Performing well in tests vs performing well in real life: as different for DACs as for people.
Live music is adequately defined as a succession of a myriad of interspersed transients with a world of variable decays and harmonic contents. Although they can theoretically be decomposed on a linear basis of sine waves, or rather via a time-frequency transforms, the audible quality of many audio components, can only be very approximately defined by measurements of constant power signals of varying frequency.
Audible distorsion is a complex, multishape animal: harmonic, amplitude and phase, intermodulation, thermal distorsion, memory effects, etc… It is long known in analog domain and it happens to be… the only domain in which we can give an opinion on the quality of digital audio.
Claiming that a simple set of measurements tells it all about audio qualities would be like a marine biologist defining uniquely jellyfish by one measurement of diameter of umbrella plus one of average filament length, to try to browse an awkward analogy. In real marine life, some jellyfish are harmless, others are not, and this is almost unrelated to size.
That’s a shame, as you say there were a lot of tricks employed that were appropriate when compute power was in short supply. I’m also a fan of Chord’s approach to reconstruction which is aspires to theoretical correctness as far as possible. Have yet to take the HQPlayer plunge for fear of the depth of that particular rabbit hole. I’m sure it’s fun, in essence that’s the problem…
What impact does the control app have on sound quality?
It simply passes instructions to the Server
there may be, but interesting though. i run roon lossless on a dedicated mac mini, so i won’t notice anything,
Warning: an old software engineer’s theory. (You were warned, don’t complain.)
Software (or firmware, same thing except for delivery method) is hard. It is especially hard on memory or processor-constrained devices (recall your first smartphone). Software development is expensive. Even hugely resourced companies struggle with getting high-quality software to their users. Boutique streamer and DAC manufacturers have a teeny fraction of those resources.
Distributed, realtime systems are the hardest of the lot. Multiple processors, with different software stacks, communicating over finicky communication channels, trying to cooperate to achieve a common purpose under fairly stringent timing constraints. Keeping such systems working through thick and thin is hugely demanding (think of the last time your cloud services provider failed because of a fiber cut in Podunk).
Roon core talking to Roon endpoint talking to DAC is such a distributed system, operating over a communications network that none of the developers ever tested in that exact same configuration. And each of core, endpoint, and DAC are themselves complex assemblies of hardware and software with their own realtime and resource constraints.
It’s actually amazing that any of this works at all, let alone work so well for most people most of the time. Yet, bugs lurk everywhere, from different understandings of ambiguous specs (yes, most protocol specs out there are ambiguous) to garden variety off-by-one bugs.
Is it then surprising that a small boutique streamer or DAC vendor with limited development resources trying to pack a lot of functionality into a small fanless processor might cut it too close and build something that seemed to work great in a particular environment, but may misbehave somewhat when the environment changes? There’s complex software/firmware moving those bits from the Roon network to the DAC’s innards. A teeny software (or hardware) bug could make an audible difference when things change outside, especially on highly-tuned systems with discriminating listeners.
Just tried the v1.8 for about an hour. I am with you in the sound differences.
That it’s better, worse, or the same?
What differences did you hear?
I agree that it is not always obvious to find the right polarity on musical signals in general. One needs something percussive, and one needs to be sure about the polarity of the percussion - is a large drum a pedal drum, hit by behind ==> positive pressure pulse first, or by the front ==> negative pressure pulse ?
I have found an easy option for everybody to check.
Downloads | Studio Six Digital
There you will find a “Speaker Pop Test Signal .zip” that one can dowload.
I assume it respects the standard of polarity and provides a pressure pop with forward displacement of the membrane, like described for example in audio university.online etc…
One of the files there is a long series of popups (speakersPopLong44.wav) stereo.
It is very easy to check the polarity on a woofer. With a decent volume one can even see the membrane moving forward (that is the + normal polarity) or backwards (that is the - reverse one).
I tested on wideband speaker - no filter at out, even less doubts, and speaker wiring is correct.
Here are the results. I compared the motion of the speaker playing the file via either HQplayer/1.8 without any Roon preprocessing, without or with the convolution filter that I have set (my preferred stream), or directly to Roon 1.8 to DAC (no options, DAC deactivated) and finally via iTunes, in order to get a Roon-independent relative assessment.
