They sound great to me, so who cares. Great built in DSP with EBA, and no faffing about with separate amps, Dacs, cables. Properly designed around human perception. I’m happy and Roon sounds exactly the same to me with every incarnation.
1 Like
@Wim_Hulpia wrote that Meridian speakers “can’t handle 192k qobuz.”
Actually, not correct.
The Ethernet input to the the 818v3 controller handles only 96, but the USB handles 192.
The connection from the 818 to the speakers is not Ethernet, in spite of using an Ethernet cable. It is a Meridian protocol which does handle 192.
But in any case, it is unimportant, this is not what I worry about.
Of my 5,150 albums, 122 are 176 or higher (2%).
And of those, 86 are from before 1990 which means they were analog and did not have super high resolution (most from Qobuz).
And the 36 genuinely super-high resolution albums are mostly audiophile curiosities, all but one track have been played less than 10 times.
2 Likes
None of the most timing-sensitive circuitry in the world (eg GPS) would work if electronics engineers had not learned how to achieve extreme accuracy in very difficult environments. If expensive modern streamers and DACs cannot manage do the same in the comparatively low-difficulty environment of someone’s home, that does not inspire confidence in the skill of their designers. Just to give a positive example, Soekris DACs are relatively low-cost compared with some of the gear mentioned on this thread, and they are pretty much immune to upstream effects with any source (USB, S/PDIF coax, AES). But then, the designer was also the founder of a then successful digital communications company. TL;DR Upstream timing issues should not be an issue with well-designed and built downstream equipment. Buyer beware.
2 Likes
Take a square wave. Its Fourier transform is a series of regularly decaying harmonics, If you change their phase, their frequency, truncate these, you change the perception of the waveform.
OK we don’t listen to square waves (although in Electro… 
), we listen to voices and instruments. If you seek the spectrum of instruments like violins, the fundamental has typically less energy than the sum of its harmonics. The first 6 harmonics are concordant, the discordant ones start at the 7 th harmonic. The instrument texture of a violin is largely determined by those.
Our own software between our ears is neither a spectrum analyser or an oscilloscope. Along with the hardware, it has been specialized by millions of years of evolution into the detection and localization of subtle transients - high frequency for small prays, lower frequency for big predators ?
A survival issue, no wonder our ears do not work exactly like a lab instrument, and do not have an optimal response to continuous sine waves, that although they form a linear decomposition basis for signal theory, were not commonplace in nature. And neural networks are not linear, by construction.
The 2 microseconds interaural detection threshold cited in the MQA paper is not directly compliant with Nyquist at 20 kHz, that is 25 microseconds. Getting back to the square wave and the phase of the harmonics, Nyquist formula was written for amplitude reconstruction (2 points per wavelength). One of my mentors in signal processing made long time ago there is no Nyquist theorem for perfect phase reconstruction, he found that with common algorithms, an accurate reconstruction required up to 6 points per wavelengths.
Or is it that to get perfect reconstruction you need to use the correct interpolation function to enter a DAC ?
That is the sin t /t sinc function. Yet to implement it with full accuracy (say 100 dB) it was shown by Chord guys (sorry for citing a brand for once) you need a much longer operator ? That is precisely why I find that using HQplayer that has a 1 million point sinc function to interpolate from 44.1 kHz (25 us) and 16 bits to 384 kHz (1.3 us) and 32 bits, that can even be used with zero dither, I am much happier with 16/44 than I have ever been, both for timbric accuracy (harmonics) and spatialization (2 us interaural detection limit) ?
I encourage you to test by yourself.
Here is a link to a pdf document by Keith Howards, about why those long filters, that can also be computed outside of DAC, via HQplayer for example, are theoretically more accurate, although I concur that 100 dB seems a lot compared to the physical capabilities of our transducers, in particular.
3 Likes
Then why do we need 352.8 or 384 mqa if an upsampled 48 pcm could do the job ? 
Good question.
We don’t?
3 Likes
You know the audiophile trope of “my wife heard the cable change from the other room” ?
