Hey guys, just wanted to drop by to let folks know that I'm running a blind test of MQA Core Decode vs. original hi-res...
As you know there's been plenty of hub-bub over the last few years in the audiophile world about the "revolutionary" MQA technology with all kinds of claims on how it improves sound. Hopefully I can gather some data to compare what MQA decoding sounds like with your own DAC especially for those of you who don't have TIDAL and may be wondering about the difference between a stream that's MQA Core decoded compared to say a non-MQA 24/96 stream.
After you listen, I would of course love to know your perception of differences. The tracks are randomized and "blinded". Tell me which you prefer and how much difference you perceive!
Test instructions and files can be found here:
INTERNET BLIND TEST: MQA Core Decoding vs. Standard Hi-Res Audio (https://archimago.blogspot.com/2017/07/internet-blind-test-mqa-core-decoding.html)
Thanks all, have fun... Feel free to spread this test around. Taking blind test survey results until September 8.
Arch
We regularly visit your blog :))
https://hydrogenaud.io/index.php/topic,107666.msg942221.html#msg942221
Results being posted now...
Detailed description of Procedure (Part I):
https://archimago.blogspot.ca/2017/09/mqa-core-vs-hi-res-blind-test-part-i.html
"Core" results (Part II):
https://archimago.blogspot.ca/2017/09/mqa-core-vs-hi-res-blind-test-part-ii.html
Thanks to all who gave this a try and submitted results :-).
The true benefits of MQA can't be realized in short term tests like these. It requires long term relaxed listening to flesh out the lower fatigue and less digititis in MQA vs non-MQA digital streaming.
50-80yr old men will experience far less or no hot flashes and menstrual cramps associated with non-MQA smeared digital long term listening.
Read carefully. MQA based on most recent neuroscience and its revolutionary design sounds indistinguishable from the full HIGH Resolution files packed into a tiny file.
Archimago just proved how well the deblurring works!
It's a revolution!
Edit:
Even the most critical experts like our member amirm welcomes the results:
https://audiosciencereview.com/forum/index.php?threads/archimagos-mqa-listening-test-results.1947/#post-51443
That amirm lately mentioned to be a fellow of BS can't be a reason he is deliberately ingenuous with MQA.
"MQA is providing similar fidelity to PCM at lower data rates/file sizes." ...like every other lossless compressed codec?
Still though, similar, how are people not getting lossless?
I keep checking those other forums but this is the only one I like.
"MQA is providing similar fidelity to PCM at lower data rates/file sizes." ...like every other lossless compressed codec?
Still though, similar, how are people not getting lossless?
That is actually irrelevant. That a format is "lossless" does not mean there is no loss when you reduce, say, 32-bit floating-point PCM to 16 or 24 bits to fit it into a FLAC file.
(I have issues with MQA being called "lossless" (as in "like every other lossless" above). But it is kind of irrelevant, for the same reason.)
Sorry I don't know anything about MQA, I read it was lossless and I was commenting about lossless formats in general. I see it like the next Pono hype bs.
The format is absolutely lossy. You get ~16 bits worth of dynamic range from a 24 bit file where the ~8 least significant bits are used to lossily encode frequency content above baseband content. Baseband in this case meaning frequencies up to 22kHz or 24kHz.
http://archimago.blogspot.ca/2016/10/musings-keeping-it-simple-mqa-is-codec.html?m=1
The format is absolutely lossy.
Yeah, yeah, but 3 out of 2 audiophiles in the room can't hear this, if you read the patent.
Anyway, these tests are of little relevance, as the dotard messiah (http://www.theabsolutesound.com/articles/let-the-revolution-begin/?page=2) has spoken now.
"I knew Albert Einstein. Albert Einstein was a friend of mine. Robert, you're no Albert Einstein."
You get ~16 bits worth of dynamic range from a 24 bit file where the ~8 least significant bits are used to lossily encode frequency content above baseband content.
Is it established that MQA has no loss up to the 15th bit?
Is it established that MQA has no loss up to the 15th bit?
For MQA CDs the magic surely needs less.
The format is absolutely lossy.
Yeah, yeah, but 3 out of 2 audiophiles in the room can't hear this, if you read the patent.
Anyway, these tests are of little relevance, as the dotard messiah (http://www.theabsolutesound.com/articles/let-the-revolution-begin/?page=2) has spoken now.
