James_I
(The truth is out there but not necessarily here)
122
Question for you: does it sound better while you are actively using Roon on the device? I.e. can you surf through artists, etc., and that sound quality improvement stays in place? In theory this should not be the case, in the sense that RAM playback is meant to eliminate processing noise…if you are processing other things, then it shouldn’t be the case.
My supposition is that it’s not RAM playback per se that is creating the difference in sound. There is some other difference between Roon and the internal player that accounts for it. Just a theory.
Yeah interesting question and one that’s too technical for me to answer to be honest.
I chatted to Nuno from Innuos at the Bristol HiFi show in February and whist he listed the many aspects they’d paid attention to for reducing noise in the SE he said that playback from RAM was the biggest factor as it created a buffer between the noise off the SSD and the actual playback.
I’m not doing his explanation justice by any means but that was the gist of it.
Meantime following the exploration of memory playback a hot topic over on the CA forums right now is loading the full OS into RAM. Doing so using Audiolinux on an Intel NUC is particularly successful and I’m experimenting myself with very promising results.
I can’t explain the science but it’s clear that there are several areas where the SQ can be elevated and I think that’s great.
Spent the morning listening to my SE playing to an NUC endpoint running Roon bridge (in Audiolinux) in RAM and it sounded terrific.
Now listening to the new mode on the SE using Squeezelite as an internal endpoint (to enable memory playback) and it sounds terrific too.
First world problems in terms of having to choose one!
And I’m not talking about you but I don’t understand why there are so many defensive posts in reply to the suggestion that Roon explore other ways of improving the SQ.
Roon is the greatest UI and user experience hands down. It’s not the greatest SQ tho and its a reasonable ask that they keep exploring how to improve it.
One problem is that loading the OS into RAM is volatile. I am experimenting with an AulioLinux endpoint at the moment and one problem is on the fly changes like plugging and unplugging USB DAC’s aren’t seen or saved. It takes a snapshot of what it sees on bootup and loads that into RAM. To make changes you need to reboot and re-enable the end point pretty much every time. It isn’t a practical solution yet in my opinion.
Some here asked about whether Roon playback to a squeezelite endpoint - configured with large memory buffers via -b - begins playback only after the song is fully cached.
The answer is no, but there is a huge burst of data over the network from Core to endpoint in the first few seconds (as much time as it takes to transfer the song).
I used some monitoring (iftop to be precise) to study this in a bit more detail. If interested, here is the link:
Another user on the same forum has even found that the ALSA buffer parameters (-a or squeezelite) can have a positive improvement to the SQ.
It would be great if we could have an option for Roon Bridge endpoints to have their RAM buffers filled with an entire song at the start of playback. If that would break multi endpoint synced streaming I’d accept that and I think many would if it’s an option.
Mhm, but it works brilliant, and I must say that …better than RAAT for me. The best end-point now in my system is Daphile PC linked to Roon Core (external server -> IP = CORE) with Matrix X-Hi and M2Tech USB DAC on route. In SQ competition easily defeat streamer for 2k.
I think, that any LMS implementation (a lots of them) could work w/out problems in Roon ecosystem like that. In any form. Hardware is not limitation.
Funny thing this. Discussion going on in Naim forum with some of the new ND streamer users saying that UPnP sounds sweeter than Roon and this could be attributed to the way they buffer songs to Ram before playback in UPnP but obviously not Roon. I preferred Roon myself when I tried comparing to UPnP but was not comparing same source PC as I cant run Minimserver on ROCK.
If you’re handy with OSes and command lines, it’s quite easy to try this. On the system connected to the DAC, download and run squeezelite. Ensure you have at least 4 GB of free memory. Run squeezelite with these additional flags:
-b 2097152:2097152 -a 52428800:4::
Point your Roon Core to the squeezelite instance instead, and compare SQ.
