What does “restored with MQA” mean? If the master is 44.1/16 then I do not nead MQA to hear that?
[quote=“VirusKiller, post:71, topic:5408”]
AES - The Audibility of Typical Digital Audio Filters in a High-Fidelity Playback System Won the conference award for the best peer reviewed paper. Basically debunked the very damaging and flawed 2007 Meyer-Moran paper which claimed that “HiRes” (>20kHz) wasn’t audible.[/quote]
It’s less about whether it’s audible, than if it’s desirable 24/192 Music Downloads - and why they make no sense
Hi Mark,
Did you read that white paper?
One has to unlearn the established thinking that’s it just about the frequency domain and then attempt to get one’s head around what is required to avoid temporal smear and bring it down to natural levels in the recording/playback systems.
2 Likes
Well setting aside my thoughts about MQA for a moment, I was replying directly to the posting of that AES paper.
I am very much of the opinion (and it’s only an opinion), that hi-res formats are the emperor’s new clothes incarnate, designed to tickle the fancy of paranoid audiophiles (we’re a great target market).
If half as much energy was expended on convincing mastering and post-production engineers of mainstream releases to cut back on the loudness and brick-wall limiting, we’d be in a far better place than all the hand-wringing that goes on about upsampling turgid Patricia Barber releases into oblivion.
1 Like
The xiph posting is bit like people going back and quoting Newtonian physics to argue Einstein was wrong. Audio science and engineering has not stopped with Nyquist and the CD even though the theory and the marketing then suggested it was perfect sound forever. Stuart and Craven are two of the most well educated and published authors in this area as well as having a long history of pioneering successful digital audio developments in use hardware and in technologies such as Blu-Ray and DVDA. I would give them much more of a hearing 
than xiph and hydrogen audio.
2 Likes
I absolutely agree with you. The actual production and post production has a much higher impact on what we hear than all these bit and sampling competitions. The e.g. 2L samples that are often used for demos sound great even in red book quality. Simply why they are produced in an excellent way and not for beeing effective on the radio.
We all can distinguish beetween a good an bad production without any doubt. I am not sure about the result when would have to decide, if we hear a red book or high res file.
Nevertheless, finding the next better technology is part of the fun. 
Really?
The article is fully referenced, so please point out which parts of the theory are outdated or have been debunked. Or are you suggesting that our ears have evolved significantly in the three years since it was published 
So given that the physiology of the ear is not predicated on hearing above 20KHz (and then only in the very young), and the best studio mics crapping out at around 18Khz, what are you trying to capture and reproduce?
As far as I am aware, there are no peer reviewed studies indicating that anyone can tell the difference (never mind express a preference) between hi-res and red book with any statistical confidence.
Actually there are peer reviewed studies that do show it. Recently there was a AES workshop about a meta analysis of about 20 double blind studies by Joshua Reiss of Queen Mary University, London which showed hi-res recordings were audible vs. CD.
the overall result was that trained listeners could distinguish between hi-rez recordings and their CD equivalents under
blind conditions, and to a high degree of statistical significance
I’m not going to take time right now to find you the link. If you’re interested you’ll look.
As far as the theory being outdated, no, it isn’t. But the theory has limitations that people like you always conveniently forget about: such as filters being limitless, no out of band frequencies in the sample, etc. (There are others). In real life the NECESSARY conditions for the theory to work perfectly don’t exist - so the results of Redbook sampling aren’t “perfect” reproduction of what was sampled in actuality.
Well the point is I have looked and couldn’t find anything. Perhaps I’m using the wrong search terms.
So if you could link it at some point I’d be grateful.
Heh.
I referenced a whole load of stuff in this thread.
Not about peer reviewed studies on audibility of hi-res though.
I spoke with Bob Stuart in early 2015 at some length and found his insights into the time sensitivity of hearing to be most interesting. This comes direct from recent studies at Cambridge Uni. It takes the debate much further than the human ear vs 20kHz etc.
This is why the new MQA sampling from the original masters is key as much more care is taken to preserve the critical aspects of timing (even from older analogue masters…despite what Dr Aix seeks to post every other day). Humans being far more sensitive to this than previously appreciated. This is also why the ADC and DAC characteristics become important for the end to end chain. They have an effect and can be accounted for revealing more of the original signal as recorded.
To get stuck in the standard hi-res arguments is to miss the point.
It is the improved timing that makes the biggest difference when you hear MQA …not the frequencies.
I will say though that it is really for the MQA team ultimately to help us all understand what this all means. I hope that the education pieces will be coming at some point rather than just being as reported by a few folks that get to demos or presentations.
1 Like
“people like you”.
Okay, I shouldn’t have written that phrase. Apologies. Just shorthand for what happens in discussions of Nyquist, where it is always pointed out how the theory/mathematics are perfect, and “prove” that hi-res can’t have any significance over Redbook. My point is that the theory makes all sorts of assumptions of perfect conditions that don’t exist in real life. Therefore the theory doesn’t show that hi-res recording/playback CAN’T be heard as different from redbook.