"I knew Albert Einstein. Albert Einstein was a friend of mine. Robert, you're no Albert Einstein."
Thanks for the link, This quote is pretty amazing:
"The new psychoacoustic paradigm recognizes that human hearing didn’t evolve to hear tones and beeps. Rather, it is exquisitely tuned to detect the sounds of nature, which aren’t composed of frequencies and tones, but of transients of indeterminate and randomly varying frequency. In fact, many of the sounds of the natural world, an understanding of which confers important survival benefits, have no frequencies. Examples include crackling leaves, snapping twigs, the sounds of wind, rain, and running water."
In fact all sounds are composed of collections of sine waves or if you will, tones with various frequencies. This were findings of Fourier and Helholtz, offhand: a century or more ago and still considered to be valid by well-informed people.
Sorry about those last 3 words...
Thinking too about that phrase "Indeterminate frequencies". I've run FFT's of many natural sounds and there seemed to be no lack of observable frequencies. I'm sure many others have, too.
Or. did they invent a new science and new mathematics while I was not looking? ;-)
The format is absolutely lossy.
Yeah, yeah, but 3 out of 2 audiophiles in the room can't hear this, if you read the patent.
Anyway, these tests are of little relevance, as the dotard messiah (http://www.theabsolutesound.com/articles/let-the-revolution-begin/?page=2) has spoken now.
"I knew Albert Einstein. Albert Einstein was a friend of mine. Robert, you're no Albert Einstein."
When blithering know-nothings like Harley reference Kuhn, I reach for my gun.
In fact all sounds are composed of collections of sine waves or if you will, tones with various frequencies.
That is, I guess, up to terminology - what one means by "composed", and how it could be misunderstood (even deliberately, since we are dealing with the marketing mumbojumbo-land).
As you know very well, it does not matter whether or not they were "composed" that way originally, but whether they were composed that way originally, but whether the signal can be
decomposed that way and then recreated for playback - and to an error tolerance beyond audible transparency. I guess that using the term "composed" fuels audiophool rhetoric based on the (convenient) misunderstanding that it must mean "originates from".
Heck, one could have used square waves or triangle waves instead of sines/cosines. Any choice of basis will do in theory, and therefore the trigonometric basis will do. No matter whether the audiophool thinks the sound originates from something that does not at all look like pure tones. All it takes is the sufficient number of bits - an imprecise statement yes, but one will anyway have to check real-world engineered implementations against
measurements of human hearing, which is the proof of the pudding no matter what.
Or, did they invent a new science and new mathematics while I was not looking? ;-)
I wonder what Bell Labs could have done if they already knew what MQA Bob knows now.
"The new psychoacoustic paradigm recognizes that human hearing didn’t evolve to hear tones and beeps.
Yep, so Harley then provides the "latest" research in support of this:
Human Time-Frequency Acuity Beats the Fourier Uncertainty Principle (https://www.google.com/search?q=Human+Time-Frequency+Acuity+Beats+the+Fourier+Uncertainty+Principle&oq=Human+Time-Frequency+Acuity+Beats+the+Fourier+Uncertainty+Principle&aqs=chrome..69i57&sourceid=chrome&ie=UTF-8)
So I reminded Robert that this type of research most likely involves double blind testing for results...you know, the type he completely rejects as flawed (http://www.enjoythemusic.com/tas/183/editorial.htm).
Ok, so on to the paper:
We use two test stimuli [21]. The first is a Gaussian packet...
Ummm, so we evolved to hear Gaussian packets. ???
Ok, now what's that second hit on Google"? https://arxiv.org/pdf/1501.06890.pdf (https://arxiv.org/pdf/1501.06890.pdf)
Hmm...
@Archimago The comments section at your blog gives the best insight one can find in short what MQA is really about.
I can't make up such stupid sarcastic writing as the real swallowers seem to honestly believe :)
@Archimago
The comments section at your blog gives the best insight one can find in short what MQA is really about.
I can't make up such stupid sarcastic writing as the real swallowers seem to honestly believe :)
Interesting how so many of the participants in this test had no choice but to engage in blind testing, but still seem to use sighted, slow-switched, non-time-synched evaluations with a possibility of level mismatch as well, for *proving* the relevance of their biases.
