Playing Roon over RAAT - Does the server matter for SQ?

[miguelito]

I am sure this has been asked before - I might have asked for all I know...

 

Chris's new sotm motherboard review brought this question back up for me...

 

Has anyone experienced a difference in sound quality between various servers when playing over the network - specifically Roon's RAAT?

 

I am always assuming that Roon's core is running comfortably, not hitting CPU or network bandwidth limits.

 

I have tried a few such servers: mac mini, mac laptop with macOS, mac laptop running Ubuntu (natively, not a VM), mac pro with macOS, NUC10i7 running ROCK. I can't say I hear any SQ differences.

 

What are people's experiences?

 

Thx.

 

Miguel

[firedog]

15 hours ago, miguelito said:

I am sure this has been asked before - I might have asked for all I know...

 

Chris's new sotm motherboard review brought this question back up for me...

 

Has anyone experienced a difference in sound quality between various servers when playing over the network - specifically Roon's RAAT?

 

I am always assuming that Roon's core is running comfortably, not hitting CPU or network bandwidth limits.

 

I have tried a few such servers: mac mini, mac laptop with macOS, mac laptop running Ubuntu (natively, not a VM), mac pro with macOS, NUC10i7 running ROCK. I can't say I hear any SQ differences.

 

What are people's experiences?

 

Thx.

 

Miguel

Same for me. Switched from a fanless server to a quiet one with fans and didn't hear any SQ difference.

[Lowlands]

Ditto. Moved from older i7 spec HP (plain fans cooled) COTS server in basement to newer i9 spec (water cooled) DIY server. With just plain Roon core & Raat to the same mRendu via the same ethernet LAN, I could not hear any difference between servers.

[miguelito]

18 hours ago, miguelito said:

I am sure this has been asked before - I might have asked for all I know...

 

To give a little more color to this... I wonder if using something like an Innuos server with this configuration (ie Roon over RAAT) would yield any improvements. 

 

I once asked a reviewer (someone I don't know personally) if and why there were any differences... He said he hadn't heard any... And I asked why he had not said so... He said if there are no differences it is not interesting to write about... Huh? It is extremely interesting to write about! I think he's concerned high end server manufacturers might get on his case and ask for their loaners back! So sad...

 

But regardless, I want to know if anyone has experienced any differences in this configuration.

[Lowlands]

I guess it is not illogical using RAAT via dedicated end-point over a LAN (with potentially some fiber optics in it) moves SQ influence from the server to the end-point. If you combine server + endpoint in a single box in your rack (amongst other kit and cabling) and plug that into your DAC, it makes sense it should matter more to SQ. Most Taiko Extreme owners put it in their racks, not in the basement (afaik). Also: anything in the rack should look like an audio component and not like a computer.....

 

In a split [Server] ~LAN ~ [dedicated endpoint] topology, server criteria for a reviewer to write about can still be simplicity of install & operate (plug & play factor), user interface, compatibility / versatility to serve different eco-systems, support for 3rd party streaming services, performance, heat & noise production, power usage, remote support, warranty, future-proof-ness, build quality, etc. etc.

 

So as long as a reviewer does not make up statements on SQ improvements or differences were they cannot hear any, I would say why not?

 

 

[Yiakubou]

The only time I heard different (better) SQ with Roon Core was when I started to run it on Euphony OS and quite a significant difference it was! Before that on ROCK, Win 10, gaming PC's, NUC's and no difference.

[kennyb123]

On 10/13/2022 at 10:31 AM, miguelito said:

Has anyone experienced a difference in sound quality between various servers when playing over the network - specifically Roon's RAAT?

Many years ago I ran Roon Core on a Mac Mini. The endpoint was a microRendu via RAAT.

 

One of the most significant upgrades up to that point came from powering the Mac Mini off an Uptone JS-2.  
 

I think it’s important to consider what changed and what didn’t change when those moved to different servers and hear no difference.  If one server was equipped with a significantly better power supply I would expect that a difference would be heard.  

[thijazi]

I've seen a notable improvement in sound quality when I moved from a HTPC connected via USB to my DAC to a ROCK NUC build which sits far from the Hi-Fi rack (ethernet connectivity streamed via OpticalRendu).... The other time I noted an improvement was when I stopped using the NAS as the location of the FLAC files that the Roon Server was pointing to and instead connected the library directly to the ROCK NUC via a USB hard drive, I found out that this was also the recommended way to set this up according to the Roon literature.

 

So for me, yes, I noticed improvements in these steps.

[miguelito]

5 hours ago, thijazi said:

when I moved from a HTPC connected via USB to my DAC to a ROCK NUC build which sits far from the Hi-Fi rack (ethernet connectivity streamed via OpticalRendu)

That improvement is absolutely expected.

