Not interested in watching the video, but thanks for the info about the files.
www.audiosciencereview.com
I'm not going to read ALL 857 posts, but after skimming through, I think the gist is that, yes, he could hear the difference between the filters, but many people had trouble getting past the idea that he was comparing digital filters rather than comparing DACs. During the video he explained that was to make his test verifiable and repeatable.
Ah, thanks. I had forgotten about that thread. TL;DR: It's really a test between two software reconstruction filters which have no real world applicability to DACs (except possibly for some niche DACs doing stupid stuff that no one should really care about).There is a whole thread on this test:
GoldenSounds passes apparently ABX test for DACs (NOT Really)
Spoiler: Apparently he does it by hearing above 20kHz. MOD EDIT: GS only tests two different filters created with a software program. At no time he tests different DACs let alone passing such a difference. Edit: The sharp filter, the sinc-L, is very similar to the MScaler x2 filter, so...
Nyquist-Shannon sampling theorem is rigorously proven mathematics, not just theory. But; consumer grade multichannel DACs immediately benefit from higher sample rates (96/192 kHz) spread quantization noise over a wider bandwidth, potentially lowering noise in the audible band (20 Hz–20 kHz) and allowing gentler anti-aliasing filters for better phase linearity and transients.I think Dirac itself can run at 96 kHz and possibly up to 192 kHz. If I remember right even Audyssey can do that.
The device itself is limiting it to lower sampling frequencies for various reasons, one being the DSP bottleneck.
I can use DLBC at 96 kHz sampling rate but it is the PC standalone version. That’s for playback, for calibration, I can only use 48 kHz because I have the Umik-1 that can only do 48 kHz.
Found the following, follow an interview with Chris K, by audioholics.com:
and below are from miniDSP support portal:
On the topic of High Sample rate vs Room Correction
We get these questions very often so we thought that we'd take the time to do our best to answer them to the best of our effort. The standard questions are typical: - Can I get Dirac Live for 192k or 384kHz or MHz...? ANSWER: Unfortunately due t...support.minidsp.com
What
Again, what? "Transients" are the eternal bigbear for those who claim measurements are lacking, but what exactly they are and why the hell you would need 192kHz sampling to get them is quite the mystery.
No, that is not happening. These modern DACs oversample 64 to 256 times. They will try to oversample to an internal rate that is the same for all incoming sample rates. That also means that that quantisation noise is similar for all rates. There is no lower noise. Gentler filters have not better phase linearity. That is just a matter of a phase linear filter. It can have any slope you want. As to transients: there cannot be perfect square waves in bandwidth limited material. If you create them anyway, it just shows things are wrong with the reconstruction.
Yes, it was a test between two reconstruction filters. That's what he set out to do, that's what he explained in the video, and that's what he succeeded in doing.
I wish it did, because you'd get something for nothing, but it doesn't work like that.
I think a linear phase filter isn't a flat phase filter, it's just a filter where the variation of the phase wrt frequency is linear, rather than non linear (like minimum phase).
What you see here is just the propagation delay of the filter. Because it's fixed, the relation of the fixed delay to the frequency is not constant, hence the sloping phase line.
It always happens in the digital domain in DACs of the last three decades...
Good to know you agreed
, and yes, that's why I used the word "theorem" (post#1069)As you said, potentially..., and the results of those stated "facts", or "evidence" being audible or not is going to be a forever argument. Those who said they could hear it don't have to prove it because no one hears what they hear, those who heard no difference, or couldn't tell if the difference is noticeable enough to say which one really sounded better, would be both a subjective matter, and also supported by seemingly numerous comparison listening tests, and if done in DBT, no published results showed the participants could reliably hear the differences, let alone which one was better (subjectively perceived by the individual).
Why four? Do you have four analog devices you want to connect? The Tide16 includes two stereo inputs (RCA and XLR).
When I read the devteam's reply in between the lines, I believe that when they stabilize their code, they'll revisit the USB audio's multi-channel capabilities and hopely iron out the issues to make it a reality. After all the USB Audio Class 2.0 specs support up yo 255 channels. Keeping my fingers crossed.
