MQA at CES

[Jud]

 

The part I wonder about is their claim to work backwards and fix time based issues with the original recording equipment to get a resulting file that sounds better than the original. Like the 2L recording process discussed here: Carl Nielsen Piano Music (2L-120 remaster MQA) Christian Eggen

 

A quote from that page:

 

Using new technology from MQA we can now correct inaccuracies in the time domain, the so-called pre- and postringing in the impulse response, that were present in the old digital recordings. The result lifts a veil from the sound image and gives us a transparent and intimate listening experience.

 

 

It's the usual apodizing filter stuff that Meridian has been doing for a while and also available in HQPlayer and some other places too. They just do it only up to 96 kHz....

 

 

Makes me smile. "Correct inaccuracies" versus apodizing ("removes the foot," i.e., ringing). Apodizing filters themselves are not completely perfect, in the same sense that no filter is, and particularly not at 24/96. (That is, they are not completely transparent other than removing ringing/reverberant behavior.) So while some inaccuracies are being removed, others are being added.

[Jud]

Jud, are you saying that Meridian's magic algorithms surely have the same problem as all others - they themselves add unwanted sonic artifacts?

 

Inasmuch as one can't have a mathematically perfect filter, Meridian's algorithm must "giveth" as well as "taketh away."

 

Does it "giveth" audibly? Two factors in favor of that are (1) it is designed to audibly "taketh;" and (2) it does its stuff at 24/96, giving it relatively little headroom in which to try to accomplish its purpose without adding anything audible.

[Jud]

Netflix has 65 million subscribers and a profitable business model. The most "successful" music streaming service Spotify lost $200 million in 2014. Likewise Pandora loses $40-50 million per quarter. How's Quobuz doing?? It's ridiculous to compare business models. Bandwidth costs money which Tidal and MQA if they are smart will take into account.

 

Actually it's very instructive to compare business models. 65 million people will pay $8/month for a limited selection of movies. Presumably if you wanted to grow the market for high quality audio you'd have to charge less. That's how Apple quickly became king of the audio world, by offering songs at a dollar apiece. Not rocket science.

 

For lossless streaming to work financially, the music companies will have to accept sufficiently less per song to enable streaming companies to charge something people will happily pay without a second thought, probably half or less what they charge today. How likely is that anytime soon?

[Jud]

Who did misunderstood what, and what is the solution ?

 

My guess is that Auralic interpreted the license they paid for to apply to streaming devices with onboard DACs, and Meridian interpreted it more restrictively.

 

Edit: The solution would be for both to agree on what devices are included at what price.

[Jud]

What makes no sense?? Does processing "perfectly good digital" with HQ player or using an upsampling/DSP DAC somehow have more inherent validity in terms of SQ? or getting good quality music to more people? Not necessarily in my view.

 

Well yeah, it actually does in the case of HQPlayer (and some other software) and some DACs. Not rocket science, the differences in frequency and time domain response and fidelity are readily measurable. To what extent the measured differences make an audible difference is always a question, but the measured differences are not in favor of MQA.

 

All this criticism from people who have done exactly zero hours of critical listening to decoded MQA. While I accept as valid the concerns about undecoded MQA, talking about engineers/producers lIke Alan Silverman and the guys from 2L and Wilson Audio like they're part of a dollar sucking conspiracy, or dupes in Bob Stuart's quest for world domination strikes me as just plain dumb.

 

I agree the conspiracy theories can get outlandish. Meridian just isn't big enough to dominate through market leverage, whatever its intent, and I doubt its intent is anything like malevolent. I figure they want, putting it as simply as possible, to persuade people they can put 24/96 sound in 16/44.1 packages. Maybe they'll succeed, maybe they won't. As you mention, folks seem (understandably) to be less concerned about whether they'll succeed in their main goal than whether undecoded MQA will muck things up for people who don't want to bother getting MQA-certified hardware or software.

