With Hugo 1 galvanic isolation made huge differences to the SQ; and this was down to noise entering into Hugo's ground plane - and this noise was from two sources - the direct connection of the grounds from source to Hugo, and noise due to the decoding USB device as it unscrambles the USB data and buffers it to then feed it to the FPGA, with the FPGA supplying the clock.
The direct ground connection has been isolated by using RF filters, both on the ground and VBUS of the USB - this very effectively isolates Hugo 2 from the source ground. The USB decoder chip has a more efficient PSU, and full RF filtering, so noise from the decoding is also isolated from Hugo 2 analogue electronics. Moreover, Hugo 2 DAC and amplifier section is much better isolated from the battery, and we can see that from the stereo separation figures of better than 144 dB - and that's driving a 33 ohm load. You can see the separation (which actually indicates how well Hugo 2 is isolated from the power supplies) on this slide here:
Indeed, following the listening tests on Hugo 2's inputs I did (briefly) entertain the idea of not doing galvanic isolation for non-portable DAC's. Indeed, if Hugo 1 was as well isolated as Hugo 2, I would not have bothered to develop the galvanic isolation I currently use.
So to answer your question - with Hugo 2 the benefits of galvanic isolation has been reduced by at least an order of magnitude; if you are using HP, then don't worry about it all as it is a non-issue; if you are feeding power-amps then it can make a small improvement. Of course, YMWV as differing sources and different power amps will have different sensitivities. You can test this out by listening to USB or optical - if they sound the same, then you have no problems, and do not need further treatment at all.
Rob