What do you mean that no recordings are not phase coherent? Lets define what you mean by phase coherent first.
Most recordings are made with a multiplicity of microphones widely scattered about. Now every microphone hears everything, but intensity dies down with distance from the source. This is called bleed.
The result is basically a cacophony of sound at every mic. It is a wonder recordings sound as good as they do.
In minimalist set ups, there are two ways of going about it.
Phase difference stereo and intensity stereo.
Phase difference is the system usually chosen on these types of set up. If you own any Telarc recordings then that is what you are hearing. This involves widely spaced omnidirectional microphones at some distance from the source. Obviously this is not phase coherent, but better then a bunch of mics scattered about.
Intensity stereo, which was invented by the originator of stereo recordings,
Allan Blumlein, involves the right and left pair being coincident in space.
The mics can be cardiod, figure if 8 or matrix M-S.
Best results are obtained with very expensive auditorium sound field microphones.
I used the Neumann SM-69 FET. I still have it safely tucked away.
The top capsule can rotate with respect to the lower capsule. The pick up patterns can be varied remotely.
I made some xy figure of 8 recordings, but mainly the very unusual M-S Matrix recordings. This was for two reasons. A lot of my recordings were for radio broadcast. An M-S recording gives a listener with a mono radio a perfect mono signal, since they only hear one capsule. The other reason is that when the session is monitored with very good accurate speakers, the recording engineer has a lot of control over the sound field, especially the depth of field. You can absolutely not monitor these types of sessions with headphones.
Sometimes I would use an even more unusual technique, especially for chamber music. I would use omni mics spaced about 9" apart with a plexiglass baffle between them. I made a special stand for this.
These techniques are virtually unknown and hardly practiced in North America.
They used to be common in Europe.
The big problem is that it to some extent precludes the use of spot mics. However since I was often working with non professional singers, I would spot mic them gently with an omni mic most often. I would follow the score and gently spot them during their solo. This did not upset the apple cart significantly, and made for a better production over all.
In these recordings I find the sound deteriorates as the phase aberrations of the speaker increase. In speakers that really play fast and loose with phase, even if they are otherwise satisfactory speakers, these recordings actually sound quite bad.
For this reason, in my designs I try and minimize phase aberrations.
These recordings also contain a lot of accurate spacial information and sound excellent with algorithms like Dolby plx2 music.
I think I was sensitized to the problem of phase shifts early in life. I was a full ranger until around 1984. I still am at heart. I certainly do not regard full range driver enthusiasts as nutters by a long shot.
Also at that time I adopted full range speakers, specifically the Jordan Watts modular loudspeaker, the BBC were using phase coherent techniques for their broadcast, especially the Proms from the Royal Albert Hall. They used large diaphragm ribbons as figure of 8 Blumlein pair.
So that is how you make a phase coincident recording. I would say it is impossible to make a phase coherent pop or rock recording at this time.
To answer your original question, if you want a phase coherent speaker you have two choices. You need to use a full range driver. I think most full range enthusiasts make their own cabinets.
Lowther have a range of complete full range driver horn loaded speakers as well as drivers for sale.
Your other option is the
Quad electrostatic.
I'm not aware of any other options, no matter what their designers may claim.