[Jud]

Jud the measures differences are for undecoded MQA which while not irrelevant are only central for those we have already decided not to take advantage of software or hardware decoding. Skepticism is one thing, "wishing it would die" before evaluating its benefits is another.

 

Miska has shown not only that undecoded MQA may impact the sound of RedBook, but also that he is able to do what MQA does (specific types of filtering at D/A stage and sometimes A/D stage) better in equal or less file space than MQA does.

[Jud]

To the extent that's true (and I'm skeptical that enough is known) that's swell for those who enjoy dealing with complexities of HQ Player/XXhighend et al and file based audio generally. MQA and its partners' goal of simplifying the delivery of CD+/Hirez and achieving wide distribution thereof through streaming for me stands as something of value nonetheless.

 

Yes, I think that is of value. For those who get their Tidal through Roon, HQPlayer is available through the same UI. So people have choices, which is generally a good thing.

[Jud]

Thank you both,

but if I am right this has nothing to do with the Decoding/Undecoding which follows later.

Is it possible to have this "B spline filtering" in a software player like HQplayer to get the same benefits?

 

Matt

 

You can have that filtering or a better version in software.

 

The encoding/decoding is an attempt, to put it succinctly, to stream higher resolution quality in a 44.1KHz package.

[Jud]

 

My worry is that some portion of music will be distributed exclusively as MQA, leaving those without MQA-enabled hardware listening to the undecoded version, which contrary to the claims by Meridian does not match up to 44.1/16, let alone 96/24 or higher resolutions.

 

I think this might be a greater worry if Meridian had greater market power.

 

A better way to package the MQA data in a format playable by existing equipment would be as a new FLAC subframe type that older decoders would simply skip over. There are other ways additional data could be hidden in a FLAC container as well. Of course that wouldn't give them same iron grip on the deployment they so clearly desire.

 

The "iron grip on deployment" is a normal part of patent law and commercial development of a patent. It's up to any company that doesn't like it to not purchase a license and/or to develop a commercial alternative, and up to any individual who doesn't like it to not purchase hardware, software, or content.

[Jud]

What interest does a manufacturer like Wilson Audio have in whether a market is open or closed by proprietary/DRM format? It is complex, but in the long run they might benefit just a little, and being the type of guys who would sell their firstborn for a SQ tweak....

 

Simply disparaging...something you either don't agree with or don't understand is no way to further yourself or a conversation

 

I see you have no problem disparaging people you've never met in extremely strong and wildly speculative terms. I agree this is no way to further yourself or a conversation.

[Jud]

Jud, it is not this simple (simplistic?). First, Meridian can easily sell MQA to someone with real market "power", or rather the know how to fundamentally change a market, such as Apple.

 

So you're against anything someone with market power may or may not decide to buy in the near, middle or distant future?

 

Second, as I have been saying MQA is not another audio product like any other audio product. In that sense it's creators and supporters are somewhat correct in using the over the top "it's revolutionary" language. It has the ability (because it is a format) to change the way every other audio "product" relates to it. A new speaker or amp simply does not have the ability to do this, and thus one can move on to another product very easily. A format is not like that. Tell me, what other formats (besides the congressionaly approved ones) do you purchase your hard copies of movies in? I know, the analogy is not perfect (MQA is not yet a DRM mechanism in the same way, etc.) but the attempt to "control the road" is there.

 

Right now 2L has more content available and more of a market with its DXD format than Meridian does with MQA. Many DACs will not accept DXD (352.8KHz) input. Shall we be very, very afraid? In some speculative future, after all, it has the ability, because it is a format, to cause the industry to change over to a format your DAC won't accept. And unlike MQA, people are actually purchasing files formatted this way right now (cue theme music from Jaws​).

 

How shall we fight the dread specter of MQA? That's easy, don't buy it. If it isn't successful, or succeeds in only a niche market, you need have no fear that MQA will take over the audio world.

