MQA Tested, Part 1

I don't think I've ever seen an audio debate as nasty as the one over Master Quality Authenticated (MQA), the audio-encoding/decoding technology from industry veterans Bob Stuart, formerly of Meridian and now CEO of MQA Ltd., and Peter Craven. Stuart is the company's public face, and that face has been the target of many a mud pie thrown since the technology went public two years ago. Some of MQA's critics are courteous—a few are even well-informed—but the nastiness on-line is unprecedented, in my experience.

It's reasonable to be concerned about MQA. It's a big deal. There's already much support from record labels and DAC manufacturers. It's clear to me that MQA's developers see it as an idealistic venture designed to fix what digital broke, sound-wise and within the music ecosystem. Others see things differently (footnote 1).

My goal for this series of articles, of which this is the first, is to subject MQA to a fair and thorough vetting—not as an expert, but as a science and technical writer. My role is not to make absolute judgments, but to do the hard work, struggling through dense technical articles, pestering people with questions, evaluating evidence in consultation with experts, and assembling what I learn into something coherent and accessible. I'll present the evidence, and people can then decide for themselves.

This is a complex business, with far too much to look into all at once with any sort of rigor. So I'll take on the issues one at a time, beginning here with an aspect of MQA's time-domain behavior: its decoder/renderer's impulse response.

A few months after my first request, Bob Stuart made available an MQA-encoded file containing a train of perfect impulses. Later, he sent me a non-encoded 24-bit/96kHz FLAC file containing exactly the same information, so that I could compare MQA's performance directly with the performance of non-MQA DACs, on even terms.

An impulse is a very short signal—the shortest possible signal, in fact—so it's tempting to think of a test of an audio system's impulse response as a test of its response to very short signals. An impulse-response test is that, but because an impulse contains all the frequencies—for band-limited systems, all the in-band frequencies—it's a useful and commonly used measure of a system's overall fidelity. The more closely an output impulse resembles the input impulse, the truer the output will be for any input (footnote 2).

MQA is, as Bob Stuart likes to say, an end-to-end technology: analog in to analog out. This test, though, starts in the middle: It skips the "analog in" stage by sending a digitally manufactured test signal directly to the DAC. We're skipping half the MQA process—the encoding part, which Stuart says is responsible for some 70% of MQA's claimed improvement in sound quality. The part we're testing—the "renderer" (footnote 3)—contributes about 10% to MQA's performance, Stuart told me.

I measured an assortment of DACs that I have on hand. I recorded their analog outputs at 24/192kHz, so each sample is a little more than five microseconds (5µs) wide. I've expanded the view so that you can see the individual samples—the little magenta dots. The major horizontal divisions are 100µs apart. To make comparisons straightforward, I've used the same vertical scale for all the plots.

John Siau, who designs the DACs made by Benchmark Media Systems, focuses on maintaining the signal's frequency-domain integrity. This is why Benchmark's DAC3 HGC includes a linear-phase reconstruction filter that rolls off very quickly in the frequency domain, and so produces a good bit of time-domain ringing (fig.1). This is typical of a linear-phase filter in that the ringing is symmetric—it comes both before and after the main peak. One of the key notions on which MQA is based is that our ear/brain system regards pre-ringing as unnatural—and there's plenty of it here. And yet, the DAC3 HGC is a brilliant-sounding DAC.

118mqaaustin.MQAfig1.jpg

Fig.1 Benchmark DAC3 HGC, impulse response (one sample at 0dBFS, 96kHz sampling, 100µs/horizontal div.).

Next up is Mytek HiFi's Brooklyn DAC, with its reconstruction filter set to Minimum Phase (fig.2). Again there's lots of ringing, but it comes after the main pulse—no pre-ringing—and the post-ringing would normally be buried beneath the music's reverberation.

118mqaaustin.MQAfig2.jpg

Fig.2 Mytek HiFi Brooklyn, Minimum Phase Filter, impulse response (one sample at 0dBFS, 96kHz sampling, 100µs/horizontal div.).

Fig.3 shows the Brooklyn again, now with its slow-rolloff filter selected. This shows what you can accomplish in the time domain by using a filter that rolls off slowly in the frequency domain. The response is very short, but it's still linear-phase (whether or how much this matters isn't clear), with just a little pre-ringing—about 20µs total. That's not much.

118mqaaustin.MQAfig3.jpg

Fig.3 Mytek HiFi Brooklyn, Slow Roll-Off Filter, impulse response (one sample at 0dBFS, 96kHz sampling, 100µs/horizontal div.).

For the next tests, I sent the same data to the DACs, this time MQA-encoded. First I sent it to a non-MQA DAC, so that you can see what that looks like (fig.4). We're now in the 48kHz domain, not 96kHz, so we expect a wider impulse. This response is mostly linear-phase, though the asymmetry suggests some nonlinearity in the phase response. The details of the response will depend on the DAC's particular filter.

118mqaaustin.MQAfig4.jpg

Fig.4 Benchmark DAC3 HGC, impulse response (one sample at 0dBFS, MQA-encoded, 48kHz sampling, 100µs/horizontal div.).

Fig.5 shows MQA proper via the Mytek Brooklyn DAC with MQA enabled, though the response should be the same for any MQA-enabled DAC. This is nearly ideal: There's no pre-ringing, and the response is fast and short—clear evidence of MQA's time-domain excellence, though the Brooklyn's slow-rolloff, linear-phase response had very similar width and only a small amount of pre-ringing. Would such a small difference be audible?

118mqaaustin.MQAfig5.jpg

Fig.5 Mytek Brooklyn DAC with MQA enabled, impulse response (one sample at 0dBFS, MQA-encoded, 48kHz sampling, unfolded to 96kHz, 100µs/horizontal div.).

Here's a surprise—or it would be surprising, if there hadn't been hints in John Atkinson's measurements over the last couple of years: I've sent the PCM impulse file—not the MQA file—to the Brooklyn DAC with its MQA decoding turned on (fig.6). Same thing, right? Looks like it to me. Apparently, as long as the MQA decoder is enabled, the impulse response is basically the same—even for non-MQA data. Stuart explained to me that, in some implementations of MQA, when MQA decoding is enabled, all data are sent to the DAC's MQA module, which detects the file type and then does the right thing. In DACs that are built this way, including the Brooklyn, even non-MQA music is sent to MQA's upsampling renderer. Don't want MQA messing with your regular PCM data? Turn it off (footnote 4).

118mqaaustin.MQAfig6.jpg

Fig.6 Mytek Brooklyn DAC with MQA enabled, non-MQA impulse response (one sample at 0dBFS, 96kHz sampling, 100µs/horizontal div.).

It's important to consider what fig.6 doesn't show. This is not MQA's claimed deblurring. Deblurring, per MQA, is the removal of time-domain artifacts remaining from previous analog/digital conversions; here there are no artifacts, since this test file was built and delivered in the digital domain. I hope to find a way to demonstrate and test deblurring—how MQA handles imperfect files—for a future article.

One of the challenges levied against MQA by its more knowledgeable critics is that while MQA's approach may improve the shape of the impulse response, its sampling method—and the resulting, presumed increase in aliasing—introduce randomness in precisely when those impulses occur. If they're right, this would offset any claim of time-domain advantage. I synchronized the MQA and non-MQA impulse responses: MQA in the left channel, non-MQA in the right. Over 30 seconds of impulses spaced 0.7ms apart, examined on a microsecond scale, I saw no random offsets—or offsets of any kind—in where MQA's impulses landed.

