Does Roon have better or worse sound quality than dCS Mosaic

My audio gear is simple, a hifiman hekse pluged into the dCS bartok streaming music from a QNAP hs543dx (the fanless NAS).

Previsouly I used Mosaic control the music play and found it easy to use. Several days ago, I installed the ROON sever on the NAS and applied for a 14 days trial.

Although the ROON is very powerful in terms of analyzing and managing your local music file, I do think it’s worse in terms of sound quality than Mosaic streaming through UPNP.

It seems quite intersting to me why is there a difference of sound quality. Anyone has a chance to make a similar comparison? Any comment is welcomed. Thanks!


About a year ago now I thought so too; that Mosaic sounded better than the same track played back via Roon (and even posted on this forum claiming so).

But in fact, they’re exactly the same - a local track played-back via Roon is bit-for-bit identical to that same track played by via Mosaic.

Perceptual bias is a real thing that actually effects our listening. The only way to know for sure if there’s real differences in the physical world is to measure it objectively - a platform like the dCS Vivaldi with it’s disaggregated components and features like Clone mode in the Upsampler, makes it very simple to objectively measure bit streams for differences.


Thnaks for your sharing.

Unfortuantly I don’t have a Vivaldi to make the test but the Roon and Mosaic do use different ways to stream the music, which are UPNP with Minimsever for Mosaic and RAAT protocal for Roon. I think we have ground to suspect that there may be a difference.

In addition, a bit-for-bit identical playback doesn’t alwasy mean the same sound quality, otherwise we don’t need to worry about the usb cable and ethernet cable.

Thats exactly what I thought too. Until I tested it and the unwrapped bit-streams were literally identical.

If you’re implying Jitter as the factor that makes otherwise identical bit-streams different, it doesn’t apply in the case Roon/UPnP as these are error-corrected and buffered streams. I can capture the bit-streams on a PCM recorder and the files are literally identical :smiley:

By the way, here’s dCS’ view;


Thank you for the feedback, very informative. Really appreciate.

1 Like

After my Qnap NAS died, I decided to get a Roon Nucleus. Not only is it my Roon Core, but it also holds my music on a SSD. The Nucleus is dedicated to just serving music. IMO, it sounds better than a NAS setup. Is the way I listen to Roon a better experience than a UPnP via Mosaic?..not sure. :thinking:

Yes, there are cheaper ways to do it (NUC etc…) but for me, the Nucleus is a lovely piece of kit, and works flawlessly with dCS. In my case, a Network Bridge, soon to be replaced by a Bartók. :+1:

1 Like

Likewise, I love my Nucleus+. Elegantly simple and high performance. As for Roon v. Mosaic, I’ve already expressed my choice, and for me, it’s no contest. I applaud the folks who run multiple front-ends. Their feedback is enormously helpful. But having done everything from JRiver (which I detest) to Minim to other UPnP solutions to Roon, I enjoy Roon. YMMV, and that’s what makes having choices so delightful. I have experienced superb, indistinguishable SQ from those other offerings as well. Why? Simple. Bit perfect files delivered over a reliable, low-noise network into an Ethernet port. There is no fathomable engineering reason why UPnP or RAAT should matter. If you think it does, or someone tells you it does, you have to ask yourself “why?”. When I pressed this specific point—why software clients on a network audio system should sound different—Paul McGowan, whom I’ve known for years and is a smart guy, carefully avoided answering (they’re developing a new server and software over there that looks interesting, but it’s not network audio . . . yet). The answer is pretty easy. If you believe network switches all over the world don’t change the sound of a bit-perfect file (and Paul once did a video on that very topic), it’s difficult to maintain that software clients should matter.

I subscribe to the notion that “everything matters.” But one has to be pragmatic. Does cosmic radiation matter much for assessing what temp to cook my steak? Probably not. We can assert that cables sound different, and some people’s ears can prove it in the analog realm, just as some measurements show why some cables sound different from others. I subscribe to that notion. But to me, in the digital realm, particularly Ethernet-based audio, the burden of persuasion is different. It’s easier to demonstrate and verify that the digital information is being delivered accurately to the DAC. Silver and copper aren’t going to affect the validity of that data. It might be possible for a cable or a port or PSU to affect EMI/RF or other noise entering the system, but not the integrity of the musical data itself (unless it is woefully non-compliant). So, if one really does hear something, ask yourself why. It’s far more likely to be perceptual bias, expectation or confirmation bias, or some wholly unrelated problem in the system.

