Most-Read Opinion Articles (Last 365 Days)
- 2016-06-01 - The Best of High End 2016
- 2017-01-01 - Our 2016 Products of the Year -- One-Sentence Summaries
- 2016-11-01 - The Best of the 2016 Tokyo International Audio Show
- 2017-02-01 - The Best of the Worst CES in Decades: 2017
- 2016-08-01 - Future Sounds -- Sonus Faber’s Sf16 and New Directions in Hi-Fi
- 2016-10-01 - Bespoke One-Bit and Multibit DACs -- Assaults on the State of the Digital Art from Aqua–Acoustic Quality and EMM Labs
- 2016-12-01 - Warsaw: The Best of Audio Video Show 2016
- 2016-07-01 - McIntosh Group on the Mediterranean
- 2016-09-01 - Danish Audio Trio -- Gryphon Audio Designs, Dynaudio, Bang & Olufsen
- 2017-04-01 - MQA One Year Later -- Suddenly, More Questions
- Written by Doug Schneider Doug Schneider
- Category: Monthly Column Monthly Column
- Created: 01 April 2017 01 April 2017
On April 1, 2016, we published “Myriad Questions About MQA,” my article about Master Quality Authenticated (MQA), a digital music format for high-resolution music files developed by Meridian Audio and now spun off as MQA Limited. (Audiophiles define high-resolution music file as a PCM digital file with a bit depth and/or sample frequency higher than the Compact Disc standard of 16 bits and 44.1kHz, aka 16/44.1.)
From what I’ve learned about MQA, it promises two main benefits. One is the claimed lossless compression of files of resolutions as high as 24-bit/768kHz to 24/44.1 or 24/48 (depending on the original sampling frequency), the results being files roughly 50% bigger than a 16/44.1 file. Since the process is claimed to be lossless, there should be no reduction in sound quality -- when the file is decompressed, all of the original information should still be present. MQA files are stored in such industry-standard file formats as FLAC or ALAC, so even audio systems and DACs that can’t decode MQA files should be able to recognize and play them.
The other claimed benefit of MQA is an audible one -- the reduction or elimination of what MQA calls “temporal blur” or “time smearing,” which, they say, occurs mostly in the upper frequencies, when a recording is converted from an analog to a digital signal, primarily with converters that use brickwall filters. To “unblur” or “unsmear” the signal, information about the original A/D converter is buried in the file when it’s encoded as an MQA file, so that an MQA-compatible D/A converter at the decoding-and-playback end can read this information and correct for it.
Some of my questions about MQA go right to the heart of what it’s supposed to be used for and what it’s claimed to do. For example, with the constant increases in cheaply available bandwidth, do we even need the compression MQA offers? And is MQA truly lossless compression, as claimed? Can MQA really fix “temporal blur” or “time smear”? And does this blurring or smearing even exist? After all, before MQA, no one had heard these two terms applied to digital audio. I also questioned the validity of the demonstrations of MQA conducted for audio writers at the 2016 Consumer Electronics Show. For most of the comparisons described, it wasn’t clear if the same source files (i.e., master recordings) were used for the MQA and non-MQA versions of the same songs being played back to back. Obviously, to assess whether MQA is doing what is claimed for it, we need comparisons of “apples with apples” in which identical source files are used.
To the best of my knowledge, my article was the first to raise questions about MQA’s claims. Some earlier articles about MQA that I’d read seemed to be rehashes of the company’s promotional campaign; others raved about the technology, saying that it could provide sound quality beyond anything we’d heard before, yet without asking if the codec actually worked as described. Robert Harley’s “Beyond High-Resolution,” published on The Abso!ute Sound’s website in July 2015, promoted MQA’s benefits wholeheartedly, without enough (or any) questions related to whether or not MQA could work as claimed.
I received a lot of positive responses to my article, both from readers and from people in the industry, who were thankful that someone was finally taking a critical look at MQA. I was contacted by others who shared similar concerns and wanted to express their opinions, but lacked a vehicle for doing so.