 

The improvement from NAS to USB connectivity - I assume you heard this improvement with the NUC, as you say, not HTPC+NAS to NUC+USB disk?

[miguelito]

14 hours ago, kennyb123 said:

I think it’s important to consider what changed and what didn’t change when those moved to different servers and hear no difference.  If one server was equipped with a significantly better power supply I would expect that a difference would be heard.  

Setups:

1- Mac mini (i7, 2012, 16GB RAM, 4tb SSD with music over USB) powered with an Uptone JS-2 + macmini MMK kit

 

2- Mac Pro 2010 on Mojave running Roon Core (at the time 1.7 and 1.8) + files on internal SATA drive

 

3- Macbook pro M1-max running Roon Core + files in USB HD

 

4- Macbook pro 13" 2013 running Ubuntu 20.04 + linux Roon Core + files on USB drive

 

5- NUC10i7 on ROCK + files in internal SATA SSD

 

All of the 5 above sound the same to me over RAAT to a dCS Rossini. In all cases network connectivity to the Rossini is over an SFP-Ethernet bridge (so optical to the bridge, ethernet to the Rossini).

 

I don't hear a SQ difference in 1-5 above.

[thijazi]

19 hours ago, miguelito said:

The improvement from NAS to USB connectivity - I assume you heard this improvement with the NUC, as you say, not HTPC+NAS to NUC+USB disk?

Yes, I had the storage attached directly (USB) to the NUC.. It appears that when I had the Roon storage configured via a NAS drive, it worked but slowed down things, and some of that extra work the Roon Server had to do may have impacted sound quality. When I connected the drive directly to the Roon Server things improved in a notable way, plus the Roon UI became far more responsive and less sluggish.

 

[Kal Rubinson]

On 10/15/2022 at 9:30 AM, miguelito said:

I don't hear a SQ difference in 1-5 above.

With an entirely different array of componentry, I have the same observation.

5 hours ago, thijazi said:

It appears that when I had the Roon storage configured via a NAS drive, it worked but slowed down things, and some of that extra work the Roon Server had to do may have impacted sound quality.

IMHO, something must be amiss/faulty/suboptimal in your connections (or the hardware in your Roon server host) for this to to be the case.  

[miguelito]

On 10/16/2022 at 4:42 AM, thijazi said:

Yes, I had the storage attached directly (USB) to the NUC.. It appears that when I had the Roon storage configured via a NAS drive, it worked but slowed down things, and some of that extra work the Roon Server had to do may have impacted sound quality.

Ok... Over RAAT I am not sure I understand the reason - I am assuming this comparison is with the DAC/Streamer connected over the network and playing over RAAT in all cases. If you're talking about a USB connection, then that's different and I could easily see a difference there.

 

As a side note, I will say that I have played the same files over Roon RAAT and also using MinimServer to serve the file and play it using dCS Mosaic to the dCS Rossini. The difference in sound was remarkable. SAME FILE (eg Shuggie Otis's Purple). Even the soundstage was different. Very strange.

[stefano_mbp]

36 minutes ago, miguelito said:

I will say that I have played the same files over Roon RAAT and also using MinimServer to serve the file and play it using dCS Mosaic to the dCS Rossini. The difference in sound was remarkable. SAME FILE (eg Shuggie Otis's Purple). Even the soundstage was different

Which is better, Roon/RAAT or Minimserver? … using Minimserver here and I’m curious …

[miguelito]

56 minutes ago, stefano_mbp said:

Which is better, Roon/RAAT or Minimserver? … using Minimserver here and I’m curious …

I have not done a thorough comparison lately but my impression is Minimserver is better.

[thijazi]

On 10/16/2022 at 6:02 PM, Kal Rubinson said:

With an entirely different array of componentry, I have the same observation.

IMHO, something must be amiss/faulty/suboptimal in your connections (or the hardware in your Roon server host) for this to to be the case.  

 

Could be... I did speak with Roon about this, and they are the ones that shared with me that their recommendation was to avoid the NAS route if possible... So I did..

[jhwalker]

Just for contrast - I run Roon Core on an old Mac Mini connected to a WiFi AP via Ethernet, the files are stored on a NAS accessed via WiFi (at the other end of the house), and the endpoint is an Intel NUC sending multichannel to my receiver via HDMI.  Sounds great ;) and rock solid.

 

Not to say it couldn't be better if I had a dedicated streamer + DAC as endpoint, but I wouldn't expect it to sound MUCH different - biggest difference will always be the final digital > analog conversion and amp.

