1. TCP/UDP and network protocols are only tangentially relevant to this discussion. At the end of the day, the one thing I think that 99% of us will agree on is that a streamer or network endpoint (roon or hqp or direct tidal/quboz) will deliver exactly the same bit-perfect digital audio from it’s *output*. How the data got from source/cloud/drives/whatever via streamer to the output is somewhat moot. All streamers are going to deliver bit-perfect digital audio output.
2. That said, while the *content* of the data is the same, the timing of the delivery of the content may differ for various reasons, something that manifests as "jitter" and could have subtle audible results. Could be due to network issues (wifi or ethernet), or processing/cpu issues on either sending source or receiving side. As @ironlung described above, this is dealt with in various ways between a streamer/network endpoint and a DAC, via combination of clock synchronization and buffering. Better streamers/endpoints will output with less jitter and thus make it easier for a DAC to reconstruct the timing with the most precision possible. Similarly, better DACs will be able overcome inbound jitter more easily than others. Things like asynchronous USB audio were also formalized in order to minimize these issues. At this point in it’s evolution, it’s questionable if jitter is really ever an issue, particularly to the point where it affects the audible end-result. Some still think it is, some don’t. As long as the protocols, devices, and network in use are of sufficient quality it should probably be fine.
3a. The remaining unknown is electro-magnetic noise and interference that might get ’passed along’ from a streamer/endpoint to the the DAC along the cable. In theory, this noise might affect the output of the DAC process with audible consequences. Indeed, this seems to be often realized when people connect DAC’s directly to PC’s/laptops. The PC’s have high-noise switching power supplies and other processes running that cause issues. This issue has been exacerbated by the fact that USB Audio, unlike SPDIF/etc, allows for a 5V power bus signal. So on most implementations, the USB connection is carrying not only the (bit-perfect) digital audio but also an inherently noisy max of 5V @ 500ma power. This power bus will transmit any excess electrical noise from source to DAC, just like a power cord may. And until recently, most USB receiver cards/implementations inside of DAC’s did not deal with this well or at all. This is why many, when doing A/B comparison, have felt that SPDIF sounded better than USB, all else being equal. More recently, some USB receivers self power and don’t pull power over USB and/or galvanically isolate the input to mitigate the issue. Just like one wants a good low-noise (linear) power supply for their DAC, they should also want a low-noise input. And the better the power supply on the streamer/endpoint the less noise it will transmit in the first place.
3b. Noise can also similarly travel along ethernet cables. So any wired connections carrying upstream/source data/audio to a streamer/endpoint maybe also carry some noise. Ways to mitigate this include using high-quality wifi, but as per above, may induce electro-magnetic interference of it’s own. Or using fibre optic instead of copper ethernet - which effectively carries only 100% data - and isolates any upstream noise.
My personal setup includes a cheap Roon core running on a PC in my basement which is then hardwired (via routers/whatever) to a Sonore OpticalRendu endpoint ($1000) via a fibre optic media converter. The OpticalRendu is powered by a low-noise LPS and has low-noise data input over fibre, buffers source data and clocks to provide clean low-jitter output. I then use USB to my DAC which importantly has a USB receiver that is *not* bus powered. While the noise issues may be overstated I sleep well knowing that I removed any potential gremlins and have not spent too much $$$ in doing so.
IMO, the ’streamer’ part of a digital audio chain has very basic needs and can be done on commodity hardware it if’s properly isolated from the DAC/endpoint. $4000+ for a entry-level Lumin streamer with a noisy switching power supply??? $10,000 with mainly off-the-shelf PC parts wrapped in a nice chassis with a 3 line UI on the front? Unless you really really have to have a fancy faceplate with the song name, bitrate showing on your rack ...
[for what it’s worth, I have spent the last 15 years designing/coding low-level protocols over both TCP and UDP for real-time financial data]
