Version: Firmware 13.1.3 + DOS 2.2.4 Date: 2019-02-12

[David A]

David,

I don’t disagree with what you wrote, but it’s not quite what I was trying to do.  I’ll try and explain, I might not do a good job so work with me.
We already know that our amps measure very well, so well that they are already way below audible thresholds in all major areas.  You mention distortion, as I remember it the lowest amount of distortion that can be detected in listening tests with music is 0.1%, but there is some research that shows that 0.01% might be audible under very controlled conditions with test tones, which is still way more distortion than our amps have, so if distortion was made lower again it would not matter, as we would not be able to hear the change.  On the other hand lots of people (probably including me) like a bit of distortion from time to time, of the right sort, but if that had been added to ‘improve’ the sound it would probably be of the 0.5% or higher of 2nd order type, which my tests would have shown easily, distortion was 0.2% across most of the spectrum in my test.  So I was not looking for lower distortion, and I agree I could not measure a decrease in distortion, but rather checking that it was not higher.  Now I would have fallen of my chair if Devialet had added distortion, I was really looking for anything else that stood out, something that would easily change a listening test, like a change to frequency response or playback level.  They would have easily shown up in REW sweeps, but they were not changed at all, so much so I had to double check as the lines on the graph they were so close to each other, I don’t normally get results that repeatable.  Nothing was found be me, as far as I can tell there is no measurable change, with the exception of below 100hz, were my tests are not reliable as background noise and increasing speaker distortion intrudes.
So I don’t detect any cheating that some manufacturers have indulged in over the years, which is good.

My point was that although I'm hearing a difference, it's not a difference I would have expected to show up on any test done by measuring speaker output.

I think the things which would need to be tested are DAC input vs output with both RAAT and AIR, and also any differences between signal output from Roon itself to signal received by the Devialet apart from the musical data in )s and 1s. The music content seems identical, frequencies and loudness levels aren't changing, so things like packet error rates, background noise levels on the connection, timing differences in how the packets are transferred and so on are the kinds of things I'd be looking at there but then there's also the issue of showing how such changes can make a difference to the audible output.

If RAAT makes a difference, and I believe it does, proving that with measurements may not be all that easy. We still can't measure everything that's out there in the world, and we don't always measure everything we could measure but rather just the things we think will be most useful.  We may need new tests to reveal a difference in this case (audio history is littered with the development of new tests to measure things which make an audible difference but couldn't be measured at the time) or we may just need to find the right test out of the many tests we have available including those not usually used in equipment reviews.