This is just one small piece of a large puzzle, but it's a start. MQA's filter—the one that in non-MQA DACs is called the reconstruction filter—is apparently very well behaved in the time domain (footnote 5).

Next time: Sure, MQA's compression has a lossy aspect—but how much does that really matter?



Footnote 1: See my "As We See It" in this issue.—John Atkinson

Footnote 2: Strictly speaking, the signal we use to test DAC impulse responses is "illegal," in that it violates the Nyquist/Shannon requirement for the signal to be band-limited to half the sample rate.—John Atkinson

Footnote 3: The core decoder is in the circuit for the MQA test file, but other than routine unpacking, there isn't much for it to do in this case. —Jim Austin

Footnote 4: But see the review of the Aurender A10 elsewhere in this issue, where for non-MQA, regular-PCM files stored on its internal drive, the MQA filter can't be turned off.—John Atkinson

Footnote 5: However, as my measurements have shown, this filter is "leaky" in the frequency domain.—John Atkinson

COMMENTS
JimAustin's picture

Does MQA marketing claim to take us back to the master? Or does it claim to take us back to the studio? The latter claim may be justified, assuming the recording in question has an acoustic origin (i.e. existed as vibrations in air at some point before it became a digital recording). From a technical standpoint, the goal appears to be to deliver the closest thing possible to the analog music BEFORE it was digitized. That's not to say, however, that they haven't ever claimed to get us closer to the master. I really haven't followed the marketing.
Jim

ok's picture

..since it is always and necessarily translated through some certain decoding-amplifying-transducing system, either this being the engineer’s or the reviewer’s or ours or whatever. There is no “faithful to the master” hardware (or software) by definition; there’s only stuff that measures well or bad, sounds natural or weird, feels good or awful. “Faithful to the source” is just certain reviewers' euphemism for expensive hardware that somehow excels in making supposedly bad recordings sound considerably worse.

Staxguy's picture

I just don’t understand how this is supposed to work. And it strikes me as complete charlatanism. Sort of like that advertisement for a casino winning system based on the authors inside knowledge of and study of random number generators, and the purchasers ability thus to take advantage of this.

It’s that unlike advanced digital photography formats, digital audio formats don’t - and I may be mistaken - encode the make and model of the ADC in the digital audio file, which would then be sent to the stream.

Artificial intelligence may be used perhaps on a recording to determine such yet stil I do not get how the MQA authors intend to do a reverse of artifacts of various devices with so little silicon such as in a Merridian Explorer.

It strikes me more of the type of thing the pro audio guys - the live sound guys - would use in the past - I can’t remember Aah the BBE Sonic Maximizer - a bit like hi fis Carver Sonic Holograpgy - to achieve similar.

Anyway, other than a database of albums, tracks, and such, and of course hope, or just listening and golden ears, does anyone know how Merridian identity ADCs to begin their process?

Staxguy's picture

Thanks for this article, Jim.

Since their white paper and what I’ve heard from hi fi salesmen has left me baffled, it’s nice to hear more about MQA.

Although I’ve never purchased their gear, I’ve always admired Meridian UK.

While I thought their compression scheme was novel and brilliant, I thought that their white paper was not fully readable in the intellectual sense. More interest generating coupled with obfuscation. Computer programmers hold such in high regard, however.

Not just chain letter writers, but young computer scientists compete to create unreadable code, which disguises its purpose.

From what I could read, however, I was struck immediately that this was lossy compression, so how could it be geared to audiophiles I wondered.

The focus on timing and complaints on ADC performance and the claim to fix. How could they even hope to do this? Were It a clock signal regeneration, I could understand. They could read the digital file, and mathematically retreated the square wave or whatever pulse which was encoded. Fix the leading edges, for clocking. But it is sound and music data.

Assuming a ADC clock is at some base rate plus or minus some error (say a quartz design, or rubidium...) and a time code is not stamped for every audio value of course, and this error or non accuracy is somewhat random, then how is the MQA to know the value of this error? So as to apply timing correction. Of course, even were this possible, there would be regenerating every sample. Or perhaps not a single sample, given the recording length of a CD, and the actually timing error or drift of the ADC clock over such.

It is not as if we are talking or a mechanical watch rather than a quartz which much be re winded each day, or are we?

Audiophiles are a perfectionist sort, but how are they happy with sample regeration?

Was anyone other than I upset with the sound degregation of Dolby A, B or C on a compact cassette? The noise was terribly poor without it, but was not the sound tone degraded with such intolerably? Dolby S, would that be the ticket? I feel it is such with MQA.

So, let us assume ADC converters then have varying timing errors, and MQA says they can fix. Ok. Not sure how, but ok. Sounds crazy.

How about any other non linearities? Ok, there can be varying ways they encode each sample. A chart could be made of each analogue voltage and the accompanying ADC recorded value. A master chart to say which is sound perfection today to MQA. A master DAC could be used for reference severed to a host of available ADCs with a master source file.

Some ADCs sound better. Some worse. Let’s see.

With the small sizes of audio files, I can see no need for compression for the audiophile market, given that decompression takes CPU cycles which then leads to worse real time behaviour, ala RTOS, or inexcusable pre play latency. That’s me.

People get bored and unhappy with the stuff they’ve got, and look to buy stuff.

I can talk about my own experience. For my PC, I was quite happy with the sound of the on board Intel audio, other than the noise. 24 192 was great. So I bought an Audio Emgine D1. Ok, the noise went away quite a lot, but the sound compared to what was before became totally crappy. Used it regardless, as the noise had bothered me, but stoped listening to music on the PC as a result. Bought a Calynx Coffee and enjoyed it, but so crappy. Bought a FIIO. Liked it, but not as good as the intel. Bought a AudioQuest Dragonfly, and quite nice but not great, at least musical, but not hi fi.

Bought an IFI IDSD DAC, as I thought that the battery power supply might improve the noise. Used it once. OMG it sounds crappy.

Add all the DACS together, and it’s as much as the HP PC, and no better than the Intel Audio. At least I supported the audio industry. Aah, I left off getting an IFI ipower and some crazy cable. The ipower at lest improves the sound of these crappy DACS.

It will likely be some sort of Chord DAC next, likely the Mojo to start. I left out a couple of DACs, but at least they are cheap, an as such encourage consumption. The fact they sound poor is good also, as it encourages continued consumption ala cigarettes or cigars.

Such, I think is the value of MQA.

Anton's picture

Conclusion:

MQA may or may not be good, but should never be used to listen to pervy Elton John music, hippity hop rap stuff that isn't music, and should never be used to listen to Tidal, which is owned by some hippity hop rap jerk.

Also, not on any crap Meridian equipment, and by nobody who rides anything other than a Harley.

We covered all the bases here, including homophobia, racism, and motorcycles as wiener extenders!

For anybody here new to the hobby, I apologize on behalf of our obvious idiot hobbyists who should all click over to this site: http://www.namenda.com/

New folks: Please don't judge the hobby based on our dopes who have to keep grinding old grudges or who think Bibles should be thumped instead of bass notes.

Glotz's picture

Thanks Anton!

spacehound's picture

Anton makes different 'life choices' that I do.

Anton's picture

Who passes your listening litmus test?

Plus, get it right: Anton doesn't judge others as you do.

spacehound's picture

However, as you asked.
It's not just 'listening'.

It's most things we spend our money on. I won't buy battery eggs either. And if we all do that the battery egg producers will go bust and the cruel process will stop.

Get the idea?

