Possibly repeating/paraphrasing what Antoine has already said, but…the Devialet seems to be entirely responsible for clocking the data received from the network interface through its audio pipeline. Based on what I can see by poking around in the AIR network packets, the Devialet sends explicit feedback to the computer allowing it to work out whether it needs to send samples at a higher or lower rate. Since the buffer is large, as Antoine noted, this doesn't have to be a particularly tight or accurate feedback loop - all it has to do is stop the buffer overflowing or underflowing. So the audio pipeline in the Devialet should be completely de-coupled from any clocks in the computer.
(Having spent the last few weeks writing software to implement an asynchronous USB audio device with rate adaptation and implicit feedback, this is a subject close to my heart and I can see that in some ways the Devialet AIR solution is simpler and more elegant.)
Guillaume and Rufus - you must be exhausted after all that listening! A very interesting comparison, and thanks for taking the time to share your results. Looking forward to see how the Aries fares in its re-match!
(Having spent the last few weeks writing software to implement an asynchronous USB audio device with rate adaptation and implicit feedback, this is a subject close to my heart and I can see that in some ways the Devialet AIR solution is simpler and more elegant.)
Guillaume and Rufus - you must be exhausted after all that listening! A very interesting comparison, and thanks for taking the time to share your results. Looking forward to see how the Aries fares in its re-match!
Roon (Mac Mini), Wilson Benesch Full Circle, Expert 1000 Pro CI, Kaiser Chiara
Warwickshire, UK