Roon 1.8 sound quality change?

The tech being used here is way above my paygrade but…
I was using dsp in 1.7 to upsample everything to 384 to my Chord Qutest.
With the change to 1.8 I am fairly certain there is a subtle but noticable improvement in SQ with the same dsp upsampling of 384 for everything.
Dont know for sure exactly why and in all honesty I really do not care overmuch.
If I am happy that is all that matters… :joy:

Apparently bats, dogs (and humor) are not welcome on this thread…


Don’t look beyond - Roon team mentioned they improved the DSP side, on top of improving CPU and memory management. Which I believe I also see directly - now 4 seconds less to start output after heavy computations in my own Roon/HQP setting, from 12 s in 1.7 down to 8 s now.

1 Like

After all the technical digressions, do we now have an idea why Roon 1.8 might sound different?

Have double checked and also found out that polarity is inverted within 1.8.

Thanks for the tip!


1 Like

For all:

  • less CPU load and memory traffic, might influence indirectly jitter and noise.
    In my case imaging improved, better transients I think “called” for a slight rebalancing of the very low end to equilibrate the spectrum.

For those using Roon DSP:

  • Further uplift of quality likely as Roon claimed they improved it.

The phase as far as I can test in my configuration, is a non issue. If Roon 1.8 inverts the phase, then so does iTunes, on IOS running machines!

Hard to believe that, with the number of iTunes users, the army of people working for Apple, and their reputation of dedication to audio and video quality, they would have let that major blunder go uncorrected for so long… And Roon team probably checks such things in the first place.
So enjoy 1.8, and see the video I enclosed in above link.

PS: On 1.7 phase, I cannot tell directly but indirectly - on my system I tested initially with 1.8 a correct low frequency phase with a filter designed under Roon 1.7. It was all reversed in phase correction but so was my amplifier.

On what speakers are you hearing this rebalancing of the very low end to equillibrate the spectrum?

FWIW I’ve repeated this test with a bit more volume and and objective testing aid (a ruler). I’m now confident that the polarity for Roon “lossless” on my system is correct and that the wiring’s correct.

38 cm mounted in a large load, very simple crossover.
Then there is a subwoofer, with a high slope, to dim it out quickly from some 45 Hz onwards. I assume that it was 18 dB/octave, I input via an additional RC filter, in that case (not a huge sub) it is important, it was obviously designed for less bass-capable speakers hence the added slope.
The only trick is the RC can’t work properly unless it goes from amplifier output to higher impedance input.

I am running two 15" woofers per side, dsp corrected to 30Hz (-3db@25Hz) and designed for the room they are in. But I can’t hear a difference between 0.3 dB more or less for everything under 40Hz let alone anyone with small bookshelf / standmount loudspeakers (lot’s of audiophiles who are afraid of WAF) Once used to the effortless bass of twin 15" drivers any speaker with a 6’5" woofer sounds like a distortion generator. That’s why I was so curious why in your case it can make or break your sound. What woofers do you use?


Agreed, in agreement also with my remark at the beginning of this thread about the missing measurements in magazines about speakers: distorsion spectrum at different output levels, showing basically what you say in a humorous tone :upside_down_face:

We are not still doing the “Nyquist said, 44.1 kHz makes for perfect reproduction” thing, are we? In 2021?

Nyquist and Shannon were doing math, and math does not have a problem with infinite.
But engineering does: infinite numbers are awkward when it comes to cost, energy consumption, starup latency, the black hole thing…

Rob Watts of Chord has done a very articulate discussion of how to move real-world processing capabilities closer to Nyquist’s theoretical goal. His current $10,500 stack can process red book data to 16 bit perfection, which is lovely but far from the Nyquist theoretical perfection.

Note that I’m not arguing that theory is not reality. The boffins may prattle but my ears are better. That’s not the point. The point is, you need to read the theory carefully.

Nyquist, the poor guy, never said that 44.1 kHz makes perfect sound.
Or if you think so, I have a 1982 CD player in perf3ct condition…


I agree, and I have been pointing out that HQplayer enables to do pretty much the same for a reasonable yearly fee, if one accepts a few seconds more delay of piping in the real time stereo convolution in the digital stream on Roon core output, to attack any DAC capable of accepting a say, 384 kHz resampled PCM. I like Chord DACs, they are excellent. Already in years 2000 I had a DAC64, with longest tap at the time, it sounded excellent, quite analog-like compared to the competition I tried. Since then, improvements were made by everyone, including them.

And yes, the resampling with 1 million samples is at best “only” accurate to about 100 dB, and still not tapping into the ultimate potential of High Res audio like 24 bits. Now, assuming nothing is lost in the digital wiring to the DAC, while 100 dB accuracy or Signal to Noise is challenging but probably achievable for top electronics, it remains to be seen what is the ultimate accuracy one can reach with even ultimate speakers. All these being considerations that we have to mitigate and rethink in terms of transient accuracy, to which human ear is most sensitive: for continuous signal all these efforts are superfluous.
I believe that we have already a good compromise with those Chord or HQplayer solutions, I am personally more concerned with this “wanning” mentioned by Claude Cellier at the end of his talk.

The easiest way to understand that DACs are doing an approximation is that filters are needed. A perfectly constructed signal would not need filters to clean away junk over 22khz because there would not be any.

I am pretty sure that Nyquist was talking theory when he said that up to 22khz could be perfectly reconstructed. After all, there where no DACs in the world when he said it.


That’s not the point I was trying to make. I do understand some of the difficulties of real-world implementations (we haven’t got the time to wait for the perfect answer). I do think that reference to theory is useful when looking at ideas/proposals. In this case I wanted to counter the idea that the solution to perceived audio reproduction issues is to start a sampling rate arms race. Theory needs to be backed up by experimental evidence and in this case there’s not a whole lot of evidence that anyone can tell the difference between red-book and higher resolutions. IMO red-book cuts things a little close but beyond 24bit and 48kHz is probably overkill. If the theory says it’s enough and experimental evidence backs this up why go looking for a problem to solve?

If sampling theory is inadequate perhaps another approach might improve things? If you believe that I have a SACD player gathering dust…

I would have stayed with SACD if the music didn’t cost so much. I couldn’t afford to feed that and the vinyl habit so a choice was made. It helps that they are still remastering and producing vinyl. SACD prices went through the roof. Sounded fab to me though.

Fair enough, and each to their own when it comes to what you enjoy listening to. I was never won over by SACD’s sound and the “dirty secrets” of the technology left me less than impressed personally. That’s got nothing to do with whether you like the sound of it though and it does appear to be the digital format preferred by vinyl fans. As far as vinyl goes, I’m in the several thousand records but no turntable camp.

1 Like

I’m in the several thousand albums camp but recorded then all in 24/96 with a turntable but mainly play them off hard drive. Half of them are old punk ones that I would hate to part with as they are part of my growing up.

1 Like

Totally agree.
I am using the Chord Qutest which for its modest new price is an excellent sounding unit.
Only possible gripe somebody might have is that it has single ended RCA outputs only, no balanced XLR.
I can live with that though for the SQ I am realising in my system.

1 Like

Sorry all I need to correct this.
I’ve learned that my tube stage preamp is inverted.
That means 1.8 is normal phase.