My view that it's "cruel" or my views on Elton John and rappers don't have to be 'correct'.

It's called the 'free market' and its just as valid for the individual as it for the corporations who keep saying how wonderful it is, how it is an important part of USA society, and a big part of what 'freedom' means.

So: Though not in the USA I'm exercising my 'freedom'. You seem to have a problem with that.

spacehound's picture

We can never know.

So any liking or disliking of MQA is purely our own personal preference. There is NO 'good or bad'.

My preference is not to have whatever came out of the singer's voices and the instruments, as 'messed around with' as much as MQA does and not have as many 'digital' distortions as MQA produces as a side effect of the MQA process.

Of course anyone else's preferences may be different.

JimAustin's picture

I think it's an interesting question whether MQA more closely resembles the analog mic feed on the recording end, as claimed. Not a complete impossibility, but hard to do and to corroborate.

Brian Lucey's picture

If you are looking at tracking a solo instrument in a great room in mono and you are using top gear with PCM and you want to argue that the MQA process in that 0.0001% of all recorded music ONE EXAMPLE is better, I would suggest it's from the subtle distortions that make digital a little warmer and fuzzier. We CANNOT argue with any integrity that a fully compressed and limited master that was poured over by 12-20 people, after many mixing passes, etc is going to be better with any changes. PCM with a great filter is not fundamentally flawed. Great engineering gets the sound EXACTLY as the artist intended, and the playback rooms and equipment of the world, at various temperatures and pressures of air, are the only right variables. Files should be files, not subjective games for $$$$$ to flow in licensing to people who are broke from Netflix and Hulu screwing up their DVD income.

spacehound's picture

the signal after it has been 'digitised' into PCM, which provided certain simple rules have been obeyed HAS BEEN MATHEMATICALLY PROVEN to be 100% correct to as many decimal places as you want.

So MQA can't possibly "more closely resemble the analog mic feed".

It deliberately discards some data, for a start, in addition to producing some mathematically proven anomalies.

And never forget, 'high fidelity' means "accuracy to the original" by definition. which in this instance you have CHOSEN to be the mic feed.

From it's very definition, 'High Fidelity' is NOT subjective.

Anton's picture

Mostly, it won’t.

MQA adulterates the original digital file. It does it the same way, every time, to every recording. Would anybody claim that doing the same ‘trick’ to an original signal on every recording somehow ‘improves’ on all those originals and also somehow makes them all more similar to what the original microphones ‘heard?’

No.

That would be one amazing piece of magic, akin to claiming that pressing an LP from a digital source ‘always’ improves the sound. All either claim can make is that we may prefer the added artifacts, not that it is more accurate with regard to the original source.

spacehound's picture

It 'adulterates' the 'original' recording as you say. That we don't have a precise definition of what the 'original' recording actually is doesn't matter, it adulterates everything it is used on.

Also we know what these adulterations are, as they have been widely discussed here and elsewhere. And they can be demonstrated by measurements and 'high school' level mathematics.

So 'straight' PCM, even when converted to its lossless equivalents, such as FLAC or ALAC, have to be more accurate, which is what HiFi is all about.

And they are MORE than just 'accurate', they are 'perfect' provided certain simple rules are obeyed. There is no 'missed data', no undigitised analog data missed 'between the samples', etc. Recovery is total. People who don't accept that simply don't understand the sampling theorem.

And that perfectly reasonable (we aren't all mathematicians) lack of understanding by 99.99999 percent of human beings is what the MQA people are mostly, but not entirely, relying on to sell their process.

What is more, MQA is neither original nor new. They are taking a long known and well understood process and trying to make it 'proprietary'.

The reason it has not been used before is that most technical experts think its known and inherent faults outweigh any advantage it may have in 'perceived' sound quality.
The 'perceived' part, which they use quite often, seems to be their 'get out clause' because it's not actually real, it's a set of distortions that some people apparently prefer.

Brian Lucey's picture

Listening to the sample/bit exact file is the ONLY REAL MASTER. Going up or down is not the same thing. Even 44.1 and a great AD converter gets the job done right, well past human hearing, especially with the average age of the audiophile :) The 24 bit files are what we all want, that's the REAL DEAL. It's too simple perhaps for the greedy to accept. Streaming gets faster by the day. MQA's only hope is to sell "improvement". And it's of course a label interest. Like remastering took off in the 90s when Grunge died, they are always looking for corporate monthly and quarterly numbers. Means nothing to what the quality of the product is.

Enjoy your listening, and remember with MQA, when that blue light comes on, 99.9% of the time it was processed in bulk with NO HUMAN CONTACT, and it's NOT the same or better than the source. And it's still a huge file.

seldomheard's picture

And the last three impulse response files posted at the beginning of this article prove that MQA is not faithful to the original signal. Fig 1 is a near perfect rendering of a digitally created impulse test signal - including all of its upper frequency pre and post spike ripple. The MQA representations have discarded this ripple that should be present in the output by raising the noise floor (adding dither noise) to mask it. It may look "cleaner" on a graph to a total newb or novice but it is not as faithful to the original as the figure 1 NON MQA output is. What is far worse is the trailing inverse voltage swing after the main spike in all of the MQA renderings. This indicates latency THAT SHOULDN'T EXIST. Mr. Lucey is therefore quite correct. MQA is actually a degradation of the original 16 bit or 24 bit master. And both MQA and it's representatives in this forum are not being honest about what MQA actually does. The response curves presented, bowever, to the extent that they have not been doctored - tell the real story for those who work with impulse signals on a regular basis. You'd think Stereophile would have chosen someone with a little knowledge and experience to interpret the file data and speak accurately/intelligently about what it says. Sadly in this case, that didn't happen.

John Atkinson's picture
seldomheard wrote:
You'd think Stereophile would have chosen someone with a little knowledge and experience to interpret the file data and speak accurately/intelligently about what it says. Sadly in this case, that didn't happen.

Whatever. But I note on the Computer Audiophile forum that you say you are professionally involved in the audio industry. In which case you need to add your affiliation to your posts and not post anonymously. If you do not wish to do this, then regardless of the merits of what you say, you will be blocked from posting.

John Atkinson
Editor, Stereophile

seldomheard's picture

I am retired. And none of the companies I have worked for in the past would want me to speak on their behalf [flame deleted by John Atkinson]

John Atkinson's picture
seldomheard wrote:
I am retired.

With respect, this is not what you wrote on the Computer Audiophile forum, where you said you were still actively consulting for audio companies. So please follow our rules and include your affiliation in your postings to our website.

John Atkinson
Editor, Stereophile

seldomheard's picture

If I ever post again to this site (don't hold your breath waiting) I will include my full name and address if that makes you happy.

cj

John Atkinson's picture
seldomheard wrote:
If I ever post again to this site (don't hold your breath waiting) I will include my full name and address if that makes you happy.

That's all we ask.

