@Christian_DEPLANTE can you describe a bit more how you use Roon DSP, is it only for bass EQ ? Do you resample? Do you hear differences between 1.5 and 1.6 on all formats (44/16, 96/24…)? On which type of music?
FYI some weeks ago I challenged the PEQ of Roon and ran some tests:
defining a multiband PEQ filter, 5 bands, various gains and Q
implementing this filter in Rephase / minimum phase, generating associated convolution filter
Test 1: using Roon PEQ with my filter parameters
Test 2 : using Roon convolution engine with Rephase convolution file
There was no difference at all, correlation depth was maximal.
Running Diracs through the audio chain in the digital domain, I could reconstruct the response of the filters perfectly in REW, they were absolutely superposable.
This rounding up MUST be removed. It is an aberration. And what was wrong with the previous display which enabled to see both all filter settings and their effect ? It was much better IMO.
The Q filters are convolution filters, aren’t they ? So yes. As I use parametric filters with Q, I use convolutional filters. It now seems clear the subtle quality degradation I observe is associated with HQplayer and perhaps using convolutional filters.
PS: I use ClosedForm/ No dither as the best setup in HQplayer Roon 1.5.
Yes, only for bass EQ.
I resample any input format (mostly CD, I made my comments from CD format) to 32 bits/float at 384 or 354 kHz in HQplayer. But I do not resample in Roon.
Given other comments from Roon that downgrading might alter the database, I am not keen to going back and forth 1.5 to 1.6 for further testing…
I prefer to wait until at least I can surely make the comparisons with exact same filters for which the system was tuned. And this is already a significant source of difference, due to the truncation to 1 decimal.
I agree with you for Downsampling (going to finer sampling) in Roon. In HQplayer, different story.
It was what, two hours until I completed my thread post with some explanations on setup mentioning HQplayer, and I could not say from scratch the problems did not appear without it, for me removing HQplayer was out of question anyway. Apparently bringing contradiction ASAP was an emergency for some. At least this enabled to circumvent the issue to HQplayer setups. And the change I detect is probably only heard with very well tuned systems. Again, different populations of Roon users. Some say same sound across releases, this has not been my experience, but so far it was always been progressing hence my disappointment.
And finally we must be rigorous: true that in systems modern DAC are usually not the weakest part but if basic measurements said it all in audio it would be a different domain out there. The software between our ears is far more sophisticated than a spectrum analyser. But this is not a discussion for this forum.
Christian, you’re right that the “Q” filters of the parametric EQ can be seen as a “convolution” in a broad sense.
My guess is PEQ in Roon is implemented purely in the time domain using recursive filtering, coefficients being derived from Q, F and gain settings.
Whereas “convolution” of a user specified filter impulse response is implemented with a fast convolution algorithm, involving some Direct Fourier Transform by chunks.
My assumption is that these are entirely separate modules, and that the PEQ/recursive filter module has not been touched between 1.5 and 1.6.
I’d like to do some test, could you send me by PM your current filter settings? PM can be in French
I was curious about this, so I googled it and learned that Floyd & Toole found that some people can hear differences of .1 dB, but only in a specially treated room, with headphones, under ideal listening conditions. No one can hear a .01 dB difference. If there is a change in your sound, there must be some other explanation.
Here is my feedback on the pure audio playback quality: At the first hearing seemingly more stage and emergence of voices. But then everything smoothed out, middle frequencies to be emphasized, heights disappear, the bass becomes almost mushy. Would immediately back to 1.5. Unfortunately, no backup …
Well in the first point you are describing my free time, in the second point you are describing my job. I work as a sound engineer and I am working with professional musicians, orchestra’s etc. I do have an ear for music, timbre, ambience retrieval, if not I would be bad at my job. 99% audiophile recordings I hear at demonstrations make use of artificial ambience and a lot of other mastering processing btw, once you know these processing you’ll hear inmediatly what has be done, heck most of the time I can even tell you wich software package they have been using. But that aside, that’s my professional ear, at home I do not listen like that. I can tell you that I don’t hear any difference between 1.5 and 1.6. But offcoarse, most musicians and sound engineers don’t have ears good enough to hear a 0,1db difference, at least I have never met one. In fact I’m very glad I don’t, it would only drive me crazy.
You do realize that one Decibell difference was considered to be the treshhold of the average human hearing in the glory of their physic. Let’s asume you have above average human hearing ,as I instantly believe you do, you are stating your hearing is ten times better then that of the average 20 year old. The only case I can think of when it becomes audible is when you apply this eq right in the middle of the crossover region in the most sensitive area of human hearing where the phaseshift can cause an additional time delay between your two drivers. Time delay is audible into microseconds, (in the most sensitive part of our hearing) that’s part of our human hearing survival mechanism. People are very sensitive to timing issues and frequency shift, anyone can hear if a guitar is badly tuned, but not that sensitive to amplitude differences, even superhumanhearing like you mentioned. It’s not how our ears work. You say you are only using eq below 60Hz. I have to meat the first person on earth who can hear a difference of 0,1dB or 0,1usec timing issues at those frequencies, seriously, and I work with a lot of people who have what I consider superhuman hearing. Move your head 1cm and the difference at these frequencies in your room allready exceedes more then 1dB and phasesifts larger then several degrees.
