My new article series on MQA.

I'm glad you will give me a chance in future articles to explain more about what happens to create those MQA files and why I see some benefit here.

I think the Archimago test may be proving that some don't hear differences between 16/44 and higher sampling rates. It also is subject to wide variability of playback systems.

 

I've mentioned this before in other posts.

But 'internet' tests such as the conducted by archimago don't fully test MQA.

Yes, the test is labeled as 'core MQA', but it is incomplete and confusing in the bigger discussion of MQA as it has multiple parts:

1. core MQA, which can be done all in software, but is only a partial origami unfolding of the possible data rate.

2. full origami data rate unfolding - only possible with a licensed/enabled MQA DAC

3. data 'deblurring'/filtering - again only possible with a licensed/enabled MQA DAC.

Discussions and comparisons confuse or don't understand or specify WHAT level of MQA listening they have conducted.

Like I said on a different forum - it's like someone trying Chex Party Mix but only picks out the Wheat Chex bits to determine if they like or don't like "it".

At least specify that is how 'you' arrived at your decision, otherwise you have to grab the full handful to really get the full effect.

 

I'll have to wait and hear in my home the difference, if any, that I can perceive before rendering an opinion.

One of the problems MQA tries to solve is one of bandwidth in an expanding streaming world and still sound like a high resolution recording this would save streaming companies a lot of money.

How successful MQA is is yet to be determined and whether or not it lives up to its goals.

 

Success is likely to hinge on the development of an MQA ecosystem. This has three parts I believe: 1. music availability from the labels, 2. hardware availability from the DAC/software app makers, and 3. some embrace from the pro audio community.
based on what I have heard #1 is a done deal, #2 is looking promising, and #3 is moving well except for some vocal mastering engineers that worry about making less doing mastering which is understandable.

 

Appreciate your piece and this thread. I would strongly encourage you, in your future pieces about MQA, to critically question what you've written here and in the article as assertions, in bold above. I think you've done a great job identifying MQA's two main sonic claims:

1. Reduce/eliminate phase (time) distortion, aka pre and post ringing from digital anti-aliasing filters
2. Compensate for/"undo" the idiosyncrasies of analogue to digital encoders in the recording chain.

There are major problems with both of these in practice, however:

1. MQA simply uses a minimum-phase filter, preserving time-domain accuracy in exchange for increased aliasing, aka harmonic distortion. I have no problem with this in principle - but the reason all the MQA vs whatever listening tests so far have yielded mixed results is that phase linearity vs frequency linearity is a trade-off. On some recordings, played on some systems, in some rooms, for some people, a linear-frequency filter might sound better. In other cases linear phase might sound better. And in other cases they'll sound essentially the same - or slightly different but not in a way that anyone would characterize as better or worse. Moreover, anyone with a CD player that has a switch for different digital filtering options already has "MQA" filtering available to them.
2. As for compensating for ADCs etc in the signal chain, that makes sense only if (A) it's known what digital equipment was used in the recording; (B) if the digital signal chain used in recording and mastering was simple enough to "compensate" for it. The vast majority of recordings - including ones still being made today - mix tracks recording on lots of different equipment, with a whole range of ADCs, D-A-D intermediate stages, and so on. The resulting sonic signature is such a muddle of ADCs and DACs that there's no clear filter signature to compensate for. It's like having a multitrack recording of 16 different people talking, all recorded in different rooms, and saying you have a magic box that can reverse "the room ambience" of the recording in order to make the voices sound totally neutral. There is no single "room ambience" - there's a mix that can't be untangled, and any filter you apply to that mess might change it slightly but won't be able to make it "better" with any consistency.

All that said, the main point of your piece - that MQA is an audiophile-friendly format that's also very streaming-friendly because it can adapt to different bandwidths (and different levels of end-user equipment decoding ability) - is a good one.

The problem for audiophiles - and people who care about sound quality in general - is that an unfolded high-res MQA file is lossy compared to PCM (FLAC, Apple Lossless, whatever), and even an MQA redbook CD-quality track is slightly lossy because the least significant bit is used to hold the unfolding/decoding info (similar to HDCD).

Finally - and perhaps most problematically, from a practical standpoint - much of the source material MQA is using comes to them from the labels in the form of high-res PCM digital masters. That means when you get a folded-up MQA high-res file, you're getting something that is partially lossy and adulterated compared to the high-res PCM original. In other words, you're getting an inferior product so that the file also can be used for streaming applications instead of the mp3, AAC, or redbook FLAC files that currently populate the streaming services.

 

Speaking of ecosystem....Astell & Kern just announced MQA support:

IRIVER To Add MQA Support For Line Of Audio

Streaming Devices

MQA Support Will Be Added To Streaming Devices Starting With ACTIVO, Then Added To Other Streaming And Audio Devices

Las Vegas, NV – IRIVER, the parent company of Astell&Kern, the global leader in high-resolution portable audio players, announces support for MQA decoding and playback will be added to the ACTIVO portable high-resolution audio player in January 2018.

