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
supamark's picture
Quote:

...I don't think I've ever seen an audio debate as nasty as the one over Master Quality Authenticated (MQA)...

Ah, have ye forgotten the still raging battles (in both pro and consumer circles) of tubes vs. transistors for over 50 years and analog vs. digital for 35+ years?

I think the real difference is that this is the only one of the three that has played out from the start on the internet.

JimAustin's picture

I think the real difference is that this is the only one of the three that has played out from the start on the internet.

That's a big difference.

Cheers,
Jim

supamark's picture

that it's a dramatic difference (pun intended).

Interesting and informative stuff so far (esp. that Mytek Brooklyn DAC), though I haven't really kept up on MQA since I don't stream music (I already own a large music library going back decades) so the vitriol is a little surprising to me... but only a little because internet.

Lookin' forward to further installments.

seldomheard's picture

I reserve Saturdays for "square wave listening" on my Ayre MX-R Twenty. Will this "clean up" the extra ringing I sometimes get when the moon's orbit causes severe high tides?

Also, is there a "square wave" version of Miles Davis' Kind of Blue? I'd like to add that to my collection....

JamieHowarth's picture

We measure with square waves so you don't have to.
And so you don't have to hear distortion, unless of course that's what you're listening for! Cheers

seldomheard's picture

Nobody I know that routinely tests amplifiers or loudspeakers in the past 10 years bothers with square wave generators. The current standard is the non causal digitally generated impulse. The funny thing is you have people in this thread using this "freak of nature" signal to show how real music will be either properly handled or mishandled by playback equipment and they are focusing precisely on the limitations of the signal generating technique to infer that a problem exists.

The prior explanation of the Gibbs phenomenon elsewhere in this thread is just one more example of how botched this "pimposium" has been. The impulse signal is a "discontinuity" - something that doesn't occur naturally in real sound because in order to create it, you need sine and cosine waves with frequencies that are FAR, FAR HIGHER THAN THE REALM OF NOT ONLY HUMAN HEARING BUT ANY KIND OF HEARING BY ANY KIND OF MAMMAL. The pre ringing which is actually more accurately described as a ripple constitutes the artificially assembled collection of sine and cosine waves in a FINITE approximation of the discontinuity function that a digital device is attempting to create. In a similar, but not equivalent way, the ringing (not ripple) just after the leading edge discontinuity of a square wave produced by an amplifier that undergoes clipping is also not a natural sound or anything one is likely to encounter in music. Impulses, and to a far lesser degree today, square waves, only have value or significance in establishing how a system behaves with a small fraction or portion of what is being visually depicted in a response graph. There is no need to delve into the world of megaherz frequency components when the ear brain combination is via repeated experiment found to be very limited in terms of time domain and frequency domain stimuli. And that is where for this "pimposium", the rubber hits the pavement and sadly in this case, the vehicle hits the brick wall.

JamieHowarth's picture

If only we were testing amplifiers... we're not. We're testing tape record/playback transformers/ pre-de-emphasis EQ, head characteristics, etc. and we're not using square waves but a square wave will look great when we're done. thanks for the lecture though, a lot of good information in there. IAD comprehension is looking to be like 5-10usecs so you're just straight up incorrect on the ear/brain being limited in terms of time domain and freq domain.

seldomheard's picture

If you are going to draw conclusions about ear brain limitations with respect to sound reproduction signals, you have more than just frequency/time limitations to deal with. You also have to know if the amplitude (intensity) variations of sound that are SIMULTANEOUSLY at play impact the net result. And in this perfect storm of a case, you really need to not only look at the amplitude of the ripples, but how these ripples could possibly affect the moving coil or suspended diaphragm in a loudspeaker in any meaningful way. Anyone can cherry pick a tiny aspect of a picture. To develop a full understanding and appreciation of the picture requires context. You should know better Jamie.

JamieHowarth's picture

I do know better, but I'm pretty sure I can't understand what amplitude variations you're talking about... the ones where the a amplitude is varying from 0 up to some peak along a varying rate of change then diminishes and crosses through zero and does a similar but likely not identical amplitude variation below zero then comes back up again?

What's a ripple?

seldomheard's picture

The apparently aperiodic summation of gradually decaying sinusoids defines ripple pretty well.

