What is left in terms of speaker design to achieve, in terms of lowering audible distortion and hence retrieving more neutral sonic information from recordings?
My first classical record acquisitions were by
Audio Fidelity, because cheap when I was poor (and remain frugal, so still own):
Although they may have been first played using a monophonic cartridge, when
Hafler surround was announced, these were among LPs with most plausible ambience.
These were reportedly recorded using "Frey Stereophonic Curtain of Sound" (Sidney Frey being Audio Fidelity's owner) with "Telefunken stereophonic microphones".
Since released in 1959, and my understanding is that Telefunken was not yet producing their own, those may have been Neumann condensers
distributed with Telefunken logos and probably monophonic. Having even today never owned more than matched pairs of speakers,
I cobbled together an active approximation of
Dynaquad, to balance levels from whatever speakers were available to graduate students at the time.
Bottom line: matrix surround playback IMO retrieved more useful information from those stereo recordings than any conventional stereo system that I have yet heard.
It seems generally accepted that a pair of microphones can suffice to capture a stereo recording intended for reproduction by a pair of loudspeakers
(or perhaps more credibly, with suitable provisions and processing, stereo headphones).
That sound pressure captured at 2 points in space should be acceptable for recreating an entire sound space
or the sound image from musicians in that space seems absurd to contemplate, but is generally accepted.
Mathematically, one would need to record sound pressure over the entire surface of the space to reproduce it.
I suppose that, were recordings simultaneously made (1)using a binaural recording apparatus and (2)for conventional stereo playback,
then playback of that stereo recording (3)
subsequently rerecorded by the binaural apparatus,
(3) would never be mistaken for
original binaural recording(1).
Consequently, perhaps some loudspeaker distortions are important for preferred reproduction.
Memory color is a generally accepted concept among so-called color scientists,
where consumers prefer grass being greener, water and sky bluer reproduced than in original scenes.
Beyond that, when rendered in smaller scale (e.g. on smartphones) even stronger contrast and saturation are preferred.
I suppose that comparable perception/memory issues apply to music reproduction.