MQA’s award-winning technology captures and reproduces the sound of the original studio master in a file that’s small enough to stream and download easily. With the TIDAL Masters integration, users can instantly stream thousands of MQA tracks on their players.

James Lee, CEO of IRIVER said, “IRIVER is committed to bringing the latest innovation and technology to consumers. We are happy to add MQA support to the new ACTIVO audio player, powered by Astell&Kern’s new TERATON sound solution. We will continue to add MQA support to other streaming and audio devices in the future.”

Mike Jbara, CEO of MQA, commented, “It’s exciting that MQA’s technology will be integrated into IRIVER devices. IRIVER’s addition to the list of MQA partners is a milestone in the creation of an environment where all consumers can easily enjoy high quality audio.”

The ACTIVO CT10 is the first high-resolution audio player from groovers Japan and features the new TERATON sound module by Astell&Kern, allowing high resolution audio playback up to 24bit, 192 kHz high resolution PCM audio and up to double-rate DSD (converted to PCM). The ACTIVO CT10 also supports music streaming services including TIDAL and groovers Japan.

The ACTIVO CT10 will be displayed in the Astell&ASPR booth during CES2018, Central Hall, booth # 18218.

 

I will address these points in the future articles but I would make the following counterpoints:

1. Based on my understanding, MQA is not lossy. MQA claims to have proven this mathematically.
2. There are other filters besides deblurring used.
3. Timing is more important sonically than FR so the trade-offs there are acceptable.
4. The use of some DACs and ADCs and mic preamps does not yet have a corresponding filter set but my understanding is that that is less than 10% of what has been in use and MQA is building filters every day to ensure coverage of the universe of options.
5. You are not "getting an inferior product" with MQA as a hirez file as it still is a hirez file versus a typically non-hirez file source for streaming. MQA simply represents a compact way to stream the higher resolution elements.

 

Is the idea that MQA software will only be distributed via streaming services, or will downloads be made available? I ask because it seems Tidal is struggling to stay afloat offering CD quality audio, which makes me wonder what the market for high res streaming looks like.

 

My posts were not meant to be hostile. What I was asking for is an objective review that addresses the popular (read known common) issues of MQA instead of a single sided report based only on interviews of the people most likely to profit from MQA. That was a suggestion to the writer and not a comment about MQA.

I didn't want to turn this thread into another MQA bashing vs hype thread and am trying hard to not respond to the misguided hype retorts (aka your second comment above) because that has been covered in the forum in previous MQA threads.

 

Because the labels are striving for ways to make more money using DRM...

 

There is no DRM in MQA.

 

You are being a bit ridiculous here. This first article is on the music availability benefits of MQA and the second article is specifically meant to be about the technical elements. You are arguing with the structure of the series here, not the content of the series. As I writer I felt there was enough content to do multiple articles.

And it wasn't single-sided. I talked to numerous people outside of the MQA team after my interview with Ken in part to confirm his statements and in part to get other perspectives. Nothing was revealed that negated the them of millions of new hirez tracks are coming. I talked to several competitors in the streaming business as well in both the streaming and hardware side of the biz.

 

Thank you for your reply. I'd like to respond to each of your points:

1. MQA is indeed lossy, and they've admitted as much - and this is something you appear to currently be getting schooled about over at the ComputerAudiophile forums - and while I certainly could have missed a comment, from what I've seen you very notably have not repeated your claim there that MQA is not lossy, instead pivoting to the benefits of the business model, and then following up with an iRiver press release about adopting MQA. I try to be fair-minded at all times, but this leads me to question your open-mindedness on this topic. I certainly hope I am mistaken.
2. Please tell us what other filters are used.
3. "Timing is more important sonically than FR" is an assertion, nothing more. In its breadth and generality, is a meaningless and unsupportable claim. With respect, you should know better.
4. It's not about about whether or not some ADCs and preamps have a filter set yet. (And "they have filters for more than 90%" is a highly suspect claim to boot). If you're running the stereo master tape/file of a modern 32-track recording through MQA, what filter set(s) are you supposed to apply? Such a recording could use several different mic preamps, multiple ADCs and DACs, mixed original sample rates, and so on, in dozens or 100s of permutations. To cite just one very typical example, an instrument is recorded through a mic preamp and ADC onto a multitrack track. During final mixing, at a different studio, the engineer decides to re-amplify the track, playing it back through a DAC and an amp and speaker, and re-recording it through another mic preamp using another ADC. This happens a lot in rock music - and it's fanciful, and borderline fraudulent, to claim that any digital filter can compensate for this complex chain of different equipment - all of which is just on one track of what could be a 32 (or more) track recording.
5. "MQA simply represents a compact to stream the higher resolution elements" is a true statement only if MQA's high-res folding process is not lossy. But it is.