I'm going to be honest here. This has nothing to do with audibility here. But I would say regular Hi-Res is better than MQA Core based on several points:
Need of prosperity hardware and software and forbidding of decoding the audio stream into a usable format for a device that doesn't support it. (https://en.wikipedia.org/wiki/Master_Quality_Authenticated#Hardware_and_software_decoders)
The fact that regular Hi-Res files is typically a change of sample rate and bit-depth that can easily be converted.
The fact that it's just another Super Audio CD type of attempt all over again.
I'm going to be honest here. This has nothing to do with audibility here. But I would say regular Hi-Res is better than MQA Core based on several points:
Need of prosperity hardware and software and forbidding of decoding the audio stream into a usable format for a device that doesn't support it. (https://en.wikipedia.org/wiki/Master_Quality_Authenticated#Hardware_and_software_decoders)
The fact that regular Hi-Res files is typically a change of sample rate and bit-depth that can easily be converted.
The fact that it's just another Super Audio CD type of attempt all over again.
Agreed. DRM is the obvious goal of MQA.
Agreed. DRM is the obvious goal of MQA.
I also thought so when it arrived, but it seems to me that it offers no DRM opportunity, i.e., no way of locking down the content to prevent copying? That looks like about no more DRM than HDCD or like Dolby-encoded retail music cassettes were back in the day before they were reverse-engineered and implemented in free-of-charge software: Playing it on non-[MQA|HDCD|Dolby]-aware gear would give you the music (unlike DTS-CD, mind you!), and a certain fidelity that is not outrageous - take that as "relatively speaking", for the cassettes. MQA'ed CDs - yes they exist already - can be ripped and copied.
So it looks more like a "premium" business model to me. Of course, it was about selling encoder and decoder licenses. But that is nothing unique (the CD format was also licensed, of course). And it looks like creating an excuse for yet-another-remastering-round in order to sell the same music over again - now if we get some decent stuff out of that, I even think it is worth the idiocy.
(But "nobody" knows about everything that the MQA chips can really do? Is it conspiranoia to consider the thought that there is indeed some decryption in the chip, in case MQA catches on well enough to start delivering encrypted signal?
And what about fingerprinting ... )
In fact all sounds are composed of collections of sine waves or if you will, tones with various frequencies.
That is, I guess, up to terminology - what one means by "composed", and how it could be misunderstood (even deliberately, since we are dealing with the marketing mumbojumbo-land).
As you know very well, it does not matter whether or not they were "composed" that way originally, but whether they were composed that way originally, but whether the signal can be decomposed that way and then recreated for playback - and to an error tolerance beyond audible transparency. I guess that using the term "composed" fuels audiophool rhetoric based on the (convenient) misunderstanding that it must mean "originates from".
Heck, one could have used square waves or triangle waves instead of sines/cosines. Any choice of basis will do in theory, and therefore the trigonometric basis will do. No matter whether the audiophool thinks the sound originates from something that does not at all look like pure tones. All it takes is the sufficient number of bits - an imprecise statement yes, but one will anyway have to check real-world engineered implementations against measurements of human hearing, which is the proof of the pudding no matter what.
Wow. Very artistic. When you say, quote
" .... All it takes is the sufficient number of bits - an imprecise statement yes, but one will anyway have to check real-world engineered implementations against
measurements of human hearing, which is the proof of the pudding no matter what."
Are you saying, suggesting common (collective) 'subjective impressions', or the actual measurements -as represented by data/graphs?
In fact, what exactly are you saying ?
MQA: It's rather simple. Three core components;
1) MQA starts with encoding (of the original recording). There are known flaws in many of the early (and possibly current) A/D converters. MQA begins there, i.e. correcting known (temporal-related) errors..
2) The original recorded Master file is used.
3) Decoding (MQA style): It has long been recognized that a passband extending to 40-50 KHz. is a desirable objective, as MQA tests (and prior) have demonstrated.
The decoding process is simple enough; the standard passband of 22.05 KHz. (16/44.1, Red Book) is copied bit-for-bit. Any frequencies above this (ultrasonic?) band is simply compressed -and expanded during MQA playback,
It appears not very different to HE-AAC's "SBR" component.
This (ultrasonic component) can be observed by MQA's bit-rate of 1,550? Kb/s (or thereabouts); 1411 Kb/s Red-Book and the added folding/unfolding of higher frequencies (above the prevailing noise floor).
As a passionate music lover, if MQA proves superior, I'll embrace it.