John Atkinson
Editor, Stereophile

ok's picture

Here’s a press release excerpt from dCS/MQA which offers some insight into the complex and variable DAC/MQA interaction, while putting forth the idea that not all MQA decoders are created equal. Royalties aside I don’t really see anything wrong in a rather complicated compression protocol provided one would not have to take the “exactly what was heard in the studio” sales pitch literally:

[...] The MQA encoder takes account of and corrects aspects of the original analogue-to-digital and studio preparation chain. Although a listener can enjoy the encoded stream at CD quality without a decoder, the best result comes with an MQA Decoder, or a combination of MQA Core Decoder and Renderer, which reconstructs exactly what was heard in the studio. The MQA Renderer performs sampling reconstruction under song-by-song instruction from the encoder, while at the same time matching and optimising the attached DAC to deliver an authenticated analogue output. MQA Decoders include a Renderer which is customised for each built-in digital-to-analogue converter. Generally, the converter includes an integrated DAC which is not wholly configurable and may have some performance limitations. For this reason, most MQA decoders include precompensation for the built-in converter. dCS does not use IC converters in its DACs; instead the process of reconstructing analogue from the digital stream is entirely custom, using specific software and discrete hardware to make a DAC. However, unlike other non-integrated DACs, the dCS is still modelled on reconstruction using oversampling, filtering and high-speed conversion [...]
Bob Stuart, Founder & CTO, MQA adds: “[...] This MQA implementation is unique, as it is the first opportunity to enable a DAC which, by providing exact rendering to beyond 16x (768 kHz), matches the desired temporal response with very low modulation noise.”

Brian Lucey's picture

The sales pitch is whole the problem. If they said it's lossy, not lossless. If they said it was not Master Quality, etc. If they said "we are batch processing all the catalogs now" ... the problem is that lying, while usurping the authority of Mastering Engineers is extremely rude, manipulative and I would say bad business.

spacehound's picture

It's interesting that your post in reply to my "Absolutely" one has been removed. Not just edited, removed in its entirety with no record of its removal. The truth about MQA cannot be told, apparently.

(It's quoted in full in the email notification to me of your post however.)

John Atkinson's picture
spacehound wrote:
It's interesting that your post in reply to my "Absolutely" one has been removed. Not just edited, removed in its entirety with no record of its removal.

That is correct. I usually apply a light hand to moderating this site, but posts that are flames, or abusive, or accuse people of lying, or are too much off-topic I delete.

John Atkinson
Editor, Stereophile

tonykaz's picture

With age comes wisdom, I suppose.

Tony in Michigan

ps. I just got 14 Elton John Albums with maybe 2 being "keepers". I'll probably restore him to obscurity.
( keeping: "Love Songs" 1996 and "Yellow Brick Road" 1973 )

I thought that E.John might end up being a Gene Harris level player but......maybe not for me. I wasn't missing much, all these years.

spacehound's picture

Though he's a better piano player than he is a singer.

And my middle name is Elton.
What's more, it is real, not fake like his (it was my fathers first name, I think it comes from Wales), and I was born before him too :):)

Look for John's "Candle in the Wind" (or "Goodbye English Rose") on YouTube and play the live altered version used at Princess Diana's funeral. (Through your system of course.)

The funeral was all terribly 'maudlin' and over the top, but his live playing/singing it in Westminster Abbey is terrific.

tonykaz's picture

I didn't seem to order that album. hmm

Most of the 14 I did get had a sameness quality. This guy might be a marvelous virtuoso in a live setting but I don't find it in his recordings. Maybe I'm outa touch for all this.

Tony in Michigan

spacehound's picture

To me, at least, being a dCS user.
BUT:
It's an MQA press release, NOT a dCS one - they just quote it on their site. (Though this wholly MQA press release does contain a short quote from dCS within it - just the usual type of 'puff' about trust, co-operation, and mutual backscratching.)

Unfortunately the part I quote below is BS:

"The MQA Renderer performs sampling reconstruction under song-by-song instruction from the encoder, while at the same time matching and optimising the attached DAC to deliver an authenticated analogue output."

The MQA 'authentication' process is mostly a 'bulk' process (as many here and elsewhere have already pointed out) applied en masse to the record label's supposedly entire catalog, much of which will be quite old, far older than MQA. So it is likely that no individual 'signing off' or 'analysis' of each 'song' will have actually been performed. And they will anyway have no clearer idea of what each 'song' was meant to sound like than we do.

Bear in mind that MQA's original claim that they (the MQA company) would tailor each DAC using a sample the manufacturer provided never happened and still isn't happening.

Sure, unlike most (all other?) DAC manufactures dCS can 'tailor' its filters rather than using the standard MQA pre-provided set where the DAC manufacturer picks the ones he likes.

But they AREN'T going to 'adjust' the filter for every 'song' ever performed, are they? They don't need to, as MQA are not doing what they first claimed.

Why do dCS support MQA?

It's another 'label on the front'. Might sell some stuff to MQA enthusiasts.

And at dCS's prices they can't NOT have what DACs between a fiftieth and a hundredth of the price are providing, can they? And to do that they have to pretend they think it's a good idea.

tonykaz's picture

It's coming to the place where Mr. Spacehound needs to shed his ( presumably his ) disguise and establish himself as an Authority in these matters.

Anonymity is only important for "Hit & Run" escape strategies.

Tony in Michigan

spacehound's picture

Following the (short) history of MQA, what it initially claimed, and what actually happens is all that's needed for the first part.

As for the rest, it's how business is often run. You should know that better than I do as I was only ever a very small cog in a very big wheel, though it became less successful after I retired :)

And I'm a dCS fanboy :):)

Archimago's picture

Hey there Spacehound,

Just a few words about the "MQA Renderer performs sampling reconstruction under song-by-song instruction from the encoder" bit.

There is some truth to that when referring to the "renderer" portion of the MQA system. I think it's just talking about the fact that the encoder system can tell the playback renderer whether to dither, add noise shaping, and which of 16 mangled low-tap length FIR filters to apply for upsampling. This is basically instructions embedded in the MQA file just like how HDCD could embed settings like "peak extension" during playback...

This can all be determined in some way by the encoder algorithm when the source files are batched through (did someone say "AI"!? Sure, MP3 lossy compression must also be AI as well :-).

Hard to imagine record labels would spend money to hire artists/engineers to specially "sign off" the final output of thousands of albums. I feel bad for anyone hired to do this job as it must be horrifically mind numbing and a questionable use of resources...

spacehound's picture

I will have to take your word for that. [off-topic content deleted by John Atkinson]

AI? Not a chance. After 30 plus years in the mainframe computer business I will state categorically that no such thing exists. Even 'self-learning" systems aren't AI. It's just a fancy name for some 'expert systems'.

I will only accept that AI exists when an 'intelligent' artificial doctor saves itself when the firebell rings, buys itself a Ferrari, or gets bored with doctoring and decides to go fishing for a couple of days :):)

And as you say it's hard to imagine record labels hiring artists/engineers to 'sign stuff off'. Very hard to imagine. Impossible to imagine, perhaps.

ok's picture

..it’s thoughtful implementation that counts the most – not “the” format itself. Any decent audio format (digital and analog alike, even some "lossy" ones) can sound excellent when properly mastered and (re)produced; it can also sound terrible, regardless of qualifications and $$. I see no real point in format comparison, since the mastering and (re)production process of one and the same track is never one and the same for different formats, even if done by the same person and gear: different functions and routes inside the same machine are actually different machines and one could reach at very different results under slightly different conditions. Routine measurements also measure certain hardware/software implementations, not “the” ideal format itself. The whole “mine is bigger” format war is mostly a way of cash extraction and occasionally fun (for me at least..), but totally outlandish in context of the current state of affairs as far as audio quality proper is concerned.

spacehound's picture

....uses 'lossy' formats for recording/manipulating in the studio or anywhere else. They use PCM or DSD.

Both of which, provide certain simple rules are adhered to, record 'perfectly'. There are no 'gaps between the samples' and there is nothing going on within the samples that is not recorded.