[Jud]

DXD can be converted to a lower rate using freely available software. The details of MQA are secret, so you must use an approved playback system. Well, someone is bound to reverse engineer it eventually, but it's the thought that counts.

 

Where did you get the idea that the "details" of MQA are secret? It's been the subject of white papers, and patent, with its legally required disclosures. I have little doubt there's DRM in the decoding that requires a license ("MQA certification") to use, but the actual manipulations to the music file are anything but secret. Someone like Miska or perhaps yourself might without too much trouble make a better-sounding result from an MQA file than Meridian does; but why would either of you have an interest in that, when you can make something that sounds better yet from a non-MQA-encoded music file?

[Jud]

If it's just another format then I'm all with you.

 

There really is not much difference between FLAC and ALAC, however I am all for open source so I go with FLAC.

 

ALAC has been open source for five years.

 

However, Meridian and MQA want to get into the recording studio and have MQA certification right from the recording/mastering process. If they do get in there, then it's no more a non-MQA world out there. Or in their own words, you will have to listen to non-MQA authenticated music and who knows how bad that is going to sound down the line?

 

The bottom line is money, always is, and always has been.

 

I for one don't believe it is only/or all about SQ.

 

Yes, I can just see Meridian forcing Sony Music (2014 revenue $4.5 billion), Universal Music (2014 revenue of owner Vivendi 10.5 billion euros), EMI Music (2014 revenue a little over 1 billion pounds), and Warner Music Group (2014 revenue slightly over $3 billion) to license MQA and make only MQA-encoded CDs and music files. After all, Meridian's majority shareholder Richemont owns a number of Swiss watch companies, and if you don't do what we like, music company execs, no more Vacheron Constantins for you!

[Jud]

Those resources provide only a high-level, and at times contradictory, picture of the format. To decode it into PCM, you need to know the actual formatting of the bits in the file, and that information is nowhere to be found. If a software decoder becomes available, it is possible to disassemble it and painstakingly recreate a format spec. A firmware file for a DAC might also be used in this way. This is how formats like ALAC have become known to the public.

 

Or you can filter out the noise and work to make the rest of the file sound like something, but as I mentioned above, why would you, when you can produce a better sounding result from a non-MQA-encoded file?

[Jud]

That's true. However, the majority of quality music labels seem quite happy to sell unencrypted high-resolution PCM today. If they switch to MQA only, it will be a serious step back. In fact, the tight DRM controls probably played no small part in the failure of SACD and DVD-A to gain much popularity.

 

It will be interesting to see whether equipment and music companies will get on board MQA for anything other than enabling "hi res" streaming at RedBook bandwidth.

[Jud]

One of the concepts behind MQA is that every ADC (and DAC chip) has specific inaccuracies that you can profile. Hence you can set up a software mechanism in MQA that corrects for these specific innacuracies. So MQA isn't improving on the ADC per se, but it is correcting the resulting innaccuracies.

 

I don't think there is really any argument that the idea behind this is correct. (That all DACs and ADCs have charactaristic "faults" that can be corrected). It remains to be seen if the specific MQA solution really leads to improved sound.

 

Seems to me that these "inaccuracies," if known, could be corrected as part of the firmware for each ADC and DAC independent of the MQA process. So why don't the chip makers know about these "inaccuracies?"

 

Further explanation:

 

No filter, whether in an ADC or DAC, is perfect. It cannot be, even by design, because of the way the mathematics behind filters works: As the filter is adjusted so "time domain" distortion (ringing, which is said to "smear" transients) is minimized, "frequency domain" distortion (aliasing, which causes harmonic and intermodulation distortion) gets worse, and vice versa. One can try to minimize ringing that occurs prior to transients for a more natural sound, but the means of doing this (often a "minimum phase" or "intermediate phase" filter) lead potentially to other problems, such as "dispersion" (where how fast the signal gets through the filter depends on frequency).