I simply don’t believe these things make so much difference to you or anyone else that it defines the line between being able to enjoy music or not, there must be something more at stake here. What did you eat last night? How much water did you drink today? When was the last time you vacuumed your listening room?, how’s the weather outside? did you leave some empty coffecups on the table? Did you move your chair back in exactly the same spot? Have you cleaned your ears lately? Have they been in touch with water or steam while you where showering? Have you cleaned your nose? All things that probably make much more of a difference in the perceived sound then 0,1dB in a notch filter. And while most of these examples (not all of them) are also on the edge of the audibility threshhold it is not a joke, after all sound is nothing more then modulated air, something a lot of people seem to forget when they talk about audio. So the point being, it’s is virtually impossible to be 100% sure the difference you are perceiving are caused by the fact of a 0,1 dB difference in an eq filter. At least it is below the treshhold of my believability.
Why do I even bother to post a long answer like this. Well before you know it people start doubting and start hearing differences that are not there, or at least not being caused by the dsp engine. And we all know how it goes, someone doesn’t hear anything, some others say they do. The one who didn’t hear a difference feels he is not as good an audiophile as the others so starts listening again, highly concentrated. All of a sudden this person hears a tiny difference, this difference becomes bigger in his head the longer he listens and before you now it we read post here stating night and day differences and accusations towards Roon about destroying your sound and your pleasure of enjoying music. Personally I can’t stand this kind of audiophoolery and accusations so there the reason why, nothing personal whatsoever.
If you are so overly sensitive to differences (no sarcasm) I can tell you that Roon’s dsp engine is not exactly being considered as top of the bill by my sound engineering collegaes. Yes sound engineers and muscians also have heated debates about the sound of DSP engines/plugins and anyone of them is willing to take a blind test with you between Roon eq and for instance a Sony/oxford eq (if that even would be possible) But that’s not the case here. I think the DSP engine in Roon is perfectly fine for the above average music lover / audiophile. If you want something really better, and I think with your sensitivity you should (again no offence), you have to go looking in a much more expensive solution. Don’t expect Roon to come up with it, don’t expect mastering studio quality in Roon, and even then the differences are there but not as big as you might expect.
Just saying, Roon sounds exactly the same to me. I never thought to comment on pre and post audio quality as it wasn’t a thing untill I noticed some posts.
It always sounded just right. I use no DSP in Roon. I just play it straight to My Meridian 218 and DSP5200Se Speakers or my Bluesound units. Cat5e cables mostly and normal off the shelf network switches.
This is one of the widely ideas around, that everyone as heard. Yet experiments have proven that to be truly inaccurate. For example, in his book named "Experiments in hearing, ASA1960,ISBN 0-88318-630-b, Georg von Bekesy elaborated methods and devices that enabled him to establish, from what he named the law of contrasts, that human acuity could far outperform the theories developed earlier by Helmholz, another famous name in acoustics. He was awarded a Nobel prize in physiology or medecine in 1961for his insights into human earing. I know at least this, although this is my hobby not my job.
I am not talking about being an absolute dBmeter. I am talking about something truly different. using our short term auditive memory to assess a change applied to a filter affects in a rapid A/B comparisons of the same discriminant piece of music. In my case I find the assessment in terms of timbric variations of sounds of instruments such as altos or cellos in an orchestra, for example, or of a recorded voice, especially one I have heard live recently. Using relatively low Q values, say about 1.6, at for instance 56 Hz still affects the lower midrange frequency, such as the balance between fundamental and harmonics on such string instruments.
I would be incapable of reaching that accuracy without the rapid A/B switching in a stable environment, on a system I know well and on recordings I have ever heard live or have heard on several high level systems. But in those conditions the difference is reproducibly audible.
Then I am aware of the risks you evoke. I am trying to take precautions to be specific and avoid controversy. I much prefer the discussion to stay focused on narrowing down the issue and finding solution. The idea is to contribute to make Roon better for all of us. If the audio flow satisfies even the most demanding, it is a winner for everyone.
And in fairness you are being very level headed with your complaint, as is @Nyquist with his response, it’s good to see some thoughtful non- aggressive discussion around these topics that we don’t all necessarily agree on.
I will speak to HQPlayer as I use it downstream from Roon. On Roon and HQPlayer “interaction”, there is none. After Roon applies it’s processing, the PCM (or DSD) is sent to HQPlayer. There is not mysticism to explain “loss of expressivity”, or rather if there is it’s a change in Roon itself. However, Roon employees have denied any change in DSP processing/filters/algorithms.
The filter I use in HQplayer is the ClosedForm approximation of HQplayer to the million-like taps used by Chord in their most recent DACs and scalers. Don’t know if that matters.
I already mentioned part of the difference would come from the rounding to one decimal of the different filter parts I used in 1.5, degraded in 1.6 by this process.
Happy to retest once the DSP parametric filter interface has been corrected, or put back to its previous form, that I also preferred visually.
Personally I think Roon need to address the sound quality as it is considered by many to be not as good as some of it’s competitors. It’s good having the best UI in the business but once people get bored with that they will start looking at what else is out there.
I much prefer the sound of Audirvana but their software (especially the remote) are absolute crap. If the software was better then I would move on from Roon without hesitation.
I am not saying Roon sounds bad but I think they have spent enough time on the visual stuff and need to get back to basics which is how it sounds.
For the sake of clarity, there is now a million-tap variant of closed-form which appeared after discussions of the merits of such high tap counts. This was around the time of Chord’s M-scaler release. So everyone is right