Reproduction is a different matter. Any lossy compression technique will reduce the quality. To me 320 AAC is as low as I will personally accept as I find its losses 'only just audible' so it's ok for 'in car' use. And bandwidth/disc space is increasing so rapidly that I suspect that soon that won't even be needed.

But MQA is merely another lossy codec. All it does different is degrade the source as well, which is fine as long as the 'original', whatever that may be, is still available to the public. Which by the way MQA appears to be going (almost on its last legs), it likely will be.

ok's picture

..commercial downgrades to 44.1 or 1-bit of the pristine PCM or DSD master respectively do always justice to the "original" (from a mathematical and also acoustical point of view), nor that the various PCM to DSD and vice versa conversions often occurring during the mastering/(re)production process are mutually interchangeable and fully reversible (let’s not even mention the analog chain in case of legacy tape recordings). The main problem with MQA, when compared to the aforementioned and various other digital manipulations/downgrades, is not so much it ain't theoretically and practically lossless; the problem is that MQA appears to be utterly useless.

spacehound's picture

With a couple of minor exceptions.

1) I don't think 'downgrading' of the bitrate will be audible provided it's by an integer and kept within reason, like 192 to 96 or 96 to 48.

2) As a 'mathematical' process 44.1 should genuinely result in the originally claimed 'perfect sound' as whatever these 'cloud cuckoo land' people claim about their 'golden ears' it is nonsense. No human on the planet has ever been demonstrated as hearing anything above 20KHz. Period.
And as 'time domain' and also 'rise time' stuff are both direct mathematical functions of the human frequency response it's the same for those too.
However, if I had been 'in charge' I would have chosen 48 rather than 44.1 as it gives more room for the filter so you can have a gentler slope.

As for MQA it is pointless for us end users. It degrades everything it touches in terms of 'accuracy' and its standard FLAC compression results in a bigger file as FLAC doesn't compress it as efficiently as it compresses regular 'music derived' PCM.

ok's picture

..I believe the main reason some people claim they can hear the super tweeter effect in certain loudspeaker designs, is not some kind of unconscious visceral process, but rather the fact that super tweeters mostly act like high-cut filters which allow the main tweeter to effortlessly operate within its linearity frequency range. As for presumably audible differences between “high resolution” and red book files, I think they all derive from the fact that the latter are either "downgraded" versions of the former or occasionally non-comparable alternate masters. By the way one reason for the claimed superiority of digitaly mastered vinyl pressings over their red book or sacd equivalents could be that the former allow for direct use of the original master.

spacehound's picture

I've just put a steep high pass filter with my first try at 10KHz and I gradually increased it in 1KHz steps to 18KHz, which is easy on JRiver, and played some 192 sample tracks from dCS, no less, and also some pop stuff at 96.
Just in case I've got cloth ears and couldn't hear anything I also had JRiver display the result as a 'live' chart.

There is no output WHATSOEVER above the very low PC/JRiver noise level beyond 16KHz.

Which makes all this 'Hi Res' stuff, including the entire content of an MQA 'fold', rather pointless.

Even if MQA had a 'theoretical' point originally the 'practical reality' has destroyed it:

The data to be 'folded' is encoded in a lossy process. The MQA people have said so..
The 'fold' creates aliasing that wasn't in the original.
The final transmitted (or stored) data file is in FLAC. But due to the previous MQA processes this file is in fact larger than a FLAC encoding of the source file would be.

There are other disadvantages to MQA but I will stay with the above as they are easily measurable.
The result, obviously, is a departure from the content of the source file so it is 'inferior' by definition.

It COULD have worked. But there are two difficulties:
The originally claimed detailed analysis of the studio equipment, whether it be 'old' or 'current' equipment is not taking place in most instances. So the claimed 'compensation to correct its faults is not possible.
The process is implemented very clumsily, with these losses, added aliasing, and so on resulting in an inaccurate output file that cannot be efficiently compressed.

And unfortunately for MQA, due to it's long gestation period there is no longer any 'need' for it (if there ever was) as it has been overtaken by the ever increasing efficiency of digital communication networks and the constantly reducing costs of 'disk' space.

tonykaz's picture

Well, damit, why didn't you just say this from the very beginning.

I've felt that Redbook is adequate for 'my' musical playback needs.

However, the Movie Industry ( who spend serious money creating soundtracks ) use 24/96, don't they?

The "Live" musical Acts such as those A National Tours use 24/96. The Music Recording Industry use even higher 'rates' which the Mastering Engineers then forge into the entire range of files, from 16/44.1 ( my favorite ) to everything else.

Advertisers, like my GM, have high-res sound tracks that have to be included in the Adverts being sent to iPhones; Bombarding end-users with "important Marketing information" about 'our' latest offerings.

I suspect, not that I agree, that MQA can take the high res master and deliver it to someones's smart phone, which is the one hell of a marketing claim, doncha think ?

Simply said, MQA will convert everyone's Audio ( sound ) tracks to a 'Standard' 16/44.1 so that all media types will function properly, thereby creating a Standard.

I read audiophile complaints relating to the MQA Standard as not being a "Gold" level Standard.

Sooooo, MQA is not for us 'hair shirt, lunatic fringe" Audiophiles, MQA is for Industrial purposes.

Why is anyone arguing about this?, audiophiles aren't required to participate in any part of MQA, are we?

Tony in Michigan

ps. MQA might just result in all of us standardizing on PCM. ( which I consider a good thing )

spacehound's picture

What I intended to reply is immediately below this one.

spacehound's picture

Is more than adequate for every human being, though a tiny minority within another tiny minority won't admit it.

Though I'm now 'semi' retired we didn't do much advertising as we ran the world already and still do, having almost 100% of the 'world running' market, though most don't know we run it at all :):)

MQA doesn't deliver the master, it screws it up considerably first.

And as I said, there is so little audio information above 20KHz (zero above 16 KHz in the admittedly fairly small number of examples I tested) that it's not worth having as you won't hear it even if it does exist.

Also as I said, transmitting 'straight' unmolested FLAC uses less bandwidth than the distorted and lossy MQA stuff.

So what's the point of using MQA's distorted, lossy, and 'proprietary' process?

My 'guesstimate' is that MQA will have vanished with a year, it's vanished from most UK audio magazines already, including the biggest and by a factor of about ten the most popular one that does AV, phones, soundbars, TVs, etc. as well as our stuff.

tonykaz's picture

Well, ok, again.

I don't see the point of going "Lewis Black" over all this.

I'm delighted that we still have 16/44.1 after all these years, it's the most successful music format ever, isn't it ? and we're still using it as 'the' basis of our personal music systems.

I'm kinda moving onto finding some 'good' sounding stuff for my personal use. My previous all time favorites were ProAc Tabletts, Magnapan MG2 and Electrocompaniet.

Now, I'm preparing to purchase ( for extended audition ) Loudspeakers and Electronics for my personal uses. I'll repeat the methods I used in selecting product lines for my Esoteric Audio Salon ( back in 1983 ). I'll start with Meridian, I love the Brit. stuff.

Tony in Michigan

spacehound's picture

But Meridian is NOT highly regarded in its home country, the UK. It has so few potential buyers that hardly any dealer stocks it or will even obtain it on a 'sale or return' basis for you to listen to. In fact Meridian is somewhat of a joke in the UK HiFi scene, they go with Givenchy, LMVH, Piaget, Maserati, and all that overpriced poseur's junk. And their continued existence is by no mean secure (assuming repairs and the continuing existence of a US agent matter to you).