I must say however I'm hearing some great "sounds" even with streaming at 60-128 Kb/s (AAC) !
Anyway you look at it, Robert (Bob) Stuart and team should be applauded for his efforts -if nothing else.
You like -you buy (or stream). If not, carry on -as usual.
pj
It has long been recognized
[citation needed].
Scientific review article, please.
(MQA style): It has long been recognized that a passband extending to 40-50 KHz. is a desirable objective, as MQA tests (and prior) have demonstrated.
It is barely established that a 20kHz filter done wrong may be spotted under laboratory conditions. There is indeed a Meridian paper so don't expect it becomes to clear.
The decoding process is simple enough; the standard passband of 22.05 KHz. (16/44.1, Red Book) is copied bit-for-bit. Any frequencies above this (ultrasonic?) band is simply compressed -and expanded during MQA playback,
It appears not very different to HE-AAC's "SBR" component.
This (ultrasonic component) can be observed by MQA's bit-rate of 1,550? Kb/s (or thereabouts); 1411 Kb/s Red-Book and the added folding/unfolding of higher frequencies (above the prevailing noise floor).
You can't copy content below a certain frequency. You need to filter. In this case near the audible band. On reconstruction you need another filter to reconstruct it together with the aproximation of what once was the HF content. It has already aliased components from the prior lowpassing. The lower bits are shaped off. From the second unfold on you add tons of mirroring because of lousy upsampling.
There is even more going on.
If you apply the filter correction they claim to be important to a simple noise shaped 18bit 96kHz file you don't need that nonsense and the resulting file most likely compresses better as the MQA file.
1) MQA starts with encoding (of the original recording). There are known flaws in many of the early (and possibly current) A/D converters. MQA begins there, i.e. correcting known (temporal-related) errors..
BS Peter/Bob. Lets see reliable audible evidence for these "flaws". This is HA not The Absolute Clown Peter/Bob.
MQA starts by allowing aliasing distortion via an audiomoron filter.
2) The original recorded Master file is used.
Wrong again, this is purely alleged fiction by MQA shills.
3) Decoding (MQA style): It has long been recognized that a passband extending to 40-50 KHz. is a desirable objective, as MQA tests (and prior) have demonstrated.
See 1) Peter Bob
As a passionate music lover, if MQA proves superior, I'll embrace it.
I must say however I'm hearing some great "sounds" even with streaming at 60-128 Kb/s (AAC) !
Anyway you look at it, Robert (Bob) Stuart and team should be applauded for his efforts -if nothing else.
You like -you buy (or stream). If not, carry on -as usual.
pj
Sure thing Peter Bob.
Just how does one "carry on" when music starts being originally encoded with MQA aliasing distortion?
Nice trolling btw ;)
Leaving aside the malarkey in points #1 and #3
2) The original recorded Master file is used.
Promises , promises, Pedro. As customers of HDtracks.com have learned, the "Master file' can be whatever digital version the record companies supply. Is that a flat transfer of the correct, original, two-track mixdown master tapes the artist , producer and mixing engineer created and approved back in the day? There's certainly no guarantee. Is Meridian being upfront about this? Nope.
I also thought so when it arrived, but it seems to me that it offers no DRM opportunity, i.e., no way of locking down the content to prevent copying?
DRM doesn't have to prevent (digital) copying. In the case of MQA, the intent appears to be to prevent unauthorised reproduction at the original quality level.
... Playing it on non-[MQA|HDCD|Dolby]-aware gear would give you the music (unlike DTS-CD, mind you!), and a certain fidelity that is not outrageous - take that as "relatively speaking"...
MQA encoding provides the option to reduce the "non-MQA decoded" quality to several levels, right down to 2-bit resolution.
I suspect they leave the high bits alone to avoid blowing up equipment.
... (But "nobody" knows about everything that the MQA chips can really do? Is it conspiranoia to consider the thought that there is indeed some decryption in the chip, in case MQA catches on well enough to start delivering encrypted signal? ...
There is no magic MQA chip. There is strong encryption of part of the data stream. Follow the money. MQA didn't buy an expensive encryption product just to use as dither.
MQA encoding provides the option to reduce the "non-MQA decoded" quality to several levels, right down to 2-bit resolution.
Link?
There is no magic MQA chip. There is strong encryption of part of the data stream.
You are saying that there is no hardware-based decryption in MQA-aware DACs?
Link?
