Good job. Quite impressive circuitry, the voltage regulators alone are pretty fancy. Output noise should be unspectacular but generally OK, there is only one stage of volume control by the looks of it.
Looks like it should be easy enough to implement some bootstrapped cascode goodness via a capacitor across R150-D30 (and whatever the other channel equivalent is). Something in the double-digit µF range should do just fine, I think, maybe even a bit less if that results in turn-on issues. That should bring distortion down even further.
I did spot one potential problem - when high gain mode is enabled, C55 is about an order of magnitude too small, so the amplifier would end up being on the bass-light side. (I presume it's a film cap?) Parallel some more capacitance if needed (maybe two 47-100µ back-to-back).
With some more C in that spot, there'd be a realistic chance of bringing output noise down further as well. Right now it's by no means terrible but just slightly better than average for a dedicated pre, being hampered by moderately high feedback resistor values. Here are some values to bring the wideband noise floor down by about 3 dB:
R140/141/142 = 1k0 / 154R / 4k22
and in order to avoid increasing output loading:
R2 = R193 = 11-12k, matched / <1%
(Should output loading be a noteworthy contributor to distortion, you can always move the input-side end of R193 to ground. The balanced output is still going to work, just 6 dB quieter and with some common-mode component. In the stock circuit, this would bring main amp output loading from about 3.7k to 14k. Not as much benefit when modified, of course.)
Resulting noise floor should be little higher than 3 µV, good enough for 95-ish dB/W/m speakers with +30 dB worth of power amp following. That's 90 µV on the output at 44 dB of gain or good integrated amp level. Those with super high efficiency speakers - use an attenuator between pre and power amp or a variable input or generally lower gain (e.g. 22-26 dB) power amp, with the power amp not being too noisy in itself (good ones will manage around 30 µV). Top-end pres from the late '70s to early '90s would often be fitted with two-stage volume controls, usually employing 4-gang pots.
I would like to share as it may be useful for others as well. Use at your own risk. It is correct to the best of my knowledge, some component type number were unreadable though, mostly diodes and a few transistors on heatsinks. 