First, @Anupc did not say that bit-perfect and SQ are the same thing. But once one has normalized the delivery mechanism–testing on the same USB setup or the same Ethernet setup—then bit-perfect is in fact a surrogate for SQ in this context. And client software and cables are two very different things. Client software only has one job: controlling what files go to the player. If the file is bit perfect, then it means the client software has not changed the file. If one changes things about the music in Roon or some other software, then the file is no longer bit-perfect; it is different. And as Anup has mentioned, it’s an easy enough test to verify. USB and Ethernet are not the same. And I would submit you don’t in fact need to worry about the Ethernet cable so long as it is certified compliant and unshielded (on dCS equipment). But even if you subscribe to the notion that compliant digital cables can alter SQ (without altering bit-perfect status), nothing about that validates the notion that client software can affect the SQ of a bit-perfect music file. It might cause you to ask, but it hardly proves it.

I know that, as audiophiles, it is so easy to get sucked into the notion that “just because we don’t know how to measure it, doesn’t mean I am not hearing it.” And I agree! But that is only the starting point. It’s not the answer. The answer is the “how” and “why.” Remain skeptical. In both directions. But enjoy the music.



I have exactly the same NAS, a dCS Network Bridge and a Vivaldi DAC.

I run Roon on that NAS, the cables are Audioquest Vodka, I use a Sotm router.

I must say that I can’t hear any difference between Roon playing a song from the NAS and the same song (file) coming from Mosaic (Minim server 2 on the NAS).

However, when I play a song from Tidal or Qobuz being from Mosaic or Roon it plays louder than the same song coming from the NAS…

1 Like

In my system with DCS NB, sound with Mosaic upnp is much better than Roon Raat.
More air and resolution.


1 Like

Interesting. Kirill, can share your full system configuration (including the Roon and UPnP setups), as well as your steps to ensure no volume differential? With dCS advising that there are no differences between the bit-perfect delivery to the DAC, there should be some explanation other than UPnP/RAAT.

My system: Intel NUC with hdplex lps->Keenetic Giga with hdplex lps->optical cable->Sonore optical module with hdplex lps->DCS NB->Goldmund digital processor->Goldmund Telos amps with built-in dac->loudspeakers.
I tried Roon Rock several times but always went back to Linux with Minim server.


1 Like

Thanks for the additional info.

Me too. Tried Roon twice but sitll went back to UPNP (Mosaic). Bought three months’ Roon at $1 last week so keep trying at the moment.

I think I go back to UPNP not only becuase of the sound quality (assuming there is a difference) but more importantly, Roon doesn’t support the folder view, which they said they would never support.

1 Like

Just a couple of general Roon questions as I always keep an eye on it in case one day I will change my mind and try it. Can anyone offer any advice on:

  1. Roon integrates Qobuz and your local music library. As a classical music listener I don’t want this because classical titles are duplicated hugely. So if I search for Beethoven Symphony 5 I would see the 6 or 7 different versions I own plus maybe 50 others from Qobuz. Can you get the application to show only your own library or only Qobuz?

  2. I keep an eye on Roon Community metadata queries and although there seems some improvement there still seem to be a majority of posts about classical music attributions which concerns me it being a minority genre so they should be amongst the fewest instead. Any comments from any mainly classical music users of Roon?

Pete, the answer to your first question is “yes.” There is a Roon search results toggle for only showing your library. Even if you don’t activate it, search results display an icon indicating whether an album is in your library, from Qobuz/Tidal, or both. Like you, I have multiple versions of many classical and some rock recordings. They’re pretty easy to identify in Roon.

As to metadata, I think Roon is in fact much improved for classical, but not yet where fans want it to be. I do find it easy to override metadata locally, and my current classical library is nicely indexed and complete, but I don’t have anywhere near the library many classical fans do, so I am not expert enough for you on this topic.

1 Like

Thanks Greg.