[SQFIRST]

On 10/17/2022 at 8:13 AM, miguelito said:

As a side note, I will say that I have played the same files over Roon RAAT and also using MinimServer to serve the file and play it using dCS Mosaic to the dCS Rossini. The difference in sound was remarkable

 

Great topic and it can be very subjective and complicated to understand all the nuances involved. Sharing some of my experience.

 

My endpoint is a Lumin X1 Streamer+DAC. It uses the Lumin App which uses MinimServer for local storage, all over upnp. Roon works with the Lumin also. I have no experience with Apple Macs or with the DCS so bear that in mind.

 

As you noted the difference in SQ between the DCS Mosaic and your Roon setup, I too immediately noted the gap in quality between my Roon and the Lumin app when I first started using the Lumin. I focused my efforts to get the network to optimal level supporting the Lumin app and the direct play of the Lumin X1. Once this was satisfactory to me, it is only then that I focused on the Roon situation. I prefer using Roon over the Lumin App for two reasons:

-Music discovery with its layered, 3D UI approach

-Ability to use Convolution filter for that last uptick in sound quality allowing some room correction. Lumin does not support HQplayer.

The Lumin App is my go-to for reference level evaluation.

Factors that I found to be impacting Roon RAAT:

Platform:

For my ears, Roon with Windows based audio sounds similar to what the Lumin App does. Archlinux based audio did not. Again, this is just my ears, not suggesting any absolute here. Since I am using the Lumin X1/App as my reference, my quest was to get Roon to sound at least as good. Even within Windows, different versions and even different releases of the same version had sound differences! This is a big topic by itself but I bring this up simply to highlight the possibilities that impact Roon RAAT sound.

Hardware:

Lowest CPU involvement and disk access yields best sound. Less is more. I use ROON without any DSP, other than Convolution, and keep everything unchecked that engages CPU, additional cores or higher usage. This is immediately noticeable in a revealing system (which includes network quality).

Networking:

The balance between isolation and speed. I have had success with Wifi isolation (as jhwalker noted above) but it caused minor speed issues that went away with wired isolation. Fiber is essential at the right junctures. Less is more rule applies and minimizing connection points favors sound quality. My source components, including the Roon server is upstream behind fiber and the Lumin is after an EtherRegen. I use an external Clock with the EtherRegen and the clocking made a big difference to the entire network not just the Lumin.

Hearing about your experience I will suggest the possibility that there is an equalizing factor downstream of your Roon core that is making everything sound alike. However, since you do hear a difference between the DCS Mosaic and Roon, this is good news (and bad news too) as you can focus on getting Roon up to that level. If you want to. :)

[miguelito]

On 10/18/2022 at 10:48 AM, SQFIRST said:

For my ears, Roon with Windows based audio sounds similar to what the Lumin App does. Archlinux based audio did not. Again, this is just my ears, not suggesting any absolute here.

Confirming: Over the network (you said RAAT which would imply that)? As I said previously, I fully expect sound quality differences if playing Roon over USB, but over the network... I have never heard differences when playing over RAAT.

 

The sound quality diff between RAAT and UPnP-served files played with dCS Mosaic varies by track, but there was one track of the lot I tested which was bizarre: Shuggie Otis's "Purple". The soundstage sounded different. Reproducibly, no need for blind testing. Bizarre. 

 

My networking config is tip-top: gigabit wired for Roon core machine (NUC10i7 running ROCK), Minimserver machine (mac pro), and the SFP/Ethernet optical bridge going to the DAC (dCS Rossini). On the Rossini side there's a gigabit SFP/Ethernet bridge powered by a linear power supply going to a SOtM Isocat ethernet filter which connects to the Rossini's ethernet port(s). I have tried both gigabit and megabit Rossini ports but hear no difference between those two when playing RAAT.

 

I am not sure what can be improved in this situation, but I would argue the main difference is unchangeable: Via Roon RAAT, the data is sent from the core to the Rossini which uses some form of a Roon-supplied player in the Rossini network board to play the file. When using UPnP file streaming, the Rossini will source the file data as-is and will use some internal player (almost surely different from the RAAT code) to play the file inside the network board. I would in principle chuck the difference to these two apps. 

[Yiakubou]

In every Roon setup, the audio file playback is performed by Roon core. Endpoints only receive the final audio stream that Roon bridge receives and turns into PCM / DSD / whatever... Every Roon certified device has Roon bridge in it (plus some additional integration I believe). Even when you run Roon from a single machine and just send the audio to DAC e.g. via USB, the way it works is the same - Roon core does the playback, this is then transferred via RAAT to a Roon Bridge component that turns it into digital audio signal.

