Reasons for SQ difference Roon vs. Mosaic?

There have been a number of threads about SQ differences between Roon and Mosaic. Today I have done a few comparisons and I have again heard a difference in SQ when playing via Mosaic vs. playing via Roon to my Rossini / Rossini Clock (via Ethernet obviously). Via Mosaic soundstage is better deeper, instruments are more clearly defined, the space between the instruments is more silent. Using Roon instruments sound more blurred, as if some artifacts were present besides the instruments.
I have been able to reproduce the same using an Odroid N2 → USB → Neukomm DAC. Using native Volumio 3 on the Odroid sounds better than streaming via the Roon plugin in Volumio. The characteristics of the differences are similar as in the dCS case.
To me this difference is not subtle. Which reproduction one prefers is of course a matter of taste.
The interesting thing here is that in both cases the hardware involved is the same. Same Ethernet port on the Rossini, same upsampling, same filters. In the Neukomm case, same Odroid, same Volumio, same USB interface, same DAC settings.
I am scratching my head what the reason for this difference is. It is hard for me to believe that both configurations would be bit perfect, yet sound different.
Can someone (maybe @James ?) shed some light on how these differences could come about? Second question: What would need to be changed in the Roon case to make it sound like the Mosaic use case (because to me Mosaic sounds much better, but Roon has the better user interface).

1 Like

Hi Rudi,

I try not to jump in on threads like this where possible but as you tagged me I’ll give it a go. :slight_smile:

I do not know of any technical reason why there would be any sound quality difference between Mosaic and Roon. Both are providing asynchronous bit-perfect data to the Rossini from presumably the same source material (are you streaming from TIDAL, playing locally stored content…?). Provided you aren’t using Roon’s DSP (like EQ), no reason for one to sound different to the other.

At the risk of sounding rather cheeky, have you tried this in a blind test, with someone else operating the apps? Last thing I want to do is tell you that you aren’t hearing something you are as that is ultimately down to your ears/brain, but I can say there is no technical reason why you would be.


Thanks James,
No, I did not do a blind test, let alone a double blind abx test.
With all the buzzwords (RAAT etc.) surrounding Roon, I should really be biased towards Roon sounding better than straight ethernet
To me the difference is not subtle, especially what regards soundstaging. And yes, DSP is turned off in Roon.
But I will try to do a blind test with a friend and come back with my findings.
The only rational explanation I have is that Roon is not bitperfect, even though Roon are adamant about the fact that they are.
This afternoon I will be getting a Melco N1A EX for doing the same comparison. The Melco can play from its internal disks, from NAS via MinimServer and through Roon

Our own @Anupc has demonstrated that Roon is bit perfect. The fact that over on the Roon forums, and even here I believe, half of people think that non-Roon sounds better, half think that Roon sounds better, and everyone else is enjoying the music, is supportive of what @James suggests: this is all in the ear and brain of the listener. This is the sort of thing that is in fact quite amenable to double-blind A/B/X testing. But somehow I just don’t think anyone really wants an answer, more like an eternal disputatious quest.


Rudi, it’s actually trivially easy to prove for yourself that Roon is bit perfect transport from source file all the way to the DAC:

  • Create a single pure-tone PCM .wav file of whatever frequency you fancy - 1kHz probably works best.

  • Use Roon to playback that track and capture the output, either as PCM out of the Upsampler (in Clone mode), or as an analog recording of the DAC output.

You’ll see it’ll be perfect.

There are other (slightly more complex) ways to do the same thing, even with music files, and utilising Roon Bridge on a PC/Mac. But all of them will just prove that Roon is in fact bit perfect :grin:

1 Like

Thanks guys. I accept your proof that Roon (and I presume streaming from NAS via MinimServer) to Rossini via Ethernet is bit perfect.
I will investigate further by doing a blind test with a colleague to hear whether the differences I believe to hear can be reliably identified under those conditions.


Did you ever try a blind test Rudi? I did, see my results here.

I must admit that I did not. I have simply stuck with using mosaic for critical listening.

Same here…

same here…roon adds something i don’t like

Roon adds a lot of stuff we may not like. However once all the dsp is turned offf, Roon does deliver bit perfect data transfer, which is error corrected if delivered via ethernet.

1 Like

All of my Roon DSP is turned off, other than…

when I’m listening to DSD 256 files, for which I currently have no option but to listen via Roon in order to downsample…

I really hope that will not remain the case for much longer :crossed_fingers:


i hear you - i have a handful of dsd 256 albums with 128 duplicates. I’d love just to clean up my library!

Currently, I have 111 DSD 256 albums…

Aside from HDTT, there are an ever increasing number of new recordings in DSD 256.

I have no doubt that dCS are aware of this, hence I’m hopeful that, in the near future, we’ll all be able to avoid the necessity to utilise Roon downsampling…

That would be an additional bonus!


I’m a Roon enthusiast, but there’s clearly something wrong in the way some of its hidden settings may impact volume, hence perception of sound quality.

I noticed by the past that the volume of the exact same track was louder if played from Mosaic than from Roon.

I finally turned those two options on to get the same volume with Mosaic and Roon:

Hope this can help,



Thank you @Erwan
Your post goes a long way to explain percieved differences between playback via Mosaic and Roon. If playback via Mosaic is louder, it would obviously sound more impactful (=better)

I think these need to be set to NO if you want bit perfect playback

Has anyone here tried the new Roon Neucleus Titan ?

Is it worth upgrading ?


Default REPLAYGAIN_REFERENCE_LOUDNESS (let’s call it RRL) is 89dB for most of the tracks (-11LUFS (1LUFS = 1dB)).

If I read well, the Roon disclaimer clearly states it overrides this tag value if present and applies -14LUFS (hence 86dB for the RRL, 3dB less than the default RRL) unless these options are set to ON.

If no RRL tag exists, then Roon applies -18LUFS (-18dB).