Serious thoughts:
In the schematics posted on the first page:
u2b (one of the op amps used as a voltage source in the PS) does not have a 270Kohm resistor on it like U2a. These resistors and the0.22uf caps block anything but DC from reaching the mosfets.
Why arent the mosfets Q1/Q2 in the feedback loops of the opamps? It would reduce the output impedance of the power supply which would probably drop the crosstalk of the amp. 60db is "good" not "excellent" if we are talking about measurements without reference of audibility.
R25 could be exchanged for a JFET CCS or a CRD. CCS's are pimp.
The bandwidth limiting resistors and caps before the opamps is a nice addition. I hope this catches on in other amps.
Im 50/50 on the idea of the "opamp before the volume control" thing. Its not conventional, but has been done before. It solves the problem of noise from the opamp elegantly, BUT the gain stage is prone to overloading with excessive input voltage. This is noted on the schematic - its worth noting if you hook up to pro-level outputs. The fist kid who figures out how to set the gain to 10 (because 10 is more and more is better :facepalm: ) is going to have a wild ride.
The selected gains of 3 and 7 are pretty tasteful. Id lean towards just one. 3 or 4 covers you for 99% of all headphones, people just like switches.
I think the 0.8db channel matching quoted in the test data may be misleading. Channel matching is primarily a function of the channel matching of the volume control, which is not generally great in those little pots. Maybe they all line up nicely at maximum volume, but at real levels Id bet on worse. Not really a fault of the designer just saying...
Coupling caps, UGGGGGGH. UGGGh, uggggh. Thats the kind of stuff you'd expect to see in the kind of tube amp a guy like nikongod would design. Why not DC couple it? Ugggggh. At the very least put them in front of the pot where the signal is larger. Uggh, really? And just no. I guess they will save the headphones (sort of, but no) if the gain opamp goes all funky with excessive input signal.
I like the option of an output resistor outside of the feedback loop for the output op amps even though nobody will use it. I wonder if this admits the importance of subjective quality in spite of "knowing" how to do it right. heh. Id put a switch across this one so the user can switch on the fly like on the XIN amps. Its a good spot for ferrites or hand wound inductors should the need arise in actual use.