 

So in reality, you can't simply counteract all the imperfections in every ADC filter, since your result would inevitably also have imperfections, just in a different balance than the original. What you can do is take filters that in your estimation err too far toward one side of the balance, and change that balance for what you think is the better. In general, what Meridian would do is to take ADC filters that in their (and others' - they're not alone in their thinking) opinion ring too much, and counteract that with "apodizing" filters to remove most of the ringing. "Apodizing" filters start cutting the signal in the uppermost part of the audible range, allowing a gentler filter slope while still doing a good job stopping aliasing. The gentler filter slope helps eliminate ringing. Meridian's filters (all apodizing filters?) also utilize "windowing" functions to cut down on the ringing. I don't know enough about windowing functions to hazard an explanation.

 

So that, in general, is what MQA sets out to do, encoded in a technology that also purports to offer hi res recordings at lower res bandwidth if you have the MQA-licensed decoder.

[Jud]

Apodising is just a fancy word for windowing, and here windowing is used to describe cutting out the part of the waveform you don't want.

 

Meridian's filter is steep and minimum phase. It has to be: in order to remove the pre-ringing of the typical half-band CD-rate AA filter you have to reach a significant attenuation at Fs/2. If you don't want to cut off too much of the audible treble then this dictates that the filter be steep. If you don't want to introduce pre-ringing yourself then this dictates that the filter be minimum phase.

 

All of this assuming that pre-ringing is actually a real-world problem at CD rate and above. IMIO it is not.

 

Yes, Meridian's filter is steeper than some other apodizing filters such as Ayre's for example.

[Jud]

I don't claim to know exactly what they are doing, but I don't think your explanation quite covers it. They claim they are going to have individual "correction" for each ADC and DAC that asks to be "certified", so I think there is a bit more to it than what you seem to think.

 

If they are going to "correct" more than a particular ADC or DAC's variation from Meridian's idea of an ideal filter response, what would that be?

[Jud]

Because it is normalized.

 

No it is not, I didn't pay any attention to X-axis numbers of the plots you posted because it is irrelevant. What is relevant is how many zero-crossings the filter has and what attenuation it reaches.

 

You can run the same filter at 1 Hz or 1 GHz sampling rate, the impulse response looks the same, just X axis numbers change.

 

Time depends on the sampling rate. What is important is how many cycles of ringing you have.

 

mmerrill99, if I can presume to speak for Miska a little here, trusting that he will correct me if I'm wrong:

 

- Miska likes to fix what is wrong with the ADC filter.

 

- For most ADC filters used to make RedBook, what is wrong is that the filter rings, smearing transients. (As Miska notes, the time frame over which the smearing will take place varies with the sample rate; what the filter designer controls is the number of cycles of ringing.)

 

- Thus for most ADC filters, what Miska offers in HQPlayer are apodizing filters that reduce the ringing.

 

- What is wrong with the MQA ADC filter is that it introduces high frequency noise.

 

- Thus for the MQA ADC filter, what Miska offers in HQPlayer are filters that reduce the noise. Because of the need to reduce the noise, these filters will unavoidably have some level of ringing.

 

I would suppose, due to the foregoing, that Miska would say the better way to listen to a given piece of music with HQPlayer would be to start with a non-MQA-encoded file, allowing the use of HQPlayer filters with good time and frequency domain characteristics.

[Jud]

To me all these plots I posted are DAC filters, not ADC filters.

 

I don't really see the relationship between these plots & Miska's filter to remove some noise in a MQA encoded file (this noise being the MQA encoding) - is this noise audible when played with non-MQA DACs?

 

MQA encoding occupies the same place in the chain as the ADC filter - it's "baked in" to the digital file that is to be converted to analog. Part of what MQA does is to encode information in a higher res file as high frequency noise in a lower res (e.g., RedBook) file.

