Haveing read and re-read the OP and many susequent posts, i admit its a little confusing. But my read is OP is not unaware of computer/DAC methods and has tried some. Heck, if he's tried Roon he's way beyond ("you can use a PC as a source").
But the key question was 'are they too noisy?"
I saw a few answers that i simply have to correct. First., yes ccomputers are noisy ask anyone who has tried to lay ou a PC circuit board or design an embedded system. The osciallators, chips and SMPSs all conribute to a lot of electtrical noise.
But inthe end, what do you think is behind that fancy faceplate with the 4 gold-plate machine screws and the engraved logo? very likely a Raspberry Pi, likly in embedded form, or something similar. Maybe a full- blown intel NUC or similar, like in a Nucleus. It MUST be a computer, the onyl question is "what efforts did they take to reduce noise?". Multipl power supplies, some linear; ground and power isolation; convection cooling, and nosie filters are all part of the game.
Youcna buy it or build it or make simple accomodations to minimize it. I suggesteed some methods of separating the sournce (noisy) fromt he DAC (must be clean, accept the point folks, the "digital" signal into a DAC is quasi-analog, with the magnitude digitally coded and the x-position semi analog. Its a cartesian system.. period.
So use ethernet (not wifi please) to a bridge (raspberry pi) and the problem mostyl disappears, IF you isolate the Pi's USB output with its own power - whcih might even be possible backward-powering it from the DAC.
When i began my jorney i sent SPDIF (issues!) to a vruy good DAC from the day. When i ripped everythign ALAC to my MacBookPro and played it back with itunes and bitperfect, over USB, it was ALWAYS better. Maybe not much, but better. So the idea that compuetrs cannot be used is simply false not only thoeretically btu in a crazy-resolvign system. I always test with a panel, even if its a drunk panel.
Note: the MacBookPro sounded best on battery power. Teh adapter added not onyl some potential hum (AC and ground noise) but deteriorated the sond - mostly the "black areas" (inter transient silence or decay). So that's antoher data point that power and isolation matter.
In thoery a really god DAC with it sown power and USB isolation renders all this moot. I hoep if done right that's true, and I'm heading down that path. But so far I'll note that small contributions remain that i cannot explain. yet.
So take those points as input. I still think Roon, for all is fauls and costs, has some real advantages not only as a multi-room and point of integration, but sonically (MQA unfold 1, DSP, excellent protocol, contorl ovr timing....)
G