Ringing more aptly describes a narrower distribution of frequencies that whether decaying or oscillating continuously, tend to be more periodic in nature.
To understand the distinction, it helps to understand the origins of a discontinuity function and its mathematical construction.
What is important in this particular instance is that the person trying to convey a difference in the capacity of D/A converters is focusing on an anomaly of the input signal and not the ability of the D/A converter to handle it with fidelity - AND HE IS SEEMINGLY UNAWARE THAT THE "DISTORTION" HE IS COMPLAINING ABOUT - PROCLAIMING THAT "DIGITAL IS BROKE", IS AN ARTIFICIAL ARTIFACT OF THE INPUT SIGNAL HE IS USING TO MAKE HIS POINT" - That is the ridiculously funny part about this entire "pimposium".

John Atkinson's picture
seldomheard wrote:
Nobody I know that routinely tests amplifiers or loudspeakers in the past 10 years bothers with square wave generators.

If you read Stereophile, you should know that we always test amplifiers with squarewaves and that such tests are diagnostic for various problems. But perhaps you don't read Stereophile.

John Atkinson
Editor, Stereophile

seldomheard's picture

I stopped paying attention to square wave tests you've posted about 20 years ago. I have found the tests involving multi tone response to be much more interesting and useful.
More importantly, I pretty much stopped reading Stereophile altogether about 10 years ago when the level of subjectivity seemed to increase substantially overall. That characteristic seemed to be accompanied by a sharp reduction in objectivity and a sharp increase in intolerance with respect to questioning, inquisitive minds.
With respect to your involvement overall, I will admit that your measurement approaches and the attempt to correlate measurement findings with perceived listening experience have been the only saving grace for Stereophile in my mind. And I don't think I'm alone in that sentiment. Most of the "industry people" I've communicated with over the years, like me, have been in the habit of bypassing all of the subjective superlatives taking a "bee line" to your measurements.
A friend recently alerted me to this MQA article and suggested I create an account to post an alternate view. Prior to that, I think the last time I read anything from Stereophile was a few years ago.

Best Regards

JamieHowarth's picture

Reading my comments you may not have noticed that I said we don't use a square wave to measure the phenomenon in tape machine phase distortion that we wish to correct. Square wave tests actually would work but they would be so ambiguous on the way to the solution that a multitone test is obviously a better choice, and we have a well-thought-throug test signal.

I'm really tired of the overqualifications, and the nitpicks. It must be a real luxury to feel so good about one's probity. But I'm just trying to make stuff sound better.

seldomheard's picture

Straight from the lips of a top Mastering Engineer (Brian Lucey)

"MQA has been targeting the weakest players in our world, the audiophiles. And they’re targeting those most dependent on pimping new tech, the audiophile press. Meanwhile, one sided presentations at trade shows leave no time for deep Q and A and any real discussion panels are eschewed by MQA. " - Brian Lucey

interview can be found here:
http://fairhedon.com/2017/11/05/an-interview-with-mastering-engineer-brian-lucey/

So is this supposed to be a three part "new technology pimposium"?

I have read the entire sales pitch and sadly I'm more inclined to agree with Lucey than run out and buy another DAC that will in all likelihood sit on the shelf collecting dust for a year or two until the next "must have" digital toy comes along.....

JamieHowarth's picture

Brian is a wicked smart guy and a good mastering engineer, but I totally disagree with his last paragraph.. By his logic (that approved masters can't ever be played on superior equipment than they were approved on? That mastering touchups as speaker systems improve is illegitimate??) then mastering kinda maybe shouldn't happen either. Just put out what was approved at the mix stage and call it a day?

Truth is more complex --- there are now DSP interventions that can null out some of the distortion that made the tape record/repro more transparent, and that moves you closer to the console output, not farther. So I don't agree at all that it's heresy to try to improve the playback. If that were out of bounds playing a Sinatra master from 1955 on an ATR102 would be illegal and I don't think he'd contend that is the case. Same difference.

supamark's picture

the finest analog tape deck I ever used, sounded great and easy to edit on. That studio also had a Studer A820 24/16 track (had both heads) with the auto tape set up and Dolby SR, and was by far the easiest analog tape deck I ever used lol. Also had an Ampex ATR-124 but it was so unreliable they just stuck it in a store room and bought the Studer (this was at UT - Austin, not a commercial studio, so selling it was... complicated) because they got a deal by going in with Austin City Limits which filmed in a studio at UT and bought one at the same time.

