Roon Optimization Guide For Increased Sound Quality

[theveterans]

I, and many other audiophiles, do find that roon's DSP is audibly lossy. The fact that ASR has "proven" that Roon's DSP volume control is better that any analogue volume control in the world - is for me ample reason to stay well clear of roon's DSP .

I really wish it wasn't so, because roon's DSP is functionally excellent, but I can only hear what I hear.

Fixed Volume should have a very slight SQ advantage over Device Volume, but the difference is very small.
Both on my Grimm MU2 all-in-one (with analogue volume control), and on my previous MU1 streamer (with digital volume control), I could barely tell the difference between Device and Fixed volume, so currently use Device for its extra convenience. Both sounded significantly better than roon's DSP.

Yes this is painfully obvious even with just a regular laptop, heck even turning on headroom management on +0 dB results to flatter sound on my streamer. Audirvana vs Roon and it's no contest, Audirvana runs circles around Roon. However, lossless RAAT signal path is pretty transparent to me as long as the streamer is reclocking (or in your case Pure Nyquist oversampling) the signal on the HARDWARE level before it goes to the DAC

 

[Currawong]

The Roon volume control is very audibly lossy to me with the Chord DACs I have here. It was also audible with the Yggdrasil, which I wanted to attenuate at one stage while I was trouble-shooting some clipping with an amp. I hadn't expected it at all. Digital effects applied to music have a significant effect on the perception of soundstage depth and instrument timbre -- I notice these things immediately. I just tested a volume control on a DAC down below -95dB (with -0dB probably being around 95 dB), and could still make out the music, which should be impossible. So, I really don't care what people claim is "audible".

There was research done recently on the effects of network equipment on sound quality, with proven results showing jitter could audibly affect the output. Annoyingly, it also proved vibration reduction could as well, as the whole test rig was isolated against multiple types of interference, including vibration. I don't have the link offhand, but I'll post it when I find it again.

Edited: fixed a typo

 

[TheAttorney]

The Roon volume control is very audibly lossless to me with the Chord DACs I have here. It was also audible with the Yggdrasil, which I wanted to attenuate at one stage while I was trouble-shooting some clipping with an amp.

The term "very audibly lossless" is a bit confusing to me. The rest of your post suggests that you meant to type "audibly lossy", but I'm not sure.

Anyway, if people are perfectly happy with the SQ of roon's DSP, then that is perfectly fine. Saves a lot of effort trying to improve it.

 

[theveterans]

The Roon volume control is very audibly lossless to me with the Chord DACs I have here.

Same experience with my SP2000 DAP as well that the Roon volume control is lossless on the following conditions:

1) Device must be Roon Certified/Roon Ready
2) Device has a hardware preamp function included (DAPs, Chord DACs, Metrum DACs, etc.)

 

[Hypops]

The Roon volume control is very audibly lossless to me with the Chord DACs I have here. It was also audible with the Yggdrasil, which I wanted to attenuate at one stage while I was trouble-shooting some clipping with an amp. I hadn't expected it at all. Digital effects applied to music have a significant effect on the perception of soundstage depth and instrument timbre -- I notice these things immediately. I just tested a volume control on a DAC down below -95dB (with -0dB probably being around 95 dB), and could still make out the music, which should be impossible. So, I really don't care what people claim is "audible".

There was research done recently on the effects of network equipment on sound quality, with proven results showing jitter could audibly affect the output. Annoyingly, it also proved vibration reduction could as well, as the whole test rig was isolated against multiple types of interference, including vibration. I don't have the link offhand, but I'll post it when I find it again.

If true, this might explain why I hear no degradation with or without Roon's digital volume attenuation or other DSP engaged. I use a Chord DAC as well as a USB isolator/reclocker.

 

[Currawong]

I did indeed typo, and meant "audibly lossy". I can hear it clearly with a change in how instruments sound, especially note decay.

 

[Jake2]

Annoyingly, it also proved vibration reduction could as well, as the whole test rig was isolated against multiple types of interference, including vibration.

It’d be interesting to know more about this aspect if you can find the link. The term ‘isolation’ as often used in the context of minimising the adverse effect of vibrations suggests the source is external to sensitive components. From experimentation I’ve found localised damping on circuitry (eg toroidals, OCXOs, R2R ladder, FPGA chip) within already well isolated components to be quite audible, but not necessarily beneficial, it is rather location and damping material dependent.

 

[Roybenz]

My Dac disappears after my core restarts (iMac) I have to unplug and plug in the usb every time for it to show up. Any solution to this? Thanks a lot for helping

 

[Rayon]

I made very interesting find when playing with oversampling in Roon. If I use headroom, it does things in this order:

1. Convert to 64float
2. Attenuate -3db
3. Oversample
4. Dither to 24bit

However, if I disable headroom and use manual volume adjustment under procedural EQ the order looks like this:

1. Convert to 64float
2. Oversample
3. Attenuate -3db
4. Dither to 24bit

At first it may not seem important, but the effect is massive with Holo May. I've noticed the same effect earlier when using external crossfeed plugin with PGGB. When converting to float 64float, 64float can represent the original 16/24 bit perfectly. However, when attenuated, the samples become approximation. If we postpone attenuation, we use all the information that the original bits contain as a base for oversampling. 64bit float can handle intersampleovers independently so those are not a problem yet. Then we can attenuate the oversampled values without any worries as we need to convert into 16/24/32 bits anyway and the conversion from 64float is going to be lossy by definition.

If oversampling from rounded values, it sounds like the more we oversample, the more pronounced the softening effect is. It's like also some smearing would be happening, like the phase information would be lost due to rounding in the original attenuation.

PGGB attenuates the upsampled values, but I think HQPlayer uses pre-upsampled values. That could be one of the reasons why PGGB sounds so much clearer. I've also written to HQP dev as it would be super interesting to hear what HQP sounds like when attenuating after the upsampling.