I don’t really know how MQA works. I do think there is something to the idea (with some testing existing to back it up) that humans may be extremely sensitive to timing info, and hi-res may reproduce this better than Redbook.
But the xiph article doesn’t say that at all. The whole point it’s making is that higher sampling rates have the potential to be audible and are detrimental.
You are right in your second paragraph - the timing claims around MQA do merit further investigation.
@R1200CL. Thanks for reply. For what I know for now, HQPlayer does not have the plug-in necessary to decode MQA. In order to get full benefits of MQA, a MQA decoder is needed.
I can’t expect any kind of bit depth/sampling until someone can show me what is final MQA playback is.
It is saying that (which may be potentially true, but apparently isn’t in real life), but it is also trying to say that there is no advantage to higher rates for playback.
@anon55914447 I’m sorry but yes really. Apologies if my earlier response was overly emotive, I will try to avoid that this time.
Please have a look at this AES paper by Bob Stuart and Peter Craven if you have not done so already.
http://www.aes.org/tmpFiles/elib/20160109/17501.pdf
This paper points out a number of reasons as to why 192/24 bit sampling rates are important, such as being able to minimize the detrimental effects of digital filters. The paper also describes a number of other developments in the area of psychoacoustics about how our understanding of hearing has moved beyond the basic physiological facts that we cannot hear above 18/20Hz and the importance of other issues such as timing. This is more than sufficient to show that there is much more to these issues than the basic theories relied on in the Xiph posting.
The Xiph posting is self-published and while it has a number of cross references most of them are to websites and only a few are to more rigorous scientific or engineering work. In contrast the Stuart and Craven paper is published in a peer reviewed journal and has over 90 references to other scientific articles. Not that this in itself means that they are right and the Xiph posting is wrong, but to me at least it gives a lot more credibility to what they say.
Furthermore I don’t see it about being right and wrong or “debunking” or demonstrating something is outdated but being open to and opening up new understandings that go further. That was what I was trying to express, rather poorly I’m afraid, with my Newton/Einstein analogy. (Apologies to all and to Newton and Einstein too!) I find that the Xiph posting is insisting on some basic theories that have a lot of value but holding on to them so firmly it cannot see that there might be additional value in things beyond that. That is what I was trying to express. (ie. if we keep insisting only Newton is right we cannot see that Einstein has more to show us.)
One of the few rigorous articles referenced in the Xiph posting to support their claim (even though they admit it does not fully support them) is “Coding High Quality Digital Audio” by Bob Stuart. Unfortunately the date of the article is not clear but judging by the latest dates of cross references it was written in the late 1990s. In the article Stuart does conclude that “the use of sample rates above 96kHz … cannot currently be justified”
What is interesting to note is how several times in this article Stuart says about things that this is his current thinking or position or says things like “if I were forced to right now” (ie. sometime in the late 1990s). He is open to new understanding and thinking and recognizes his own limitations and the limitations of the current theories and that his current understanding might not have it all right and can be improved upon. You can see how some of his understanding has grown and changed in the 2014 AES paper from this earlier one and that he has added other insights.
I could go on more but I‘ve probably said too much already that no one will read but at least I got if off my chest!
5 Likes
Whilst it’s pretty much a given that Bob Stuart is cleverer than me in the area of digital audio, I still think it’s worth not leaving our critical faculties at the door when it comes to this sort of subject, especially when the corollary is that Meridian have to keep selling stuff, and everyone is looking for the NBT.
The link to the AES paper (conference paper not journal) is broken, but assuming it’s the same one as the abstract Viruskiller linked to, then the comments below it are quite illuminating and definitely worth reading.
It also contains some contentious methodology, not least:
Analog Devices’ Bob Adams (designer of the first successful IC
asynchronous sample rate converter for AD) pointed out that the shapes
of the filters used in the study were pathologically selective compared
to normal commercial practice, having a much longer impulse response (by
about a factor of 4x) than filters commonly used. So the take-away from
this paper, at least for me, is that one can design an audible
linear-phase digital filter whose transition region is above 20 kHz if
you make the impulse response long enough.
The marketing angle has been handled poorly, imo. I mean who in their right mind, especially with a credible techincal background presents in front of this with a straight face?
That’s hardly a straight face…
Were you there ?
If so…cool. Maybe you got chance to query this directly.
If not, then you cannot really take a pic like this and judge what was said, in what context, to what audience, etc.
I have seen this presented and it is a context slide. It sets the scene for general “where are we” position. It isn’t a science presentation, does not explain MQA in any way and it does clearly say “Notional”.
I am not saying that MQA marketing and general education of the masses is great either (quite frustrating IMO). That does need to come, but, let’s not become like tabloid journalists and paparazzi and draw conclusions from snapshots like this.
2 Likes