If GaN acts like a V-Fet or a SIT, then they might be on to something.
The non linear gain curve of a modern transistor will generate odd order harmonics, which is where the problem in audio lies.
A SIT or V-Fet, produces predominantly even order harmonics, which is tied to possessing what is principally a linear gain curve. More akin to a tube. A triode tube, specifically. This is critical. Big time critical.
To test this theory, one has to make a modern Class D amp with a modern V-fet, which means Tokin devices. Which are available on ebay as the last of their kind. These are high current high voltage devices. (the Tokin versions)
The future of audio is very likely V-Fet/ SIT.
Class D, as it stands, produces what is predominantly odd order harmonic hash, which is filtered.
The shape of that odd ordered hash is mistaken for detail, as the components of it (mix of odd ordered harmonics) meander around (shift and change) in the presented mix.
I heard it for what it was the first time I head a class D amp, and it is there in all of them. Garish, faulted right to the emergent core, unlistenable. For me, that is....as I heard it and recognized it...that it is totally against the aim of audio technology, in my view, hearing and understanding. To take the worst and unwanted part of what solid state devices do and make it the ENTIRE waveform. Jebus.
This is mostly testable in a finished item...as Sony did it right and made a class D amp back in the day, with SIT or V-Fets as the output devices.
Not as much of a modern refinement as the current crop of Class D amps, but the build/design clues for working with SIT devices are all there in a pre-existing finished item.
Interestingly enough, modern V-Fets/SITs are designed to be high frequency high powered choppers of the like that are required for Class D design.
We had it in hand, we were on our way... but mosfets killed the audio star....