One has to remember that digital transmissions are engineered to reject noise (RF or other) that may have been injected into the cable.
It is a certainty that no perfect square waves are received (or even generated) at either end. Then there may be additional noise coming from the environment that will mangle the perfect square wave some more.
This is why there are operational margins in terms of amplitude and time (and shape). As long as the received signal is more or less at the high (or low) value for about the right amount of time, it gives a 1 (or zero).
Overshoots, undershoots, ringing and delays… Still works!
Source: EE|Times - Jitter, Noise, and Signal Integrity at High-Speed: A Tutorial–Part I - By Dr. Mike Peng Li, 12.10.07 - https://www.eetimes.com/document.asp?doc_id=1275083
Once the signal has been evaluated as a 1 or a 0, that’s the end of noise on that segment of the transmission.
Noise is rejected at every digital hop, external or internal, and is not carried over. This is valid not only for USB and Ethernet, but also for every transmissions inside digital equipment, inside and between chips.
If noise gets bad enough to make evaluating a 0 or a 1, because a piece of equipment is broken or very badly built, then error detection and correction (usually with a retransmit) kicks in. You do not have to worry about timing, because those transmissions are really quicker than anything that could worry digital music data decoding, and because buffers are in use in most cases anyway.
If noise becomes so bad that there are too many errors detected and too many retransmissions that end in error too, then all breaks down. You go from no errors, to some detected but corrected errors to total failures, without any degraded in-between state.
To illustrate, remember how analogue TV would go from fine to snowy degraded pictures, and how digital TV goes from fine to nothing, no snow (transitional visible artefacts are created temporarily as buffers empty with partial data). It’s the same for digital music, it will go from playing to not playing, without any state of degradation in-between.
Also, we are talking about digital transmissions of data, not music. Those transmissions happen at fixed frequencies. So invoking the fact that HF audio is transmitted better inside USB cable so and so at $1500 because it handles HF aufio better is pretty hard to read.
To make it short, noise in digital transmissions is irrelevant. Totally. Not by chance, but by design. It’s one of the great advantages of having moved from analogue to digital systems. The only place where noise is an issue starts in the analogue half of a DAC and beyond. Never, ever, before.
That means that esoteric digital cables can have nothing on properly built cables sold at affordable prices (which can be bought at any local supermarkets for low two digits values), playing from memory or underclocked or gui-less or optimised systems makes no sense, linear power supplies are a waste of energy by lack of efficiency. Cheap or expensive, there is no “audiophile” digital equipment that will work better, be it a network switch, a computer, a hard drive, SSD or a cable. If it works already, it cannot be made to work better.
What we have here is a lack of understanding of the engineered nature of digital transmissions, applying concepts relevant to the analogue world where they does not apply.
Also remember that transmitting, decoding and playing digital audio is a low-level effort for any computing equipment. If a cheap Chinese smart watch can do it, do not worry about the capacities of any equipment capable of running Roon, or being used around Roon.
Save your money, do not buy into the audiophile digital stuff, buy more music, or book a week of vacation on one of those beautiful islands if that’s more your thing!
EDIT: if you want to put more that $10 into a USB or ethernet cable because it looks better built or because it looks better next to your nice equipment, then, by all means, go right ahead. We are all entitled to non-practical features. I like nice looking expensive watches, when a $2 piece from the supermarket would give me the exact same time with the same accuracy.
Knowing what I know about embedded software, that might might be as outlandish as it sounds. After all, some of you might remember the recent XMOS USB receiver kerfuffle…