Absolutely. The audio clocks 22.5792MHz/24.576MHz (of multiples thereof) should ALWAYS be as close to the DAC chip as possible.
I don't think this is the architecture Alex. I believe this is a master/slave configuration with the XMOS the master for USB. It generates MCLK for the CD4398 DAC:
XMOS needs to know the output clock rate anyway so it runs as a master, driving the CS4398 as clock, creating MCLK signal above.
Alternatively there is a CS8416 which extracts the clock from input and through a MUX drives the CS4398 as a slave just the same.
This is why there is no clock/crystal around the CS4398 at all. They are all near XMOS which is the device that needs them. It, internally then generates the appropriate master clock for CS4398.
Here is the block diagram in Amir-CAD
:
The little circles are the local oscillators for XMOS as I have shown.
Again, I am just guessing at this. I have not tried to trace routes and such. Don, what do you think?