Nucleus + Sound Quality

Having your Roon core in another room is actually a plus. If your wifi works well stick with it.

It is very interesting to hear that the Nucleus + had such an effect over a network. Since the bits are getting from the Nucleus Core to the BDP2 endpoint using TCP via RAAT (it was UDP up until Build 234 Roon 1.3) I have always assumed that the packets and messages could be carried back and forth by carrier pigeon or smoke signal and not affect the sound. It is hard to imagine what influence the server could have through that interface.

Are you using shielded Ethernet cables by any chance @Billt1 ?


It’s not all left to imagination. Some DAC designers have shared their thoughts on this with me. After going into hardcore technical stuff one of them quoted: “it’s not obvious which is better: bypassing the audio stack of drivers or bypassing the network stack of drivers…” , among other things that went over my head but sounded plausible because I’ve seen multiple experts discuss the same thing.

There are experts out there that understand these possible technical mechanisms… and it’s far beyond the ‘bits are bits’ argument.

I was using my i7 Macbook as a Roon Server for a few months, to an ethernet isolated Allo USBridge. I recently took Roon Core duties off my Macbook and my new NUC7i7DNHE (quad core i7) running Roon OS (via ROCK) sounds noticeably better… I was surprised at the difference with my desktop headphones setup. I can’t hear any difference with my speaker system…

So it doesn’t surprise me to read that people notice a SQ difference with Nucleus, even over the network.

I no longer have my sonicTransporter i7 Roon Core (highly optimized for low latency performance and other custom OS optimizations) since I gave it to my old man, but I’d be semi interested to compare, only for shi!ts and giggles. I can’t say I’m in a rush though.

But a Roon supported NUC running ROCK or a Nucleus is definitely the way to go for best SQ (and other reasons long term perhaps), to my ears, with my experience (not expertise). Whether or not you can hear the difference to care enough (again - I can’t on a speaker system but I can with closed back headphones…) is another thing.

It’s all good fun though :grin:


One possibility is that there may be a difference in how “full” the packets are between different servers running different OSs. That could result in increased or decreased processing at the endpoint.

Agreed. My i7 Macbook was always doing a whole lotta other stuff, non Roon related, when it was my Roon Core.

My NUC running Roon OS has obviously been highly optimized by the Roon Team - they’ve shut down all processes that they can, that they don’t believe is required for Roon to function.

My Macbook is still where it’s always been and still doing a whole lotta other stuff but it’s now just a remote and the Roon OS --> Roon Bridge path/activity is now very optimized for Roon. Maybe this plays a part in perceived SQ improvement? Who knows.

One of the big differences between Roon OS and the others in this space is that Roon OS never “shut down” anything. It just doesn’t include them from the start. Roon OS built from scratch, not built on top of another server or desktop operating system with items that need to be disabled. Every option, every piece of software on this box is there on purpose. I’ve yet to see another Linux based audio device that works this way (ARM or x86/x64).


I’m guessing the primary goal of this is to reduce CPU loading, mainly for thermal dissipation reasons, especially Nucleus? i.e. reducing ‘waste’ (heat)?

And if there’s a bonus improvement in SQ (which my ears say there is, when using closed back headphones) then that’s a great bonus, but it’s not your primary goal?

This is why the tinniest part of me wants to compare with my old sonicTi7 (Fedora based for others that don’t know), for sh!ts and giggle only (there’s no other reason to care about this stuff). Just to hear if there’s any SQ difference there. My NUC7i7DNHE travels nicely in it’s strong slim fanless case, so I’ll take it with me the next time I visit my old man (dad).

You’re still talking about the Core though. It is hard to understand how a server can affect SQ through a TCP connection. No one would be willing to pay a premium to Tidal if they started offering servers with optimised power supplies. I think any explanation has to look at how what is being received makes a difference to how the endpoint is working.

But when you say “what is being received” doesn’t that have to include the Core and what it’s doing (and not doing)?

Only to the extent that there is a difference in the packets. If there is no difference in the packets, why would it matter whether the Core was running on a Beowulf cluster or a monastery full of abacuses (abacii ?).

Dunno. I’m not trying to say anything but at the same time I don’t know what you’re trying to say. I think it may go into deeper networking stuff, which neither of us has any expertise in.

I’m waiting for the usual suspects to wake up (timezone differences) and jump all over the mention of linear PSU’s with Tidal servers :grin: I think I know what you’re trying to say there but I don’t THINK that’s relevant here… but I have no qualification to say it’s irrelevant so I’m just waffling (as usual) :grin:

1 Like

This is probably a bad example, but remember how not so long ago seeing ‘galvanic isolation of USB input’ on marketing material for a DAC meant no worries (and some people still believe that).