Well, “my pet bats trill along to Mongolian nosewhisle vol. 12, but only the HiRes version.”
6 Likes
This has to be the dumbest statement I have ever read. Just kidding, right? Right? 
7 Likes
Absolutely not; you should read this statement aloud to yourself for a week straight, not sleeping for the entire time, and with 100% pure copper pennies in your mouth (don’t use the zinc ones), until your vocal cords burn in.
[sorry, this thread has elicited amazing amounts of sarcasm, I’m just tossing more on the fire]
4 Likes
Except this isn’t actually true In the real world. Even the cheapest DACs are able to suppress any electrical noise over USB.
Hey look, someone even measured this Do USB Filters Make DACs Sound Better? (Video) | Audio Science Review (ASR) Forum
3 Likes
To be honest, it’s genuinely very hard to tell the people who are taking the piss and being serious apart on here.
Sometimes those making statements in earnest outstrip the jokers in outlandish ness.
In this case, your leg has been pulled.
That said, I’ve given up trying to be rational on here, abandoning roon ship for ASR, where at least people argue about things that actually exist.
3 Likes
You never know with some of the characters that hang out here…
2 Likes
Buh bye, roon will miss your rational guidance.
At least here you won’t get robbed! At least here in the forum. In the hi-fi store it looks different again.
1 Like
The guys over at ASR are just as bad. They don’t know how to listen, only measure. Opposite end of the spectrum.
3 Likes
This bit perfect debate will never have a consensus.
Stepping sideways, I have had many digital sources in my system, many of which sound different.
Many CD players, Pioneer N70a, Cambridge CXNv2, RPI4 USB, RPI4/Pi2aes, Laptops, NUC’s, customised PC’s etc.
Point in case, I have an RME ADI Dac2 Fs, which has excellent noise and jitter rejection.
I clearly hear a difference between my Ropieee RPI4 via USB and RPI4/Pi2AES Hat via Spdif.
Both are bit perfect devices, but the Pi2aes has better detail retrieval, more realistic high frequencies, and just sounds more transparent and open. The RPI4 via USB sounds a little congested when directly compared.
Sources were compared by direct switching between Roon endpoints on the same system, so switches were instant with no waiting. Music was both time matched and offset for comparison over many hours of playback across many genres of music. Differences were more obvious on complex music like Jazz.
IMO, there is clearly a difference even though both are ‘bit perfect’.
1 Like
We are hearing differences because our ears and brains are different, right? I propose that this variability is far higher than any potential change between 1.7 and 1.8; (I heard others saying the same between 1.6 and 1.7 )
- on the anecdotal evidence in this forum: some hear worse, others hear better, these anecdotes are truth for the person saying them, but completely useless for the observers reading them;
The Roon philosophy was never about SQ alone. It’s a way of life…
H
So what does this prove ?
-
Two models of filters tested.and this discards ANY type of filter. Well, imagine you apply that to, say, image stabilized lenses… You test two model of lenses; find them not to work for tracking fast moving objects, and infer that NO stabilization enables to track fast moving objects… No comment - besides I have no opinion about these filters that I never tested or even saw. My experience on USB was related above, and I spoke also of galvanic insulation.
-
Then the graphs are shown ONLY in the audio band. Those filters have obviously very little influence in the audio band, noting that the graphs show increasing noise with frequency, and that these LC filters do not do much in the audio bandwidth.
There are issues with HF noise (NOT tested here while it can adversely affect electronic equipment). I once burned a pair of output transistors on a semi-prototype amplifier due to HF latching upstream of it, so don’t tell me it cannot affect equipment.
I believe HF noise insulation remains a difficult problem as one comes close to DACs, particularly because of radiations of elements of circuits behaving like antennas at high frequencies corresponding to very short wavelengths.
Take Nagra DAC circuits, I was told they are extremely carefully shielded, and that might be a more critical reason for high musicality than the circuit itself.
Hahaha you could be right - just exactly what is “ the new computing signature”?
H
Hey I’ll raise you… here’s one that baffles me…