But not everyone was happy with my piece, including some writers who felt put on the spot -- they’d written positively about MQA, and had taken part in the demos I was asking questions about. Some of these writers thought I was pointing a finger directly at them, and poured out their anger at me on audio forums. Some saw me as a “hater” who was complaining about MQA for no other purpose than to annoy others. I was OK with that -- if I’m going to put my opinions out there, I’d better be able to take the responses to them.
I don’t know what the folks at MQA Ltd. thought about my article, but I have to suspect they weren’t too thrilled by it. A few days before it was published, I received a call from Bob Stuart, who used to run Meridian Audio but is now in charge of MQA Ltd. We talked for about an hour, but as he seemed to have no real answers for my questions, nothing he said convinced me that I should change anything in my piece, and it went online as planned.
During that conversation, I offered to send Stuart some original digital recordings as well as digital masters that I have access to -- some new and others decades old, at varying resolutions. He could give them the MQA treatment, send me the MQA versions, and I and the recordings’ original producer could hear any differences for ourselves. I never heard back from him about that, which surprised me -- I thought he’d jump at the chance to prove his claims.
The other day I was talking with Mark Waldrep, founder of AIX Records, who’s written extensively about MQA in his newsletters and on his website, Real HD Audio. Over a year ago, Waldrep, too, offered his files to MQA and, from what I understand, even sent some to the company; but as of the day I’m writing this, Waldrep hasn’t gotten anything back either. He told me that he, too, was surprised at this nonresponse.
The one person who did get some of his files subjected to MQA was John Atkinson, editor of Stereophile. He detailed his experiences in “Listening to MQA,” a Follow-Up originally published in the September 2016 issue and now posted on Stereophile.com. This article was particularly interesting because it was the first time I was confident that the same masters had been compared. The most compelling part of the article came near the end:
After doing all of my formal comparisons, I subjected myself to a sort-of-blind test. I created an Audirvana playlist that randomly mixed MQA and non-MQA files, and pressed Play. I then went into my test lab, which is in the room next to the listening room, to begin measuring some of the products in the review queue. At irregular intervals I returned to the listening room and made a decision, MQA or non-MQA, before looking at the [Meridian Prime D/A headphone amplifier]’s front panel to see what was playing. I scored four out of seven correct; though this is insufficient to prove formal identification, I feel that it is relevant information.
Because I trust John’s hearing and believe him to be honest, I, too, feel that what he said is relevant -- but not necessarily for the reasons that some might think. As I pointed out to John when we met near Chicago in February, at Vivid Audio’s launch of their Giya G1 Spirit loudspeaker, and in public on the Audio Asylum forum: Statistically, four out of seven is not considered significant, as he knew and had stated. Furthermore, had he done one more test and guessed it correctly, that still wouldn’t have been compelling, because it still means that he was wrong three times out of eight. And had he guessed incorrectly that eighth time, four out of eight would be dead in the middle, or the same results as tossing a coin.
What I take from this test of John’s is that it’s not easy to distinguish a standard audio file from its MQA version, even when you know the source material extremely well. In some ways, this is a good thing for MQA -- it appears to do no harm. But what about the sonic benefits that MQA is supposed to provide? Perhaps they aren’t all that apparent -- or apparent at all. And maybe that’s why getting a public apples-with-apples A/B comparison of MQA seems more difficult than getting a straight answer out of President Trump’s press secretary, Sean Spicer. As I mentioned in my article last year, I was skeptical about the MQA comparisons being done -- or, in most cases, not being done at all. For example, just over a month after my article was published, at High End 2016, in Munich, Bob Stuart did a public demonstration of MQA recordings, but with no comparisons, even when asked for some. It was one of the strangest things I’ve witnessed at a press conference -- there he was in a room full of audio journalists, demonstrating his MQA technology, but making no effort to prove his claims.
Bob Stuart at High End 2016
The closest thing to a proper comparison of MQA and non-MQA files I know of was done for our own Brent Butterworth, who attended a brief, private -- not public -- A/B demo at T.H.E. Show Newport 2016. Listening through headphones, Brent found the differences “subtle,” but couldn’t say which was better or worse -- just that he thought he could hear slight differences. On the other hand, he did point out that the reduction in file size from uncompressed hi-rez to MQA was not subtle -- so at least there’s that.