Some DAC designers have explained that not all isolation methods are created equal (i.e. effective).

I guess the point I’m trying to make (doing a poor job) is that more will be learnt and shared about what’s happening in networked audio, over time.

I went to demo some Meridian DSP loudspeakers recently and Meridian marketing says that their DSP loudspeakers are immune to the environment.

But for some reason, a Bluesound Node 2 source sounded different (a bit brighter…) to my WiFi enabled and 5Vdc battery powered Pi2Design digital coax source… and the owner was mumbling away at an explanation (which tells me he heard a difference too) but the real technical reasons would be over his head, and probably mine.

There’s still a lot of unknowns in technical explanations for SQ observations unfortunately but we’ll learn more over time.

When using Roon over a network, the server platform really doesn’t matter. Roon pre-buffers approximately five seconds of audio to the RAAT endpoint, per the devs. This means any network hickups, jitter, etc are all negated, since the endpoint is playing from a huge local RAM buffer, and receives the same data from the server no matter what platform/OS the server is running on. The server is also doing all the heavy lifting, so any arguments about WAV vs FLAC are bollocks as well. And again, if there is a slight burp in processing because you are running the Roon server on a PC/Mac that is doing other things, it won’t matter because of the huge amount of audio pre-buffered to the endpoint.

Further, Ethernet is isolated; has been since day one of the spec. And 99% of people will have a switch or router in-between the Roon server and their endpoint, especially if they are using Roon for multiple listening zones. Thus, those that believe in the “leakage current” malarkey would have their sound influenced by the switch/router the endpoint is connected to, and not the Roon server system.


Hehe you’d think I would have learnt by now (sharing observations) :grin:

I switched from running Vortexbox OS to Using ROCK on the same system. Noticed no difference in SQ. I miss the ability to rip so may switch back. I have mulled over whether I should get a Nucleus but I honestly cant see the benefits it would bring given its hefty price tag especially since I keep my servers hidden away so silent machine is not a requisite. I have also switched to using local attached storage on the MOCK and not using my NAS. Again no change in SQ. Roon dont claim SQ benefits with ROCK or the Nucleus just that its simpler way of doing things thats specced for one purpose run Roon as efficiently and effectivley as possible.

Somewhat cynical view: some home networks are really bad (bad wiring, bad power, obsolete servers, buggy routers from ISP, obsolete switches, …). Replacing some of that junk with a new carefully built server and associated wiring makes everything better, which is then subjectively verbalized as better SQ :wink:

Today I installed a NUC with the Roon Core. Previously it was on an old gaming PC (in another room). The NUC was built by Quiet PC which in now on the HiFI rack. I also moved the Cisco switch close to the NUC. The Cisco feeds a DCS Bridge near by. So I changed more than one thing. Difficult to say precisely which change made the difference (I’m not one to speculate) but I have to say music has never sounded so beautiful and emotional. I’m delighted.



Sorry, this might be controversial, but I just don’t believe that human recall and is good enough to make a completely subjective assessment of the sound quality of things that can’t be switched relatively quickly from one to the other and done so blindly - especially when you factor in unconscious bias. In the big changes that people are making to their systems above, are they able to readily switch back and forth between old and new sources to compare directly, or is the comparison made with what they “think” the old system sounded like from memory?
Unless the sound quality difference is huge, I really can’t believe that a gapped listening comparison has any validity. Nor a non-blind listening test.

1 Like

I suppose what matters is someone has made changes which align with Roon best practice and his experience has improved. I am not sure anything negative can be taken from that. We all know the claims can be challenged in any number of ways but to be frank all I see is a happy Roon customer!


It depends of your definition of “huge”. Perception can easily fool you in many ways and the memories of an experience are never fully reliable.

But with this said some things have an impact that not just are heard and felt, but with an ease even though it’s not “huge”. I have yet to encounter this in servers, and I’m a bit sceptical that the difference would be easily heard or remembered. (providing you have a reasonable server and setup to start with)

With other audio equipment such as DACs I have had several eye opening (ear opening?) experiences that I later could easily replicate in blind tests with friends. One especially is when playing a quite nice Anthem 510 receiver with and without the original Chord Hugo. Even my friends who are not into music (using the original earbuds that comes when purchasing a phone) where blown away. But at the same time the difference wasn’t “huge” for most people, had it not been for the swap back and forth they wouldn’t have the interest to perceive the music reproduction as better. For most of them playing music from a small portable Bluetooth speaker at 85db is a nice musical experience.

Some people in this hobby fools themselves. Probably on both ends of the spectrum. But what I think is that we train this perception to a degree that sometimes just makes the hobby more expensive. For me it’s like food, I can’t eat much of the food served even at “decent” restaurants anymore. And it’s a shame sometimes. :thinking: