SQ comparison: Rock vs Windows OS on Roon

After listening Rock vs Windows OS on Roon on the same setup, my conclusion, Rock is indeed a leap in sound quality. My impression is it has darker background, instruments and vocals are more refined especially at the edge and top ends. Everything seemed to snap in focus and some degree of tightness and overall control.

I think Rock and Roon make almost perfect combination for high quality playback. Well done guys!


What is your setup?
Connection to the DAC?

I’m using Intel i5 NUC 64GB SSD, 4GB RAM. Music storage from NAS. Both NUC and NAS are powered by a linear power supply. It is connected via USB to the Holo Spring R2R DAC. My amps are VTL 2.5i pre and monoblocks VTL-MB125 driving Dynaudio Contour 1.8Mk2 speakers.

Both Rock and Windows 10 with Roon are tested on the same hardware, just swapping the SSD with different OS. My opinion is Rock has probably the least threads and TSR running on the backgrounds, thus this creates less ‘noise’ compared to Windows based OS. Rock+Roon is more of ‘dedicated’ audio playback.

The direct connection of the DAC to the NUC could permit propagation of electric noise.
An IP connection, or a USB isolation box, might reduce that.

On the other hand, if you are happy with the ROCK setup, no need to tinker further.

The ideas that 1) general purpose computers are computationally busy and electrically noisy, 2) those factors negatively affect audio quality, so 3) separation of CPU and audio output device with distance and electrical isolation are key get bandied about in this forum a lot. However, ideas 2 and 3 may not find much support in objective science. Sure, exceptions exist, and caution has its place. But for a counterexample, something of a worst case scenario, see analog output test measurements of an internal sound card crammed inside an old, busy, noisy computer.



Perfect example of why archaic measurement standards don’t mean much in today’s audio.

Actually meaningful measurement techniques are preformed in house with many manufacturers today. However most of what you find publicly published today is mainly for marketing purposes. Why would manufacturers want to give away their trade secrets?

On this one, I thought I saw Brian and Danny mention they’ve seen these experiments during their time at Meridian.



I really don’t know enough but I do trust that they trusted the experiments they saw at Meridian enough to make these recommendations.

1 Like

However, a typical minimalist Windows 10 PC still have near to a thousand threads running on background (from Task Manager) and some TSR. Besides Windows update is a major headache that can sometimes screw up functionality of programs. Rock basically removes all these.

If I can think of another way, the latency, delay and timing issues do play a part. Having a minimum transition layers between the playback software and the OS also help.

1 Like

Most of those threads aren’t doing anything. In former jobs I looked at lots of kernel dumps and found many, many, many threads that had gone hours/days without actually running.

Science is, indeed, the best way we have to be certain that a proposition is true. However, what makes it so good at that also limits its reach. Not for all time, though: someday science will be able to come up with experiments that make falsifiable those out-of-reach propositions. Take the brain, for example. We don’t really know much more about the brain than things like “when we think about math, this general area of the brain gets a lot of blood and electrical activity.”

Despite such limitations, the role of science as the best truth-gathering tool we have gets confused as the ONLY truth-gathering tool we have. That confusion extends to the idea that if science can’t prove a proposition is true then that proposition must be false.

Is that what you mean? That if science can’t find a difference in the digital stream between a Windows machine and a ROCK machine, then no difference exists? If that’s what you’re saying, then you’re making a logical error. An absence of scientific evidence for a proposition does not prove that the proposition is false. That’s why the scientific method demands falsifiability.

If you’re simply pointing out that @MusicEar is making a cognitive error himself by believing that ROCK sounds better without the benefit of scientific proof, then you’re being rude. Do you believe that you love your family? Do you believe that a quiche made with older eggs tastes better then one made with fresher eggs? Have you ever believed that you were treated unfairly? How would you feel if the response to the expression of that belief was, “well, you can’t scientifically prove that belief, so I’m going to dismiss it”?

Next time you tell your significant other or child that you love her/him, be sure to ask yourself if you really believe that and why.


Did you know the earth was flat until science proved it was round? Before that point in time, the science of the day proved it was flat. The way I see most science is if it can be scientifically proven today, it’s damn near a guarantee it will be debunked once science advances.

Really! I always thought that the earth was ‘believed’ to be flat’ before it was proven to be spherical. Who proved that it was flat?


I’m sure someone did at the time. But the scientific community probably burnt those records :slight_smile:

Beliefs are often proven to be wrong by science. It’s rare for it to be the other way around.

It’s also very rare for the law of physics to be disproved.

It seems snake oil has been prevalent since the beginning of mankind. Until science proves otherwise.


This is all about improving the speed of your computer to avoid glitches when playing audio. It is not about making the sound “better”. For that, you will need a better DAC/amp/speakers, etc:

“All of the tuning tips below should be implemented if you want to achieve high speed, low latency audio recording and playback without glitches with your audio interface. If you are experiencing pops/clicks when recording or playing back audio please try following the instructions below. Please note that the steps below assume that you have the view in Control Panel set to ‘Small/Large Icons’ rather than ‘Categories’.”

I would stick my neck out and say that if science cannot measure a difference between the digital streams, then it would follow that no-one will hear a difference between the digital streams.

If anyone wanted to get to the bottom of whether there was a difference the digital streams could be captured and compared bit by bit.


So do you believe in 100 years the AP machines will offer the exact same functionality as today’s units? I bet the human ear already functions at the same level it will 100 years from now.

AP machines?

If abstract items such as bits in a digital stream are measured as being equal, then they are equal. It’s a mathematical tautology. Debating that means you don’t understand digital streams. If you are finding differences in the sound results, then you have either measured wrong, or you have external factors outside the digital stream of bits that are impacting the SQ. Note, that the bits are not being perceived by your brain, there are many many analog stages between the bitstream and your brain, and all may be impacted as a side effect of some other process.

For example, imagine if processing the bits one way used more CPU cycles than another way (and therefore more heat), and that heat caused expansion of some analog component near by. That might alter the sound. The bits can still stay bits and be bits all along, but the sound could be very different. Getting the bits right is a requirement (because garbage in, garbage out), but once the bits have been delivered, the analog components must make sure they don’t screw it up (and the DAC chip is an analog device (by virtue of not being able to error correct the digital side).

The crazy part in in the above example is that you could argue that “more processing power required” is equal to “bad SQ” in this one setup, but putting a more efficient CPU in there could also fix the SQ equally as reducing CPU load or moving the analog components far away. Which is right? The one that sounds good.