Of course, you realize I'm cheating with that picture. For example, I've overlapped the two lines of text by a couple of pixels, a real screen at that size would obscure even more of the visual because the descenders and ascenders couldn't overlap like they do in my faked image. And a lot of song titles would be too long for the screen and would have to scroll (a lot more than with the current font because that font is larger). Etc...
Still, I think something like this could be do-able as a userland app launched by Hijack. Have the player do Info:Off and figure out a way to query hijack for when the menus are up (so you can make your text disappear so it doesn't obscure the menus... doesn't emphatic do this already?).