes, especially engrossing ones, I sort of just spend the extra mental energy trying to picture the scene rather than "hearing" the words. I have a very hard time paying attention to what the characters look like or are wearing. Usually imagine it briefly as it is being described and move on.

Oddly enough, when I'm reading aloud, sometimes I sort of stop imagining what is going on as I'm too busy trying to sort out how to speak the dialog (especially in some different voices) that I no longer pay complete attention to the scene.

For non-fiction, I generally spend more mental energy trying to visualize the concepts being described and stop "hearing" the words. Or if I'm merely skimming and want to get a sense of what is being said without the details.

So I think a "voice" is something I can and do switch on and off depending on whether my brain thinks it is important and probably on the difficulty of the text itself.