This is a multi-faceted issue (which is what makes it so fascinating), and I don't know of any other way of approaching it than starting from two assumptions:
But most people start with these assumptions, I think (- tell me where I'm wrong on this):
- Humans separate source from room in their hearing; the room sounds like spaciousness; the listener hears the source itself effectively 'dry';
- While each natural source has its own dispersion characteristics, the ideal speaker should have equal directivity at all frequencies (it doesn't need to be omnidirectional, however).
But most people start with these assumptions, I think (- tell me where I'm wrong on this):
- The job of an audio system is to get the recording as accurately as possible to the listener's ears;
- The most important aspect of audio is frequency response magnitude, so this should be used as our basis for assessment of accuracy