The difference is that in the 80’s and early 90’s, when CD players were relatively primitive, jitter mattered quite a bit, as there was limited buffer between the CD bits and the DAC.
These days, there are about half a dozen buffers in-between the digital transmission and the analog audio that prevent any of it from mattering. Especially not at the network/LAN level. It’s simply impossible for it to matter.
TCP/IP transmits data via a series of buffered packets, with error correction and checks for accuracy in both directions. The data is encapsulated inside those packets. The timing of the packets is irrelevant, even the order is irrelevant. Often, packets are sent out of order or retried if one fails. That is commonplace and normal for the protocol. It’s all irrelevant.
Once the packets get to the computer, they’re decoded and reconstructed into a perfect series of bits, the same as would be if transmitted over WiFi, ethernet, serial cable, copying them to a USB drive and carrying the USB drive to the destination computer, or writing them down and painstakingly typing them in the destination computer in a text file. The computer doesn’t care, they’re all completely identical bits regardless of how they’re transmitted. This is guaranteed by the protocol, and is mathematically provable. If it were not mathematically perfect, then not only would audio be impacted, nothing in computers would function—that’s how certain it is.
Once on the computer, those bits can be sent to the device actually doing the audio conversion. Often this is an optical or USB DAC. The way those devices work is via another buffered data transmission method, a simpler one but the same concept. Optical audio is slightly more strict on order and timing, but the result is still buffered and guaranteed to be identical to the source data. USB is even more reliable and guaranteed, and there’s a chip on either side of the wire that ensures that’s true.
Once on the DAC device, then the DAC chip is responsible for taking that stream of bits—again, mathematically guaranteed to be identical to the samples on the Roon server—not just maybe or probably or by theory, but absolutely impossible to be different—and converting it to an analog signal.
This is where the quality can start to be impacted, and I’m personally not a believer in such nonsense as all DACs sounding the same or all that. Your DAC is doing an analog job, and can absolutely have nuances and impact the sound. That’s where the quality matters. But the digital bits up to that point? Does not matter one bit how they get there, so long as the DAC chip is implemented properly to read them with the right timing and design. Which, in 2023, nearly all are—it’s a solved problem.
The network in particular has absolutely no bearing on the sound, and again I’m open to many many quality possibilities that impact the sound, but this is not one of them. It’s simply impossible. This is one place where if you can hear a difference, it’s guaranteed to be placebo, and you need to revisit your assumptions.
