It's real simple: everything involved in a system's video emulation that isn't a CPU goes in video/ and everything involved in audio that isn't a CPU goes in audio/. For the N64, the RDP and VI both go in video/ and the AI would go in audio/ if it weren't about 50 lines of code.
I was just questioning how a component is classed as a CPU, audio or video. Is a generic DSP being used as a sound coprocessor in one system, but for graphics T&L in another, better classified as a sound, a video or a processor? I'm not trying to change things, only to understand them. If I submit any code to assist in restructuring conceptual organization, I'd like for that concept to be the same as the MAME / MESS projects' concept. If the verdict here is to keep it in video/, I have no issue with complying.