Brian Lucey's picture

You're attacking the Straw Man, not the Wizard.

ok's picture

..so is it just my guess in f6 that MQA reconstruction filter is ironing the DAC's native response – save when that is already MQA-straightened out?

JimAustin's picture

The way I understand it, hardware designers have some leeway in how MQA is implemented. With the Mytek DAC, when MQA is enabled, all data is sent via an MQA-specific path; that way, the basic DAC hardware doesn't need to figure out ahead of time whether the data is MQA or not; the MQA chip can do that.

Once it's sent down that path, it uses as its reconstruction filter MQA's "upsampling renderer"--the same filter MQA uses.

Best,
Jim

ok's picture

..if this was really the case, shouldn’t MQA-on DAC’s final impulse response stay always the same with non-MQA files regardless of the hosting DAC’s type – unlike Bob Stuart’s statement at manufacturer’s comment?

JimAustin's picture

In his Manufacturer's Response, Stuart was referring to Fig. 4, in which an MQA file was sent to a non-MQA DAC. So in that case, my comment (just above) doesn't directly apply. I'm not sure I completely understand this, but I'll take a shot. The original file is nothing but a single sample near FS, but apparently encoding (to MQA) imposes on the impulse a bit of MQA character--what's normally called "minimum-phase". So you're seeing a convolution of the encoder and the non-MQA DAC's linear-phase reconstruction filter.

ok's picture

..you're probably right, I didn’t notice Stuart’s final remark – note the phrase "we expect" though – which reminds me anyway that an MQA-on DAC might still be ironing.. well, I don’t know what.

supamark's picture

The effect of sending a non-MQA signal through the processor in the Brooklyn DAC looks quite appealing in:re ringing, and I'd be curious to A/B it switched in and out for myself in my system to hear what the trade-offs are to my ears (and if it's audible to me).

dce22's picture

There is no ringing (aka oscillation)in Fig.1,
What you see there is Time domain representation of all the frequency that the system is capable,

If you connect infinite sine generators from DC to 22khz and mixed the signals togethere and look at the output on the scope you will see Figure 1 signal
you can connect infinte AD to DA to AD to DA and so push pulse signal thru the chain it will be still the same Figure 1 signal,
If there is ringing the ringing portion will be amplified with every conversion AD to DA That is not the case.
This is not ringing Retards.....

Any deviation from Figure 1 is Time and frequency Distortion

[Flame deleted]

https://en.wikipedia.org/wiki/Sinc_function

dce22's picture

Apologies if im too beligerant, but this pre post ringing miss terminology is very bad the ripples you call ringing are essential to preserving the time-phase of the high frequency in the system and should never be classified as ringing because it is Not!

Just like adding sine wave harmonics to create a squarewave you get ripples on the start and the end of the step is not ringing you just dont have enough high frequency and bandwidth to represent perfect squarewave you dont have bandwidth to create sharp pulse Everybody knows that this is not ringing

https://upload.wikimedia.org/wikipedia/commons/b/bc/Fourier_series_for_square_wave.gif

You can phaseshift the high frequency to get the ripple after the main lobe but you have to know that you are destroying the audio timing resolution not improving it, offcourse in the mind of some people KNOW IT IS RINGING AFTER NOT BEFORE (ringing is a total lie),
If you wanna remove the ripples before and after the lobe just get a eq on your PC and roll off high end from 12khz up and it will be just the main lobe congrats you stopped the "ringing" even with oversampling filter active.

JimAustin's picture

Hey dce22, calm down! "Ringing" is just a word--and it's a reasonable word to describe what you're seeing.

I think what you're trying to say is that the cause of the ringing--or choose your word--is limited bandwidth--that if you had enough frequencies, you wouldn't see what we're calling ringing. True enough. So what? It's also true that, as these (and many other) measurements show, the choice of filter has a major effect on the shape. It's not only whether the high frequencies are delayed (as in so-called "minimum-phase" filters). In general, the sharper the filter--the more sudden changes, or "kinks" it has, the more you see this example of what's widely known as the Gibbs phenomenon. For the time domain, gradual is better.