Here are the results
Roon 1.8/HQP with my preferred convolution filter: +, in phase
Roon 1.8/HQP with only the phase part of it: +, in phase
Roon 1.8/HQP with a flat filter (also defined in rePhase) -, out of phase
Roon 1.8 alone: -, out of phase
ITunes alone: -, out of Phase
Well, it happens that the convolution filter that my preferred listening had the correct absolute phase. I did not check it before, I just tested and decided to retain the rotate compensation for speaker loads defined in rePhase,… by the wat there is no tool to revert polarity in HQplayer3 “speakers” setting, and I prefered not want to use any of Roon DSP, to be more “flow-perfect”
So I find that everything else than my preferred setting, has a negative phase - upon playing the pops, the woofer moves backwards, no doubt.
And Roon 1.8 has the exact same phase as iTunes.
Well, it seems that my amplifier might be inverting and I did not know it… but somehow I corrected it.
I will however do a test with flipped Roon polarity, removing the phase part of the filter I defined in rePhase.
I could not do the test in Roon 1.7 though… A point rather for Roon team to confirm now, unless someone has not YET moved to 1.8 and reads this while still in 1.7.
I encourage everyone in this thread, to do the same tests.
I hate to do myself down but long posts are wasted on me. I love to learn but I’m too lazy to read that much. Bite size please!
I prefer to give a bit more details to avoid misinterpretation. Sorry if it feels verbose to some.
I concur that making a synthesis of this thread, although useful, takes a bit of reading.
Maybe use a large screen - it helps digesting more information indeed.
There is no substitute to getting correct information - sometimes hard to reduce to one sentence.
It would be “Roon1.8 has no phase issue” but with uncertainties about variants of usage, older versions ?
Take your time, the best way to go fast is to never go backwards.
PS: alternatively, for expressing emotions only and little information, Twitter is short, by design.
As I type, I am listening to Lukas Nelson and The Promise of The Real in MQA and it sounds incredible as it always has done, whatever version of Roon…
One Thing to be aware of is that the vast majority of three way speaker designs employ passive crossovers where the mid driver must be connected in inverted polarity with respect to hi and lo, in order to continuously blend the step responses of the drivers together to get a flat frequency response.
Now, which driver’s polarity is more important:
Midrange, where your hearing is most sensitive, or bass maybe because you happen to visually confirm proper movement?
Interestingly, even very expensive and highly regarded speakers are employing such designs.
And if absolute polarity were so enormously critical, they would probably not make much money.
Additionally, there’s also a extensive body of research by Floyd E.Toole that i.e. shows absolute polarity not being a relevant factor for listener preference in loudspeaker sound.
That said, you are free to prefer whatever you want, just not make it the absolute truth.
And that said, I prefer my full range line source speakers, being spectrally corrected in room via convolution and having a perfect step response by design as well as respecting the recording’s absolute polarity. I think it contributes to a very natural presentation of music.
I found this on Qobuz to check phase and other audio tests. Listen to the end of the phase track where they play the drums. The out of phase should sound slightly out of time.
Now if it did a bit perfect test as well it would reduce the arguments on here. There again it would make less amusing reading then.
I can quickly change the phase on my Focal Arche and it sounds better with it set to normal when they say it’s in phase on album I posted. I am using a roon endpoint so not sure if roon bridge software different from directly out of roon core.
My theory on that is it’s all bit perfect unless you apply upsampling in the software. Believe any sound difference is due to RFI noise from system processing. Best to use separate endpoint for best sound.
Mike, if you download the speakers pop file via the link I posted, there is zero ambiguity on polarity of speakers.
The visual inspection is easily made on woofer. With sufficient amplitude, you will see it moving forward (normal) or backwards (reverse). It is clear. There is no reason why a woofer would be connected out of phase (unless there is a subwoofer and the mains are filtered).
Same for testing the up sampling, or any other step between the audio file and the DAC. I really think that bit-perfect is more of a marketing argument with little technical implications. I would prefer another neologism, flow-perfect, that implies a clean flow without accompanying jitter or noise. More difficult, especially the HF noise carried over and polluting the analog conversion as per the recent discussions.
Better to check than on midrange. Midrange speakers quite often in 12 dB/octave crossovers, should normally be connected in reversed polarity so that after the crossover-induced 180° phasis rotation the reversed polarity puts them in correct phase.
Here’s another test on phase. Album in darker green is correct phase and lighter one is reversed. You can Google search about being released with phase inversion. Can hear differences in the way the drum pressure sounds.
Hi @Christian_DEPLANTE , I gave this a go. Easy is going a bit too far but with a little volume, I found my sub cone a reliable indicator. For me with Roon and a “Lossless” signal path the cone movement was forward. That said I’d not quite call it easy and I’ve never tried it before so don’t put too much faith in my findings.