There's a discussion elsewhere on AudiogoN that discusses the perfect speaker. An interesting exercise, and confusing when the adjectives start flying with no description of the meaning of the words. First, a speaker should do no harm. In other words, it shouldn't change the sound (especially instruments), in any way, that was captured on the recording. 1) A flat frequency response is the first requirement (and easier said than done for most speaker designers). 2) Near zero relative phase response (how the drivers relate to one another in terms of phase). This will permit a proper leading-edge dynamic. It will allow a speaker to disappear as a source of the sound. 3) With those two goals intact, the speaker must be able to faithfully track the dynamics of the recording – both micro and macro dynamics (don't underestimate the importance of faithfully resolving micro dynamics). 4) Details: the elimination of anything that veils the sound on the recording. And the ability to represent all of the ambience (reverb) the recording has on offer. If all done well, the speaker will enlist your attention and draw you into the music's beauty and intent.
At the present time the speakers I find doing this the best are from Vivid Audio. Both the Kaya series (4 models) and Giya series (4 models). The best example of this is certainly the Giya series (G4s2, G3s2, G2s2, G1 Spirit).