 

This noise is at a level the MQA folks say is non-harmful; that is a point on which some of the people here, including but not limited to Miska, differ.

 

If one does consider this high frequency noise deleterious to the SQ of the file if played with a non-MQA DAC, then the thing to do is remove it to the extent possible. A filter which does this to an undecoded MQA file, because of the level of noise and the fact it is (in our example) a Redbook file, unavoidably results in ringing of the type Miska showed. Thus it would be better, if one must start with a RedBook file, to have one without this amount of noise, since that will permit the filter designer to achieve an acceptable degree of anti-aliasing without making a filter that rings a lot.

[Jud]

Right, but what I don't understand is why anyone would buy an MQA encoded file if they didn't have a DAC capable of decoding MQA (unless there was no alternative version of that file - which I very much doubt).

 

I posted my plots in response to the quoted criticisms of MQA by Tony Lauck which was that MQA was no better than high-res audio. To me, Miska's criticism about the noise degradation of MQA files when played with non-MQA enabled DACs was of no real consequence for the reason above.

 

I was more interested in what psychoacoustic improvements might be realised in MQA encoded/decoded audio files

 

This evening I drove home listening to satellite radio. It's broadcast in mp3 (only) which the satellite radio folks state and/or believe is as good as FM and perhaps CD. Though it's got lower resolution than CD, there was a lot of psychoacoustic research that went into the mp3 standard, so theoretically it should sound just as good, and the lower resolution allows the service to offer more channels per satellite.

 

So let's say Tidal is assured/convinced MQA sounds just as good as, if not better than, CD (this is the Meridian folks with decades of expertise and patents behind them, after all). Yes, technically there are fewer bits of dynamic range, but there's a lot of psychoacoustic research that went into MQA, so theoretically it should sound just as good even if your non-certified DAC (which isn't from a big company and may even incorporate some DIY) doesn't decode MQA. And being able to broadcast a 48/24 stream instead of a 192/24 stream allows Tidal to offer more music to more people over the same "pipes" and thus make more revenue, and if you have one of the MQA-certified "best with Tidal" DACs it sounds just like hi res!

 

See where I'm going here? Nothing nefarious, just normal commercial arrangements, but for the folks who do most of their music listening through streaming services in general, Tidal in particular, and have the stream running into XXHighEnd or Roon/HQPlayer, we could very easily be looking at a problem with noise degradation and no non-MQA Redbook or higher resolution stream as an option.

[Jud]

But why would you send 48/24 stream that has 17-bit worth of data and requires special decoder, while consuming more bandwidth than ordinary FLAC compressed 192/17 stream would?

 

Note that the bandwidth consumed by FLAC is not simply related to sampling rate and bit depth of the uncompressed material, but also depends on compressibility of the source material. MQA encoded material has huge amount of noise that FLAC encoder cannot compress and thus it consumes relatively much more bandwidth. So yes, indeed 48/24 encoded FLAC can be larger and thus consume more bandwidth than even 192/24.

 

Results of my compression tests of the 2L content was that I can consume less bandwidth than MQA using ordinary FLAC while preserving more (all!) relevant information of the source!

 

The answer to that is easy:

 

Tidal has been persuaded people with decoders will get "high res" quality (192/24 or 96/24) sound from a smaller-than-192/24 or -96/24 bitstream, and people without decoders will hear the same quality they've been hearing.

 

They are not interested in or even aware of your 192/17 compressed FLAC stream because you did not come to them with a reputation and a sales pitch (and a way to make more revenue if the commercial idea works out), Meridian did.

[Jud]

I'm guessing for more audiophiles this would represent opportunity rather than problem. If MQA enabled DACs become common in this nightmare scenario of wide industry adoption, most audiophiles might prefer to have Tidal and MQA do the dirty work rather than having to mess with XXHighend or HQPlayer.