 

In every streaming setup I tried, the quality always degraded when Roon was in the mix. Either it does something weird during playback, or there is something in the RAAT protocol, but that is my experience. Even when using it with HQPlayer, it's still worse than using HQPlayer alone - again because Roon pushes it's own continuous audio data stream to HQP, it's not just a remote HQP controller.

[musicjunkie917]

34 minutes ago, Yiakubou said:

In every streaming setup I tried, the quality always degraded when Roon was in the mix. Either it does something weird during playback, or there is something in the RAAT protocol, but that is my experience. Even when using it with HQPlayer, it's still worse than using HQPlayer alone - again because Roon pushes it's own continuous audio data stream to HQP, it's not just a remote HQP controller.

 

This makes zero sense. Unless Roon DSP is involved, Roon is sending bit perfect data to the DAC if directly connected to the Roon server or bit perfect data to the Roon endpoint. Bit perfect! When Roon is used with HQPlayer, HQPlayer gets the data from Roon and then sends it to the DAC if directly attached or to the HQPlayer NAA. HQPlayer does this...not Roon. HQPlayer is not being messed up by Roon.

 

Now, if you want to say that the RAAT protocol , which uses TCP, is somehow coloring the bit perfect data sent to the Roon endpoint, you need to explain this in some way that makes sense. When Roon sends bit perfect music data to an ultraRendu acting as a Roon endpoint, and then he ultraRendu passes that out the USB port to the DAC, again in bit perfect form, how is that RAAT is changing the sound?

 

When using an HQPlayer NAA, the NAA has no idea how the data got to HQPlayer. It only receives a stream of data from HQPlayer. How, pray tell, does that change the sound quality of what is coming out of the DAC when it gets the exact same data from HQPlayer either way?

[Yiakubou]

I don't know the answers, I'm not a Roon developer. I'm just describing my experience based on comparisons made on my setup. And I've been testing lots of setups in the last year or so, listening and comparing each of them quite extensively. Everytime Roon was in the mix, the quality got worse. I don't know how to explain it. And I don't think I have some expectation bias here because I actually wanted Roon to work in my setup due to its superb interface. The only thing that improved Roon sound for me was when I was running it on Euphony OS. Apparently bit-perfect is not the only thing to look for. I believe it also matters how the data is processed before it's sent out to an endpoint.

 

One example that makes me think that there is something wrong with Roon sound quality: I've ended up with a dual PC setup. Both PC's are running on Euphony OS to make things even. Both PC's are Intel NUC's, powered with exact same LPS. 1st PC works as a server - either in Roon mode or as Euphony's Stylus player. 2nd PC runs in HQP mode and connects to USB DAC or an endpoint via NAA. When running Roon -> HQP, the sound quality is noticably worse than when running Euphony Stylus -> HQP. The only difference between the two scenarios is that in the Roon/HQP setup, Roon sends it's own continuous data stream to HQP. In Stylus/HQP setup, Stylus only acts as a remote control and it's HQP PC that downloads and buffers the stream before playback (either from local files or stream services). Logically, that makes me think (and hear) that there's just something wrong with Roon streams.

[SQFIRST]

1 hour ago, miguelito said:

Confirming: Over the network (you said RAAT which would imply that)?

Yes. The Lumin X1 is ethernet based, not USB.

 

1 hour ago, miguelito said:

I would in principle chuck the difference to these two apps. 

It sure seems that way as I too had experienced the same difference and was describing it as how you and some of the others here have.

 

However, in my experience, after having spent way too long on this effort, it turned out to be a case of digital audio age challenge and was a combination of the factors as I mentioned above. Roon can sound as good as your baseline system. Is it easy to get it there? I will have to say, No. But just wanted to add my perspective and this is a good topic to hear everyone discuss.

 

In the end this is about personal music enjoyment and it doesn't matter how as long as we are enjoying the music!

 

[musicjunkie917]

23 minutes ago, Yiakubou said:

Logically, that makes me think (and hear) that there's just something wrong with Roon streams.

 

I think there is some kind of expectation bias issue here. Especially with the HQPlayer setup with the NAA. There must be zero difference there as the NAA has no idea where HQPlayer gets the music data and it gets the same exact data whether Roon or HQPlayer Client is doing the playing.

[Yiakubou]

Forget about NAA. I was testing it also while having the HQP PC connected to USB DAC as mentioned. And it does not matter anyway. Again the only difference between the two scenarios mentioned was in Roon continuously streaming to HQP vs. HQP grabbing the data files on its own, not relying on any external continous data stream. Whether HQP was connected to USB DAC or via NAA to an endpoint does not matter in this comparison as that was not changing when switching between the two setups.