Which brings us to the state of MQA today. The audiophile press doesn’t seem to be covering it with the same vigor that they were a year or so ago; in fact, I see little about MQA in the same magazines and websites that seemed very supportive of it in early 2016. Nor have I written much about it -- this is the first time since April 2016 that I’ve sat down to write something substantial.
MQA-equipped Mytek Manhattan II DAC
Some manufacturers now include MQA decoding in their DACs -- AudioQuest, BlueSound, Brinkmann, Cary Audio, Meridian, MSB Technology, Mytek HiFi, and NAD are the ones I know of. Yet MQA is nowhere to be found in products from Ayre Acoustics, Benchmark Media Systems, Boulder Amplifiers, Bryston, Chord Electronics, dCS, EMM Labs, Hegel Music Systems, Marantz, Meitner, PS Audio, Schiit Audio, Simaudio, or Soulution.
That’s not to say that MQA is dead or floundering -- far from it. The audiophile world is a small one, and MQA has made huge strides into the mainstream. Last January, at CES 2017, a big announcement came bundled with a bigger surprise: the Tidal music-streaming service would begin streaming MQA files made available from Warner Bros., the first major record label to partner with and endorse the codec. When Warner and MQA made this agreement, it was indicated that, over time, Warner’s entire catalog would be converted to MQA -- potentially, thousands of albums.
Shortly after that announcement, subscribers to Tidal’s premium HiFi service, which costs $19.99/month and provides CD-quality (16/44.1) streaming, saw a “HiFi/Masters” option in the Setup section. Having selected that, they could then choose the “Masters” option under “What’s New” to access MQA titles in Tidal’s library. Tidal includes this at no extra charge.
The big surprise was that you didn’t need an MQA-compatible DAC to enjoy part of the MQA process -- Tidal’s desktop app “unfolds” (MQA’s term) the compressed stream, up to a maximum resolution of 24/96. If you want even higher resolutions (provided the source file itself is of higher rez), then you’ll need an MQA-compatible DAC. Last year there had been hints that Tidal would begin streaming MQA, but no one expected that Tidal’s own software would do any of the unfolding.
Tidal’s addition of MQA files seems like a win for everyone, me included. Although I’ve been critical of MQA and still have serious questions about it, when I saw that it was available at no extra cost to Tidal HiFi subscribers, and that the desktop player would “unfold” MQA files up to 24/96, which you could then feed to your DAC, I thought, “Go for it! Stream away!” To me, it’s another source -- and, basically, a free one -- of hi-rez files. That’s a win.
I don’t have an MQA-compatible DAC, so I’m limited to the Tidal player’s “unfolding.” But that said, I’ve listened to MQA files via Tidal HiFi, and so far the results have been inconclusive. For titles that Tidal offers in both 16/44.1 FLAC and MQA that I’ve compared, I can’t confidently say that I’ve heard any differences. As a result, I’m sure that if I did a pseudo-blind test, as John Atkinson did, my results would probably be similar to his -- basically, the equivalent of guessing. For MQA titles of which I have hi-rez versions on my music server, I’ve heard more substantial differences, but which ones I preferred varied between MQA and non-MQA hi-rez, so that doesn’t tell me if MQA is good or not. More important about these comparisons is the same thing I’ve been complaining about since last year: I have no way of telling if the masters used for my non-MQA recordings are the same as the masters used for Tidal’s MQA versions, so it may or may not have been a comparison of apples with apples. But this is just the beginning -- I plan to listen more, and perhaps find a way to identify MQA and non-MQA tracks that I do know, with 100% certainty, were derived from the same sources, to better assess what I’m hearing. Until then . . .
On February 16, 2017, an announcement was made that seemed like a boon for MQA: Universal Music, another of the three major record labels, would begin offering their content in MQA format for streaming. I read that on Billboard.com -- MQA now seems to be something that people other than audiophiles care about.