 

Of course for most people it will not be a nightmare scenario. There will be wide industry adoption only with wide public adoption to support it. It seems to me in commercial concept to be aiming along the lines of "THX certification" some years ago, which I think was pretty commercially successful.

 

I'm one of those folks who actually likes "messing with" this stuff, so I am presumably not at all in MQA's target market. I think I stand a far better chance of getting great sound from HQPlayer, XXHE, or Audirvana+ and my semi-DIY DAC than with an MQA-encoded stream and an MQA-certified DAC. I doubt I will have to worry a great deal about whether Tidal and MQA take over the streaming market, since as we've already discussed I'm one of those people who investigated Tidal and could not find there the vast majority of what is in my music collection. I'll keep buying CDs from Amazon, SACDs from Japan, downloads from artists or various other sellers, and the folks who like Tidal/MQA will forge on quite happily without me.

[Jud]

It is not rocket science, so it is hard to believe someone running a streaming service wouldn't have competence to run such comparative analysis.

 

 

You're kidding, right? Or is it customary in Finland for people who run businesses to be familiar with what they consider the "technical aspects"? Here the executives don't need to know about such things. They "have people for that," or they hire "the best people, I'm told" as consultants. Like Meridian, for example.

[Jud]

The problem with MQA is threefold:

 

1) Closed. Closed in an environment that evolves towards openness, towards empowering the consumer w.r.t. his music collection. Closed in a reality that supports many different audio system architectures (think open-source streamer SW running on a tiny computer, into a digital amplifier with speaker/room equalisation), many of which simply do not suit the MQA paradigm with its controlled decoding stage and protected decoded signal.

 

Closed because this is a way to monetize technology in an area where (at least it is my impression) there are not stunning new breakthroughs, and thus to the extent it's open everyone else can do it if they want to without paying you. Thus I have the impression it is closed for the same reason Miska does not release his algorithms and Schiit put out their filter in hardware form rather than as a separate piece of software. It makes your particular customization difficult to duplicate exactly and thus enables you to sell it.

 

Since I don't think Meridian/MQA has the market power to keep people from listening to RedBook, hi res FLAC, etc., other than potentially Tidal subscribers specifically (and only potentially, it is still early days), the rest of us will still have the opportunity to seek our own solutions. So I don't begrudge Meridian/MQA the chance to make some money on their technology, any more than I would be bothered if Tidal had decided to purchase technology from Miska, for example. (Well, I would have been happier if they'd employed some of the solutions Miska is suggesting. )

 

 

2) unconvincing arguments. Their papers are shady. They refer to 'recent' auditory research to support their notion of problems and solutions, then bomb the reader with an impenetrable reference list that includes crap like Kunchur's. If there is any basis in demanding lumped impulse responses be < 10 us for monaurally applied sound then we should expect them to point to this evidence, instead of hiding it in a waste dump.

 

Yes, I thought the reference to Kunchur would be a trigger for some people. These papers have two purposes: Market the technology while not exposing too much. Neither purpose is served by the clarity and detail one would expect of an ordinary technical paper.

 

My impression is the paper was discussing mostly applications to 192/24 resolution files, and the balance between impulse response and aliasing would be trickier if we are talking about applying this to 96/24 or particularly RedBook.

 

3) if their assumptions hold, IF, then their solution for packaging 192kHz into 96kHz with preservation of the original impulse width by means of allowing controlled aliasing, can simply be applied, single-endedly, at the mastering stage. There is no technical need for a closed system. Calling it an end-to-end system, guaranteeing to convey what the artist heard is marketing BS. The only way to hear what the artist heard is to sit in the same studio, replay at the same levels, and take the same dope. And even then you'll hear something else, because you are something else.

 

I used to call this audio origami 'very clever, very useless'. Today I'd call it 'very clever, very undesired'.

 

Seems as if what they are fundamentally offering is a different balance than usual between impulse response and aliasing, leaning more toward better impulse response (so worse aliasing), packaged in a way that allows them to control and sell it.