Audio ultimately leads to a 1D data track. Ways to play that track is comparatively well understood, and well abstracted from the workload itself. We've been using 3.5mm jack at scale since the original Walkman, for example. That's 1979.

Graphics on the other hand constantly changes. Apple II around the same timeframe had 280×192 resolution -not even 4:3 aspect ratio- and 16 color that only existed because of a hack in the NTSC spec. Now we have dozen different common aspect ratios, 2 common sizes that we have to support (mobile and monitor), dozens of different resolutions, running on thousands of different hardware, that are near impossible to test the combination of their configurations, let alone the variety of behavior that comes from their software - the list is just endless.

Audio is played by speakers, there isn't much variance. Graphics on the other hand changes constantly, and most of those changes result in throwing out a lot of software. The situation is so incredibly bad that most cross platform products threw the towel and resorted to using perhaps the second most complicated software humanity ever devised - the browser - instead of trying to get a window and a button to work on 2 different operating systems.

And don't even get me started on prints :)