I like Naim, but would not buy one of their power amps lower in price than the NAP250DR which would likely be about 5,000 dollars in the US.
You don't need to buy their recommended 'matching' preamp, a lower cost Naim one will be fine. But remember that almost all the Naim preamps require a non-optional Naim power supply at extra cost.

As a 'left field' choice, Sugden. They are very long established and concenrtrate on Class A power amps. Their 'Masterclass' power amps are very similar to the original Krells in concept but preceded them, and have advanced in the many years since.
They at GENUINE 100% Class A all the time, unlike the Pass amps which are only class A up to certain level and then become class A/B.

Avoid Musical Fidelity. They keep producing a new claimed 'worlds best' amp, change their minds a year later and make something totally different and instantly discontinue last years 'miracle' amp.

Quad are Chinese, as are Audiolab. That doesn't make them bad but it makes them non-British.

Mission and Cyrus are nothing special.

Monitor Audio speakers are mostly good but don't buy their expensive 'Platinum' ones. They are tilted up at the treble end to impress in a showroom quick listen.

Tannoy speakers are good but only buy the ones with ONE speaker, a dual concentric one. The ones with an additional same size woofer but without the 'tweeter' dual concentric part just sound blurred.

Source? Any laptop or 'tower' Windows PC will do. Don't believe all this guff about 'specialist' PCs or expensive streamers made by HiFi companies sounding better. And JRiver as a player is as good as anything else. With a bit of ingenuity taking all of five minutes. you can also make internet radio stations look like the tracks on an album, which makes the whole lot of them look like just another album in the 'Album view', which is very neat. (I used a picture from the internet of a Zenith Trans Oceanic as the album cover art :))

What do I use?
A Lenovo 'tower' PC, JRiver 22, a dCS Rossini DAC/streamer (because I'm a fanboy and anyhow I won't buy a DAC costing more than 500 dollars if it has an 'off the shelf' DAC chip in it), A Naim NAP250DR power amp, and Tannoy Kensington speakers.
I can find no fault in any of it so don't intend to buy anything else for a very long time.

tonykaz's picture

I just had a 'Closer Look' at the UK Dealer networks. Egads, the UK still has Dealers, hundreds of them.

I was one of those Dealers, back in the 1980s, for gods sake. All or most of all 'our' Audio Dealers are looooooooooooonnnnnnnnggggg gone!, replaced by a small few Home Theater Shops ( that sell the hell outa Meridian, by the way ).

Anyway, we have no Sugden, except for Tone Imports who barely have a functioning web page and are the US Importers of the wonderfully important Shindo Labs stuff.

I'd like to let you understand that pre-owned LMVH type gear sell at yard sales & flea markets for attractive prices, Becaaaaaaaaause absolutely nobody in the public has ever heard of the stuff. The inheriting grandchildren give the stuff away for next to nothing. Go figure, they don't want grandpa's stuff cause they get their music from their iPhones, hellllll oooooh.

So, we can buy a Threshold Amp for $2-3K., we can buy used Krell or just about anything else because there are very few of us audiophiles left to sell to. There's gotta be about 1000 available amps for every interested buyer.

So, I ask you.

Is there something new & improved in amplification ?

Is there something new & improved in loudspeakers design ?

I had a hell of a music system in my high-end room at my Esoteric Audio, I'd be delighted to have something like it now, even if it's pre-owned.

Tony in Michigan

ps. I'm feeling you to be one of those Fpga/R2R DAC people, who express a deep seething hate for all things DAC chip. I kind of agree in that I accept the importance of getting it "right", right from the start: "garbage in / garbage out. Except I seem to like the DAC chip DACs and realized the difference between a solid state Class A Amp vs a Tubed Amp is greater than any difference between any DAC types. A good 6SN7 will blow away any DAC known to mankind.

spacehound's picture

It seems to be because I replied direct from the email 'notification' link.

spacehound's picture

...their dealer list.
Like many others, they tend not to remove dealers from the list when they drop out.
The closest genuine HiFi dealer to me who has actually got any Meridian stock is 50 miles from me (and it's a small 'dense' country). I've got five 'brick and mortar' dealers within 25 miles and there are two 'mid Fi'/TV chains in addition, and a large chain department store that sells Naim and Bowers & Wilkins.

And all that '50 miles away' Meridian dealar has got is two different sorts of AV centre speakers, each in black or white. Four AV centre speakers in total. Nothing else.
And they are 12,000 dollars each. Not many are going to buy them.

As for their other dealers I looked at they are merely AV 'installers' with no stock at all. And some of them don't have operational websites.

As a result Meridian is mostly confined to their flashy 'Boutique' in central London.

Your other comments are FAR more interesting :):)

If the 'common herd' have never heard of your stuff that's good. You should not buy a Rolex watch for example, you want stuff that only the 'cognoscenti' know about. You don't want people such as used car dealers and the 'oaf' who owns the local Starbucks franchise knowing your watch was expensive, do you?

As for tube and class A amps it should not matter. HiFi means 'accuracy' by definition so the higher you go up the scale the more everything should sound the same.
Also the US fondness for 'high end' is silly. It's mostly not HiFi, it's a 'body double' for it, sounding falsely 'impressive' but it's not the real thing, just overblown bling. As someone said elsewhere, the US, particularly, moved far away from 'accuracy' since about 1980.

'Chip' DACS? It's not that they don't sound any good, many of them are fine.
But I am not going to pay several thousand dollars for a 'chip' DAC where the important stuff is done by someone else's 10 dollar chip and you just read the data sheet and merely add a simple analog amp, the recommended power supply, and a fancy case for several thousand dollars extra. And sometimes a visible tube or two so it looks 'specialist'.

dCS use neither ladder DACS nor FPGA ones. They do use an FPGA for the filters, but the DAC itself is discrete hardware taken from their long military, space, BBC, and studio experience. And that hardware is 'controlled' by a small processor running dCS's own software.
Both of those things mean they can implement MQA very easily whether they 'approve' of it or not and have recently done so, presumably as it adds another 'function'.

tonykaz's picture

Oh Hell, I'm not a believer...., never have been.

But I do like Meridian stuff. It sounds good, as far as I'm concerned. I'll consider good sounding to be a good trade-off for accuracy. Besides, Hi-end stuff doesn't strive for any sort of accuracy, does it?, if it did it would sound like ATC stuff which is Pro level Accuracy ( with a screw 'good-sounding' attitude, isn't it ? ).

Oh Well.

Y'all's Rossini is a bit too pricy for casual ownership experimentation .

Stereophile's JVS seems to own dCs. The Analog Planet guy makes exaggerated claims about dCs ( and exaggerated claims about everything, for that matter, I'm afraid ). On the whole, dCs is liked by everyone that gets access to one of their $100,000 systems , that's One Hundred Thousand Dolllllllllllllars ( plus ) ! For christs sake ! Geez, talk about LMVH.

However,

You seem to have a rather nice tidy music system, about $50,000 dollars worth of stuff, with the Lion's share of the funds in dCs's pockets. I like your approach enough to claim it as my own: my systems is iMac based. I intend to find an equally impressive performing set of equipments.

Just now, coming out of a 30 year hiatus from the audio world, with the exception of some Schitt ( shit ) headphone amps & Sennhieser headphones.

I encountered the MSB guys and nearly bought in to their DAC stuff but held off because I couldn't understand DACs.

Now I realize that DAC chipsets are immensely capable and not to be chided or belittled, they are the product of decades of man hours worth of work.