Oops. It was 4 bits, not 2. (Memory is the second thing to go with age. I forget what the first one was.)
Mans Rullgard's code analysis of the MQA Core decoder:
https://www.computeraudiophile.com/forums/topic/30572-mqa-technical-analysis/?page=9&tab=comments#comment-622783
https://www.computeraudiophile.com/forums/topic/30381-mqa-is-vaporware/?page=165&tab=comments#comment-715562
(Possible TOS violation - relates to encryption of content. Note that the code under discussion is published under an open source license.)
You are saying that there is no hardware-based decryption in MQA-aware DACs?
In a DAC with full MQA decode, there's code running in a standard processor chip (XMOS or FPGA). There is no custom chip. The same code runs in Tidal's software core decoder. Mans also looked at the "render" code, the second part of the process, that upsamples with a leaky filter. Low (processing) power DACs can only do the second "render" stage. It doesn't require DSP, just decoding the control bitstream and loading the selected filter.
Oops. It was 4 bits, not 2. (Memory is the second thing to go with age. I forget what the first one was.)
Mans Rullgard's code analysis of the MQA Core decoder:
Thanks for enlightening me. 4 bits 44.1 kHz quality, I take the chance that I will not get hit by a TOS#8 binning for dissing that.
So in the (unlikely) event that MQA really catches on, then the RIAA will of course switch over to the least-listenable option, to win over the then-already-extinct breed of pirates who share mp3s rather than lossless?
The true benefits of MQA can't be realized in short term tests like these. It requires long term relaxed listening to flesh out the lower fatigue and less digititis in MQA vs non-MQA digital streaming.
50-80yr old men will experience far less or no hot flashes and menstrual cramps associated with non-MQA smeared digital long term listening.
If that's true (" The true benefits of MQA can't be realized in short term tests like these. It requires long term relaxed listening ...") and have great doubt it is, there'e simply no point in MQA.
The SQ improvement should, literally, "hit-you-over-the-head" in distinction. There wold be simply no other reason (well there'e many reasons) for the concept to exist. And, if such a subtle Or questionable) improvement wold NOT have passed the critical listening skills of many.
I look forward to some "quality" time (lol) with MQA. I'm in no rush since I'm thoroughly enjoying whats before me now.
We'll see (I'll see) just how much once (preferably an MQA-imbedded CD disc/similar file) a test drive can be scheduled.
pj
If that's true (" The true benefits of MQA can't be realized in short term tests like these. It requires long term relaxed listening ...")
I always tell the truth, even when I lie
Understood. In which case I shall refer to MQA's own research (based on previous effort's) data -to be forwarded soon.
pj
Understood. In which case I shall refer to MQA's own research (based on previous effort's) data -to be forwarded soon.
pj
If you are Peter Jasz, this would be expected
In partial response to the request of "Scientific:" data, I'll offer this to start:
https://www.xivero.com/blog/hypothesis-paper-to-support-a-deeper-technical-analysis-of-mqa-by-mqa-limited/
Specifically, (5. 5.3) ; https://www.xivero.com/blog/hypothesis-paper-to-support-a-deeper-technical-analysis-of-mqa-by-mqa-limited/
This is a fine analysis of some core features and components of recoded music and auditory impressions.
This will be followed by specific research noted by the MQA patent team. In fact, I'll search (as anyone can) the MQA patent; within it will be references to the 'science' you request.
Keep in mind, although absolutely essential/crucial (technical analysis) collective subjective impressions hold far greater weight and consistency than ever-evolving scientific methodology.
peter jasz
Understood. In which case I shall refer to MQA's own research (based on previous effort's) data -to be forwarded soon.
pj
If you are Peter Jasz, this would be expected
And who are you ? Full name please.
pj
And who are you ?
Sherlock Holmes. Retired to Florida now.
Hello there: I am Peter Jasz, managing director (Audio) at ALLIED TV & Sound, 1558 King St. E, Hamilton, Ontario, Canada. We specialize in higher-end A/V systems. I am the high-end audio guy.
I have (and continue to) search high and low for high peformance cables at reasonable cost.One of the very best I've encountered under $100 are the XLO Pro 150 (1M). A rare combination of transparancy, detail, staging and musical body define this interconnect. The Pro 125 is a cheaper version, still rather good, but the Pro 150 is definitely better. XLO also have the Pro-100 interconnect that offers greater bass extension and dimensionality, but surprisingly doesn't seem to have as good transparancy.
