Sound quality Qobuz/Roon vs. Qobuz direct

All looks fine there. Roon on Android always resamples to the freq the device reports it supports under the Android audio stack so in your case the DAC in it is saying I support 96/24 and Roon will upsample or downsample if higher to make it fit rather than let the Android OS do the resampling.

As for your PC it’s likely that you are hearing a difference. Roon unlike other apps is very active and heavier on CPU activity and this includes Qobuz app which is pretty benign. Roon do recommend to not have the core attached to audio equipment due to this to get the best from it. As excessive CPU activity can affect sq and not all pcs usb sections are the best for audio. However It does seem to be rather DAC dependant though. I use my core attached to my RME DAC and it doesn’t sound any different to having a lighter device such as a raspberry pi running Roon bridge or the Bluesound Node 2i I used to have. Others report they do others don’t. Lots prefer Roon to the native apps and others don’t it’s a very up down scenario with lists of variables it seems. Your not doing anything wrong but separating might make a big difference . I do use ROCK which does cut down on activity outside of Roon as it’s designed to run Roon and Roon only but I don’t compare to other apps as they don’t do what I want , run headless like I want and work across my entire range of kitnb. Your DAC is supposed to be pretty good I believe at rejecting extra noise but perhaps it’s not as good as it should be or Roon really is messing with stuff but it’s a bit perfect stream do not sure how.


Yeah I’m definitely aware that they recommend a separate machine, and I’ll get there in a month or so–looking at an i5 NUC10. But Roon is absolutely not taxing this beefy box (6 cores/12 threads, 32GB RAM, nVME). When playing locally, Roon’s CPU utilization hovers between 0.17% and 0.40%. It does use about 1.2GB of RAM, but I still have 16.5GB free so that shouldn’t be an issue either.

Interestingly enough, I was playing around with the DSP this morning, and found that the Audeze preset for the LCD-X actually sounds really good with my setup. So if all else fails, maybe I’ll just do that. :slight_smile:

Maybe interesting for you too in regards to headphone profiles in Roon next to Audeze: + RooHead (not ready so far) claims to allow to select from several headphones.


Hi there,

You will find similar reports across various subforums here.
Different hardware / platforms / apps.
Root cause unknown.
Some imply Roon to be bit perfect and other software to add DSP coloring. However that would be a rather widespread issue it seems.

Nevertheless, you just joined a great / diverse/ open minded / friendly Roon forum community with great people, probably the best audio forum out there.

Welcome to Roon


RooHead does sound really cool, I’ll keep an eye out for release announcements!

1 Like

Veering off topic sorry but I suspect Roohead is going to feature the auto EQ headphone filters available on GitHub. There’s a good thread about them here:


@killdozer Not off topic at all. I downloaded and applied the convolution filters for the Clear, and it made a world of difference. Now, turning on my EQ curve on the DAC/Amp (as a test) makes the sound muddy and flat. I take that as a good sign that the filter is damn close to the mark.

I haven’t done an A/B against Qobuz with EQ enabled yet, but my gut says Roon with the convolution filter enabled (and no additional EQ) is now the better-sounding option. Opportunity for some minor tweaks? Perhaps. But much closer to ideal.


I saw this in Roon weekly and would like to add a remark in the confirmative. I have tested Roon against Audirvana. Last month. It was using a 2.4Ghz as roon core.

No doubt, despite Roon reporting perfect signal chain, Audirvana was several steps better, in an direct audible manner. More air around voices, more “blackness”, audibly tighter/deeper bass. (I have full output down to 18Hz)

My setup is without any DSP and for both Roon and Audirvana I used the fresh and default available after install. No tweaks of any kind was done. Others have reported the same on Audirvana forums.

I dont have time to investigate at the moment, but if some installs Audirvana trial and experience the same, one clue could be to investigate and compare the respective default settings and spot a difference. But it could also be things like different handling of the USB dac output, etc.

The machine is a macbook pro 2013 running Windows (much better than native windows machines as a digretion). And then perhaps Roon is less compliant to older USB outputs? These were my thoughts…

Unfortunately (for me at least), while the Audirvana is better sounding, it is just too buggy in their remote systems, especially the Android remote is too immature for practical use at the moment. The user comfort of Roon is in another league.

I decided to stick with Roon solely because of those bugs. But I will probably check in on Audirvana in half a year or so. I was intrigued!

The question is, of course, how two bitperfect streams on the same “core” device can sound so different. Audirvana HAS done something that Roon has not.

On Mac or Windows? On Mac I prefer Roon no doubt.

Edit: Nevermind. Windows on a Mac…

That has always been my theory.

Doesn’t explain this thread that @Protyreus alluded to -

The Lumin rep must have gotten bored with that thread or just couldn’t be bothered with me as I never did get a reply to this -

I find Audirvana 3.5 to be extremely bright compared to Roon on my Macs. In my setup this is not complimenting the sound in any way. Just too bright but it may sound awesome in other setups.

Well @danny might definitely like to contribute here and clarify some prejudice.

I tried Audivarna on my laptop, Roon core is on rock on a separate server so different hardware config so I can’t compare server to server but comparing playback on the same laptop via Audivarna and Roon in both cases via Dragonfly DAC. Both sounded different and I didn’t like Audivarna one bit. To me it stood out as sounding false and processed I would swear they are doing some DSP but can’t prove it. Will have to connect my RME and do a bit perfect test. It’s hard to compare both properly though without countering for differences in volume and that our it’s not a seamless comparison so bias can come in quite easily either way.

I am also using an RME, and a well reputed DAC. After reading these comments I stick to my own, that the difference could be related to e.g. difference in handling of the USB chip on the particular device, or similar, and that therefore, we perceive results different. In my setup, Audirvana is not bright, but more natural sounding.

I have a newer macbook 19 pro as well, and also this machine has both WIndows (bootcamp) and MacOs installed. They are separate partitions on the disk and I can start either. On one occassion, I found the newer mac better sounding than the older one, but this is a premature observation that really needs investigation. And again, some difference in the laptop hardware could be the cause.

I find Audirvana sounding differently when turning on exclusive mode on my macbook smh… but yeah I agree with you. Audirvana does change the sound. If you like it or not is up to you. Audirvana 3.5 is a good Software with some minor flaws imo. Audirvana Studio is just crap imo… I dont like it at all.

1 Like

In terms of bit perfect output, no - unless it is misconfigured (usually not exclusive mode may imply the OS mixer is destroying bit perfectness). If you are talking about CPU / network utilization, every software is different.

This can be easily proven by using RME proprietary bit perfect test, HDCD album to HDCD DAC, MQA album to MQA DAC, etc.

For a notebook computer, also check whether you hear a SQ difference when it is charging vs the power supply unplugged from both the notebook and the electricity socket.

Over all these years your posts indicate you refused to accept SQ difference from bit perfect streams. The truth is out there.

The key thing is to realize SQ is an analog thing. You are not hearing bits, but an analog conversion of the bits.


But it’s the same DAC converting the same bits, right? That should be a well-defined and repeatable process.

On the exact same device chain? Yes, that’s what I don’t understand.

That’s OK, doesn’t matter even tho you chopped my question off where it suited you.

I’ll leave this SQ discussion with a quote from the CTO: “Have you ever heard that sound that car stereos used to make when a GSM cell phone was about to ring?”

1 Like