 

I cannot tell if you are shilling for them or are ignorant of the fact that MQA enables DRM. THAT is why the RIAA loves it so much.

This is very obvious from engineering analyses of the code.

Would you please disclose any affiliations you have with them, Stuart, or licensees??

 

Where is your evidence that DRM is part of MQA?

I have no affiliation with anyone on the MQA team. I know them but I am an independent journalist.

 

Where has MQA admitted that their algorithm is lossy? If you are right I need to see some evidence as the MQA team is telling me the opposite.

 

I am just not interested in MQA at all. After reading this entire thread, I still don't know what it is. But, after having listened to comparisons twice now, the LA Audio Show and a LA/OC Audio society meeting, I lost interest anyway.

 

MQA, Pono Player, Dodo Bird, T-Rex... Not necessarily in that order.

 

Here's Stereophile admitting it, based in part on correspondence with Bob Stuart and others. Here is the full quotation and a link to the piece:

Link: MQA Tested Part 2: Into the Fold

The piece, like most pro-MQA pieces, this one immediately deflects the question - but it is plain as day that it is indeed lossy.

More specifically, what it's referring to is that the "folded up" high-res portion is lossy. Personally, I'd also argue that even the CD/redbook portion is partially lossy since it use the 15th or 16th bit (can't recall which) to store the high-res unfolding encoding info, similarly to HDCD.

EDIT: And here's Stuart himself deflecting and dancing around it, implicitly admitting that it's lossy. Scroll down to the "losslessness" section in this linked article:

A Comprehensive Q&A With MQA's Bob Stuart

It's also interesting to note the degree to which the Stereophile piece parrots Stuart's line almost exactly.

The Stereophile piece is Exhibit A for why there's so much skepticism about MQA: You have Stereophile justifying 20-20kHz frequency response and lossy compression, after years (decades?) of being high-res, anti-lossy partisans.

 

Here's an even clearer and more comprehensive explanation (though completely unaffiliated with MQA and Stuart):

Archimago's Musings: MUSINGS: Keeping it simple... MQA is a partially lossy CODEC.

 

If I understand Bob Stuart correctly, he is saying that while a hi-res ADC capture is not losslessly encoded with MQA's 'origami' processing, the whole purpose of the MQA process is to ensure that what is heard as the master mix in the studio is as close as possible to the playback heard by the consumer. In other words, being lossless in the conventional sense of data integrity being preserved is not the be all and end all if the playback is not accurate to what was heard during mixing due to differences in the output of the DAC used for playback in the studio compared with that used by the consumer.

Furthermore, by identifying the ADC used, Stuart asserts that further corrections can be applied to get as close as possible to the original audio signal that was encoded. In the case of modern recordings where audio signals may have been through many conversions (ADC and DAC) by the time the final mix is made, this approach would seem unlikely to be of much help but I can see that it may be possible to correct for some ADC error in simpler scenarios.

If, as Stuart claims, MQA can significantly minimise the influence on the sound from any converters used in recording and playback, then it could mean improved fidelity over traditional lossless audio, despite MQA being a technically lossy process (when applied to hi-res audio). In this way, Stuart is suggesting that while technically not lossless, MQA offers a way to present something closer to the original than is possible with conventionally lossless methods. Hence, I assume, his reluctance to allow people to use the 'lossy' argument against MQA although it is perhaps a mistake for him to call MQA 'lossless' when what he means is it is more accurate than conventional lossless formats.

Whether we accept any of this or not, I think the concept is at least an interesting one.

MQA - 'lossy and lossless' »

 

This is true, but why bring it up? If there is some significant number of music lovers who don’t hear a benefit when listening to music that has been mastered at higher resolution than 16/44, and if an additionally significant number of music lovers listen to music on a wide variety of playback systems, where in all of that does MQA benefit such music lovers?

You’re also implying that the wide variability of playback systems somehow prevents the touted benefits of MQA from being heard. So then what good is MQA?

 

I find it concerning that you accuse unnamed engineers of self dealing without applying a similar skepticism to the motives of those who are in line to profit directly from this technology.

It seems that the major obstacle to an objective evaluation of the benefits of MQA has been the lack of a controlled set of before and after files. If Spence provides a complete set of your files encoded for MQA that will be a brilliant start. Please don't even consider evaluating anything less. Cherry picked data and cherry picked listening environments do nothing to settle these questions.

 

Isn't the problem with that test-which concurs with my own findings that a MQA DAC wasn't used?

 

Given that most streaming music is consumed on $10 earbuds, I suspect the market looks small.