Back in the audiophile world, some not-so-positive things were happening for MQA. About a week before the Universal announcement, electronics manufacturer Linn Products Limited, of Scotland, published on their website an article by Jim Collinson, Linn’s Designer and Digital Marketer: “MQA Is Bad for Music. Here’s Why.” Calling MQA an “outright land grab,” Collinson didn’t go after the codec for the technical claims made for it, but for its commercial aspects. He implied that MQA’s real purpose isn’t playback quality, but control of music distribution from source to consumer, taking a little bite of profit at each step along the way.
Obviously, no one can blame MQA Ltd. for trying to make money -- that’s any company’s goal. However, Collinson pointed out that MQA works by being inserted into the entire recording-and-playback process, from the encoding at the file-creation side to the conversion at the listening end -- all of which is very much like a copy-protection scheme: “there is a form of fingerprinting in the file that will check that at each stage of the production and distribution process MQA has been paid.” Collinson points out, correctly, that consumers abhor any sort of copy-protection mechanism, whether in regular computer software, movies, or music.
Collinson also called into question the little light on MQA-compatible DACs that illuminates when an MQA stream is played. MQA claims that this is done so that “you can be sure you’re hearing exactly what the artist approved in the studio.” Collinson points out some implications of this:
Now, ostensibly, this is a quality assurance check for the customer: if the little MQA light comes on, then I know that this file is legit. When in reality this is actually a quite masterful way of painting every other recording as inferior -- when exactly the opposite may be the case -- unless they are produced, distributed, downloaded and played via their approved supply chain. I could be playing a 24-bit 192kHz file straight from the studio, delivered to me in person by the artist herself and yet I am left with the feeling that this file is illegitimate; I’m not greeted by the warming glow of the MQA branding.
After Collinson’s article was published, I found many who agreed with him, particularly manufacturers already dubious about MQA. But scouring the Internet forums, I saw many criticize it because it came from Linn, a direct competitor to Meridian, where MQA was born. Some saw it as evidence of Linn’s corporate envy. Read it and decide for yourself.
Then something more damning occurred. HighResAudio.com, a German hi-rez-audio download site, had begun selling MQA titles in May 2016. Then, in the first week of March 2017, HighResAudio issued an explosive Facebook post stating that they would stop selling MQA files, mostly because of problems they were having verifying MQA files as being genuinely hi-rez -- something HRA can do with the other kinds of hi-rez files they sell. This had mostly to do with MQA not supplying HRA with tools to decompress their files to their original states in order for HRA to analyze them and determine their resolution, which is part of the quality control HRA promises.
The post was subsequently deleted from Facebook, but John Darko of DigitalAudioReview.net, also based in Germany, snagged it before it was taken down and managed to talk with HighResAudio’s owner, Lothar Kerestedjian, over coffee. On March 8, Darko published on his website an article, “HighResAudio.com Calls for a Deeper Technical Analysis of MQA.” It includes the original Facebook post, as well as Kerestedjian’s current thoughts on MQA, which make it clear that he now shares many of my concerns about the technology. Near the end of the article are reprinted the contents of an e-mail from Kerestedjian to Darko that mentions Stephan Hotto’s “Hypothesis Paper to Support a Deeper Technical Analysis of MQA (Master Quality Authenticated) by MQA Limited.” (The original link provided by Darko, to a Dropbox location, no longer works. The link I provide here accesses the same information on Hotto’s corporate site.) Hotto’s article is a must-read for anyone interested in the technical side of MQA, and in knowing if it’s all it’s claimed to be.
I read Kerestedjian’s criticisms, and then Hotto’s paper, which brings up points I made in my article a year ago, in addition to many others I hadn’t considered; e.g., the problems with aliasing created by the MQA process, and the decreased signal/noise ratio that results from MQA processing. My first reaction was to sit back in my chair, look up at the ceiling, and ask myself, “Why in the world did it take so long for someone else to bring up this stuff?” I’m still wondering. And now I’m wondering why others still aren’t bringing it up.
It’s obvious that, even after all this time, there are still too many questions about MQA that need to be answered before audiophiles can accept it as superior to the high-resolution digital files that have been available to us for many years now -- because it’s possible that MQA is a step in the wrong direction.
. . . Doug Schneider