If you can swing the extra dollars consider the Kimber SELECT series of interconect KS-1011 (single-ended) or KS-1111 (balanced).
peter jasz
"Full" unfold will require purchase of $$ MQA $$ hardware, correct Peter?
Always a pleasure to see new members only registering to spread some pointless marketing fluff not even remotely able to deliver verifiable facts on their own.
Now, now, remember the threads by members lamenting how poorly we treat innocent new shills and trolls, how rude, unfriendly and inconsiderate that is.
I hoped to circumvent that by using the word pleasure.
With 4bit music you can listen to silence much louder!
(http://stream1.gifsoup.com/view4/4862370/welcome-to-the-real-world-o.gif)
1) MQA starts with encoding (of the original recording). There are known flaws in many of the early (and possibly current) A/D converters. MQA begins there, i.e. correcting known (temporal-related) errors..
BS Peter/Bob. Lets see reliable audible evidence for these "flaws". This is HA not The Absolute Clown Peter/Bob.
MQA starts by allowing aliasing distortion via an audiomoron filter.
2) The original recorded Master file is used.
Wrong again, this is purely alleged fiction by MQA shills.
3) Decoding (MQA style): It has long been recognized that a passband extending to 40-50 KHz. is a desirable objective, as MQA tests (and prior) have demonstrated.
See 1) Peter Bob
As a passionate music lover, if MQA proves superior, I'll embrace it.
I must say however I'm hearing some great "sounds" even with streaming at 60-128 Kb/s (AAC) !
Anyway you look at it, Robert (Bob) Stuart and team should be applauded for his efforts -if nothing else.
You like -you buy (or stream). If not, carry on -as usual.
pj
Sure thing Peter Bob.
Just how does one "carry on" when music starts being originally encoded with MQA aliasing distortion?
Nice trolling btw ;)
Wow. Your brilliance is deafening. Before you start (or continue) with insults (that I have no issue with) but, I'd ask that you have the decency, the manliness, to identify yourself. Hiding behind some goofy profile pic without a name is cowardly.
(Whose the troll -that won't identify himself/herself ?)
pj
Not too bright this one.
Btw Peter, that's who's or who is, not "whose". You seem a bit unbalanced, look, I'm not a Commi from Russia, if that helps your paranoia.
Just a hint, if you are going to peddle MQA like you do magic cables, etc, you may want to add a bit of detail to your profile and signature. It's frowned upon on many forums, to be a peddler without identifying as such.
In partial response to the request of "Scientific:" data, I'll offer this to start:
https://www.xivero.com/blog/hypothesis-paper-to-support-a-deeper-technical-analysis-of-mqa-by-mqa-limited/
Specifically, (5. 5.3) ; https://www.xivero.com/blog/hypothesis-paper-to-support-a-deeper-technical-analysis-of-mqa-by-mqa-limited/
Nothing more than fiddling with filters. Of course filtering can be audible if it affects audible frequency.
https://www.xivero.com/blog/hypothesis-paper-to-support-a-deeper-technical-analysis-of-mqa-by-mqa-limited/#55
I copied the png files and attached here just in case the website silently replaced or removed them.
The MQA apodizing filter starts to attenuate high frequencies at somewhere around 10-15kHz, notice the line thickness and color when compared with the original signal and steep-filtered signal.
In this post the OP obviously prefers the steep filter used in foobar's SoX plugin instead of the one used by his DAC.
https://hydrogenaud.io/index.php/topic,110058.0.html
Regardless of audibility and preferences, such kinds of filtering doesn't require specialized hardware. I found nothing innovative about MQA apart from the "revolutionary" 4-bit decoding ability which I recently read.
There are SPC decoders with user-selectable filters 10+ years ago. 8)
http://www.foobar2000.org/components/view/foo_snesapu
(http://stream1.gifsoup.com/view4/4862370/welcome-to-the-real-world-o.gif)
Excellent deblurring example! Much better than MQA.
From the second unfold on you add tons of mirroring because of lousy upsampling.
A very nice find that illustrates the effect in action at ASR (https://audiosciencereview.com/forum/index.php?threads/mqa-bad-for-music.1378/page-5#post-55740) while the discussion itself is pretty weightless.
Origami explained:
https://youtu.be/vFl31p89VCM?t=4m49s