Thanks Jim, and welcome.
It's a nice, straightforward digital design; no programmable logic, no microcontrollers. I'm not sure why they use the ACT logic gates in this configuartion as cable drivers; maybe they have some inside knowledge that each gate's driving impedance is about 150R (IME you're looking at closer to 15-20R for that technology), otherwise the output is not matched to either 75R or 50R. Perhaps a miscommunication between engineering and production: either use a single gate and the specified 51R1 termination, or have three gates with a 150..180R series resistor each in parallel.
It's not going to win any prizes for jitter. The signal passes through a lot of gates, and those HCT393s they use to divide the MCLK down are rather noisy. They spec an unqualified 20ns jitter. I might believe that if they have very low-noise crystal oscillators, but for either that TLC2932 PLL or the AES in through the CS8414 I'd expect several hundreds of ns worth.
All in all a nice design, with some cute mixed-signal tricks. The performance appears limited by the engineer's budget rather than his skills, which is always good to see. I wouldn't recommend DIYing this unit; due to economy of scale your build cost will easily exceed a second-hand or even new unit of equivalent specs.
JDB.