Besides, I've done numerous "hands on" tests and evaluations that finally revealed the DAC chipset itself is wonderful but the implementation ( that you mention ) provides the summary performance yield. I can accept.

Overall, I'm say'n that my approach to music reproduction is nearly identical to your's. We seem to both be stuck on the same dam page: MQA.

Funny how this sort of thing can happen.

Tony in Michigan

ps. that Tannoy stuff looks lovely, I've never gotten within 1,000 miles of one of their transducer systems, have I ? I'll keep an eye out for it.

spacehound's picture

Your comment re ATC is interesting. If I purchased speakers other than the Tannoys it would be ATC. (As I said before, I haven't got the courage to buy the Quad Electrostatics.)

Simply because my 'baseline' for it all is accuracy, I see no point in spending considerable amounts of money for a result that isn't accurate.

If I don't like what I hear I buy a different recording, not change the equipment to make the recordings I have sound 'nice' or even worse, 'impressive'. The 'impressive' part is why I very much dislike this "High End" attitude of many people. It's the price they are mostly impressed by, so they convince themselves that at those prices it has to sound 'good' whatever that is.
"If you like your Ongaku you can keep your Ongaku". And your nonsensical Wilson or Magico speakers, not that an Ongaku would drive them very well :):):)

'HiFi' or 'Audiophiledom' itself:

It's nothing to do with music, everyone listens to that at some time or another.
It's just a geeky hobby, like model planes or model trains.

And as it is entirely passive it doesn't require any skill or ability whatsoever, unlike most others. All it needs is money.

And on that basis I am sometimes tempted to sell the whole lot and buy a Yamaha 'music system', thus 'cocking a snook' at the entire HiFi industry. At least Yamaha knows how most musical instruments actually sound :):)

tonykaz's picture

Lucky you,

Bob Katz is reporting on amplification.

I'm talking right now, today.

Tony in Michigan

spacehound's picture

just a page filler, A few pictures of naked ladies would be nicer.
Only you and me will be crazy or bored enough to read it to the end. I did, and most of it is gibberish, as in "Distortion is compression." or "An amplifier with very low distortion will likely sound 'dry'."

And he's a headphone guy.
When the majority of bands, groups, orchestras, singers, etc. develop the magical ability to stay in front of me no matter how far and how fast I turn my head will I buy some headphones.
Until that happens I won't.

tonykaz's picture

and anyone else interested in CD playback.

Paul McGowan's YouTube Channel has a insightful description of most of what we are here discussing.

Please see : Paul McGowan ,
Do CDs sound different than rips ?

Tony in Michigan

Ktracho's picture

For what it's worth, I just listened to Diana Krall's album "Turn Up the Quiet" over Tidal on my Audioquest Nighthawk headphones connected to a Meridian Explorer 2 DAC. Over the last two days, I listened to a couple other of her older albums, which I noticed did not have MQA encoding. With the aforementioned album, I just assumed it wouldn't have MQA encoding either, but after a few songs, I thought to myself, the sound quality seems noticeably better on this album. It didn't seem as irritating as the sound of the other albums. Then I took a quick glance, and saw Tidal flagged it as an MQA encoded album. In the other two or so MQA encoded albums I've heard over Tidal, I noticed the same kind of sound quality, and it reminded me of how my old Stax headphones sounded when driven by my tube amp, which I unfortunately damaged due to naivete. Granted, that's been several years ago, but I can't help wondering if there's something to the timing issue. According to Lavardin's website, transistors also have an issue with timing as well (tubes less so), though I've never heard their amps. Anyway, this is just one person's opinion, and a non-expert at that, but it's probably as close as I'll get to a blind test - I was expecting to not hear any difference in sound quality, but I did, even though I wasn't focusing on the sound quality - I was just at work listening to music while trying to finish my tasks.

Geoff1954's picture

I'm not qualified to discuss most of the issues raised in this article or the posted comments. Moreover although I am a fairly literate person, I am not educated enough on the issues involved to follow or understand most of the discussion. I would only like to share one recent experience.

I listen to Tidal and have enjoyed many of the MQA titles that are available on my desktop Mac application, running through my Geek Pulse DAC into my Marantz/Paradigm home theatre equipment. I am only getting the single unfold, the on provided by Tidal software.

Recently I listened to some of those titles that I had previously enjoyed in MQA and was disappointed. After doing some checking I learned that when Tidal offers an update -- and I had just accepted one -- it changes the preferences set for streaming in individual accounts. Or it did in mine. Master quality was no longer selected in my preferences. I had to reset it. As soon as I did that the MQA albums sounded glorious again.

Every album does not benefit equally in my experience, but many do. That's good enough for me.

deckeda's picture

OK, so MQA appears to supply a good reconstruction filter to DACs that may or may not otherwise have one. Said another way, some DACs remain better than others, because all DACs are both hardware and software.

Riddle me this: If MQA knows about ADC flaws, why not get into the ADC business? Sell or license matching MQA DACs to audiophiles to complete the chain. A more troubling question: If all ADCs are flawed then why are we even here ... how can you offer a fix to something that inherently is unfixable?

Quote:

"Next time: Sure, MQA's compression has a lossy aspect—but how much does that really matter?"

For a streaming experience (Tidal HiFi?) that reportedly delivers better-than-CD sound quality, it doesn't matter that it's lossy to the end user. Better sound is just that: sounds better than a CD. You can't stream a "legit" hi res version from a service, so stop comparing MQA to what doesn't exist. Full stop.

Here's where MQA won't matter: physical media, where it's mysteriously irrelevant. It SHOULD be relevant to CDs, because of the encode/decode aspect of correcting ADC/DAC errors ... or in the case of LPs cut from digital files (which are the vast majority) it SHOULD be relevant, because "they all use" flawed ADCs, right? Why don't I hear anyone talking about this? Should I play my new LPs in real time through something like Pure Vinyl just so I can hear it through an MQA-enabled DAC? I'm not being pedantic, it's a serious question!

We live in an age where Netflix HD is becoming the norm. I realize many rural areas remain bandwidth challenged, but a lossless audio file (even a hi res one) pales in size to what it takes to stream a movie or TV show. So I don't understand why MQA's file reduction/compression is even a consideration. For whom? The Spotify subscribers listening to 320kbps files? Or the tiny handful of Tidal HiFi subscribers?

************
What I LIKE about MQA is that it presumably takes whatever lossless file the streaming service gets (again, presumably something higher than 16/44) and delivers it in somewhat smaller form, in a consistent "format" (MQA.) I'm good with that aspect. Compared to CD, LP, hi res sales ... streaming appears to be our commercial future, and we'll need some standardization for it to truly bloom.

What I do NOT like about MQA is the claim about fixing ADC errors. Sounds like bullshit, for the reasons stated.

What I do NOT like about MQA is the complete silence regarding its use for CDs and LPs, areas MQA could and should touch on, for the SAME sound quality reasons it's so strongly advocated for streaming.

David Harper's picture

We don't need MQA. We don't need hi-res. All we need is for record companies to master and record undistorted uncompressed CD. Playback technology cannot improve shit mastering. Audiophiles have failed to identify the real villain. Spend all you want on overpriced high-end components. Chase all the snake oil playback technology you want. This is what keeps high-end audio alive.

spacehound's picture

It's a 'hobby' and mostly just one of the several ways we all have to use up our disposable income.

And you are right about CDs. When they first appeared the producers were VERY careful about mastering etc. as they wanted this new music medium to 'take off', which it did.

Now, as both CDs and their computerized alternatives are commonplace nobody bothers, except a few 'specialist' labels, the output of which most people never buy as such outfits don't have the money to use respected artistes. So we get excellent recordings of rubbish performers.

As for 'high end' audio it now has little or nothing to do with 'high fidelity' so the recording quality doesn't really matter.

It's just 'image' stuff and the market is getting smaller and smaller as the prices are constantly increasing while all other 'domestic electronics' are decreasing in price.

dce22's picture

Whenever there is a discussion about PCM filters always Hansen and Craven names come up with there theory of dissinformation.

Charles Hansen: "When a linear-phase filter is used, there is no phase shift at all, but then half of the ringing occurs before the impulse, and the other half occurs after the impulse. This never happens in the real world."

dce22: Yes it does, when you produce multiple tones with the same start time and level the higher frequency tone will peak early then the low will build up slowly, that is the reason the small ripple is there when you produce the whole spectrum dc-22khz at the same starting point 20khz will ramp almost instant as the slower and slower frequency's buildup and combined with each other, because of the same start time 15-20khz will be in full swing and mixing with eachother while the 10hz has moved only a fraction and is not noticeable by the time that all sine waves meet up at full wing you got the full lobe peak and as the low frequency sine tail down you got high freqency chattering at full speed and at the end the high freqency going at full rate as low sine waves are near zero so you get small ripples again and if you repeat the pulse again and again you get band limited white noise aka the whole spectrum dc-22khz with all sine waves in time coherent state (phase linear).

Taking this function (sinc) mixed with other sinc in 44.1khz time rate you will get DC strait line and if you reverse every other sinc you will get pure 22.05khz sine wave with no harmonic or time distortion,

Why all this ripples (ringing) dont messup the 22.05khz sine waves you may ask,
Because the riple are orthogonal to eachother in time so as one ripple goes from one sample the other sample support the ripple from reverse phase and can create any signal from dc-22khz without "preringing" and distorion in time.

Any i mean Any deviation from the original sinc function impulse,
is bad for the sound quality, no matter what these nice guys like Bob, Peter and the late Charles say.

Saying that preringing is bad because in real life sound echo happens after the main sound event,

Is like saying Planes fly because Frogs Jump.

Clear explanation and visual depiction on Page 7
http://lavryengineering.com/pdfs/lavry-sampling-theory.pdf

For layman easy type of understanding sampling
Read Dan Lavry Document.

spladski's picture

I have read the technical and non-technical discussions about MQA and wish to articulate a view that seems under-represented.

Firstly MQA is a business case that purportedly claims to improve audio quality. It does so by repackaging methodologies that are not entirely of its own invention. So called apodizing filters for controlling impulse responses originated elsewhere. These filters can be implemented anywhere in the digital audio chain without MQA. Attempts to quantify the recording process is a waste of time because it will inevitably lead to variable results. An extreme case is: what are you going to do with computer generated music? Therefore if this aspect is removed from the MQA business case, the 'Master Quality Authentication' concept evaporates.

Self published material will increase and the recording processes will become better. Apodizing filters are finding their way into chips. Therefore MQA is retrospective and not forward looking.

Audio designers, myself included, normally jump at the chance of technology that improves the audio cause. However a number of us are lining up against MQA, not because it is technically flawed, but because it is placing controls where there should be none, to further its business case with increased costs ending on the consumers lap.

There are those that argue that if it does improve quality, why shouldn't we use it? I will answer that. When the CD standard was established there was provision for pre-emphasis. This was to improve the signal to noise ratio of the recording and playback chain. It works. It does what it says on the tin. Therefore the logical conclusion, following the MQA argument, is that all masters should be encoded with pre-emphasis. I would be interested to know if John Atkinson has bought into this train of logic.

We have endured a plethora of attempts to mess with audio: Dolby A, B, C, SR, AC3, DBX, NICAM, MP3, AAC, CD pre-emphasis, cassette etc. Shouldn't we have learned something along the way, now that technology has given us the opportunity to have less fudges? When the media industry acted in a protectionist manner with DVD and CD, they failed to acknowledge the future. MQA is repeating the past. Therefore I would ask that Jim Austin and John Atkinson consider the business case and set aside the technology, even if it is an entertaining exercise. I believe they would better serve Stereophile's 71,000 readers by answering the question - Do we need MQA? Rather than, Does MQA work?

rt66indierock's picture

This post is definitely a contender. John is trying hard to stay away from the business case.

dce22's picture

Apodizing filters came up from photo-video dsp's where brickwall filters are not process efficient and unusable,
so to remove alias byproducts from processing they tradeoff resolution (a bit blurry-fuzzy picture) but it losses the artifacts (halo's around objects).
Example:

https://upload.wikimedia.org/wikipedia/commons/4/4b/%28a%29_measured_beam_profile%2C_%28b%29_modeled_beam_profile%2C_%28c%29_measured_beam_from_above%2C_%28d%29_modeled_beam_from_above._%281_pixel%3D0.05_mm%29..jpg

a , c --- Apodizing filter
b , d --- Convetional filter

The b ,d video filter in audio world is Non oversampling DAC.
The audio FIR Phase linear filter (the one with Pre-post "ringing") equivalent in photo-video processing will render perfect picture like b , d example but without concentric rings artifacts.

In audio world the processing requirements are very low compared to video so we can do perfect filtering, apodizing filters are low complexity, they need less memory so they yield low time latency that's why they are integrated into chips.
Apodizing filters are meant to be used in wireless mics, wireless broadcast headphones where you need to sacrifice sound quality for low time delay.

Apodizing filters are designed to combat procesing power not to achieve quality processing.

Apodizing filters are acceptable flawed filters (need less cpu)
Phase linear filters are perfect filters (need more cpu)

Fokus's picture
Quote:

made available an MQA-encoded file containing a train of perfect impulses. Later, he sent me a non-encoded 24-bit/96kHz FLAC file containing exactly the same information, ... We're skipping half the MQA process—the encoding part

So what you got in each case was a 96kHz file with a train like FS 0 0 ... 0 FS 0 0 ... 0 FS 0 0 .... One file was pure FLAC, the other file was 48k MQA, and thus had undergone the origami folding which burries the 24..48kHz band of the signal
beneath the 0-24kHz baseband. If the folding works then both files are essentially equivalent. It is known that this can work, the only errors introduced here due to the 6-7 bit quantisation of the > 24kHz band. But you won't see that on a scope on screenshot. So you tested something from which reasonably could be assumed that it would work. MQA advocates and MQA detractors will agree on this: this is mathematics.

Quote:

MQA's ... sampling method—and the resulting, presumed increase in aliasing—introduce randomness in precisely when those impulses occur. ... I synchronized the MQA and non-MQA impulse responses: MQA in the left channel, non-MQA in the right. Over 30 seconds of impulses spaced 0.7ms apart, examined on a microsecond scale, I saw no random offsets

Your digitally-generated impulse train started life in a 96k space and thus is de-facto synchronised to the sample rate. Of course you won't see alias-induced temporal offsets in this case. The test is invalid. What should have been done is to generate an impulse train of varying spacing in an analogue signal space (or failing that a 192k or 384k sampled space), and then send it through the complete MQA sampling chain. MQA's aliasing occurs in the 192k to 96k encoding step (which is nothing else than downsampling with extremely weak AA filtering). This aliasing will displace the impulses relative to each other and to the original signal. But this test won't happen, because MQA would argue that impulses are not representative and that the MQA system assumes (or rather: needs) a music signal, with its spectrum falling towards higher frequencies.

Pages

X