Filters often referred to as "minimum phase"--also, and perhaps more accurately, called "causal", since that's really the property we care most about, and not all of them are truly minimum-phase--do indeed delay the high frequencies so that they arrive after the low frequencies. In the consumer-audio world, such filters have been argued to sound better than linear-phase filters. The late Charles Hansen was an early proponent. I'm not sure who first proposed this idea, but MQA's Peter Craven was in there somewhere, early on.

This is, of course, controversial. It is open to question. I've spoken to several digital designers who have made strong cases for the importance of linear-phase filters and the distructiveness of "causal" filters. One of the smartest and most convincing is John Siau from Benchmark, who's assistance and advice have won my eternal gratitude, for what it's worth. There can be no doubt that in a pro context, where several or many passes through ADC/DAC are anticipated, it would be foolish to screw up the phase alignment too much. The accumulated phase effects would be too large.

But what about in a consumer context, where there's only one trip, and the two stages--encoding and decoding--are able to work together? That's a whole different thing. They're not designing DACs and ADCs. They're designing a digital music delivery system.

"Ringing" is just a word. Call it whatever you want. It is what it is--an extended time-domain response that, in the linear-phase case, partly precedes the main impulse. Whatever you call it, and whether the cause is the subtraction of information (correct) or the addition of information (wrong and never implied), that is what it is.

There is one other key point to make in an MQA context. Even well-trained digital engineers tend to learn a frequency-centric version of sampling theory. One that assumes a fixed bandwidth and asserts that the sampling rate should be twice the bandwidth. There are more recent, more mathematically sophisticated formulations. A more modern treatment rejects the assumption of frequency band-limitation and treats time and frequency domains symmetrically. One domain can be traded off against the other. When you frame things that way, you get different basis functions--NOT sync functions--and you get new filtering options. What I'm getting at is, it's not only a question of delaying high frequencies. There are other ways to improve--shorten--time-domain behavior. You may need to learn some new math.

Meanwhile, you'd be better off not calling people names. It's not nice, it makes others dislike you, and it makes you look foolish when you prove to be--well, not wrong, but inadequately informed.

Be Well,

Jim

cgh's picture

Ringing and Gibbs phenomena are different things.

JimAustin's picture

You:

Ringing and Gibbs phenomena are different things.

Wikipedia:

The main cause of ringing artifacts is due to a signal being bandlimited (specifically, not having high frequencies) or passed through a low-pass filter; this is the frequency domain description. In terms of the time domain, the cause of this type of ringing is the ripples in the sinc function, which is the impulse response (time domain representation) of a perfect low-pass filter. Mathematically, this is called the Gibbs phenomenon.

I can provide a scholarly citation if you want one, but most people these days are satisfied with Wikipedia.

JimAustin's picture

Note the word "ringing" in the Wikipedia entry.

cgh's picture

Funny Jim, I didn't google either word for my response. I was simply responding to the other gentleman's post about summations. From a signals/filter perspective I still view them as different, but I can see the similarities from your quote now that I think about it, and can see how an ee would think of it like that.

(As a post-script, I should probably think about updating my dated doctorate in mathematical physics with the university of wiki... hopefully this will not invalidate any of my scholarly articles that often make extensive use of various linear integral transforms.)

JimAustin's picture

Hey, I know all about outdated doctorates; I've been out of research since 1996. I wouldn't have known that fact myself if I hadn't been doing the research very recently. It is, though, a textbook case of Gibbs, from a mathematical perspective. I'm sure you'll see that with a few minutes study.

Seriously though--and respectfully--maybe there's a lesson here? There's a lot of people piling on, assuming other people are idiots. You immediately assumed that I was wrong because it didn't ring true. You were sure enough to post it on a forum--maybe your standards are lower because no one knows who you are here? And you were wrong. Maybe it's time to slow down and reconsider. I don't just, or mainly, mean you. There's a lot of it going around.

We may not be right, but we're not stupid.

Best,
Jim

cgh's picture

No, Jim, you are wrong. I wasn't piling on at all. I assumed nothing about you. You addressed me as "You" in your response. I was simply reacting / responding to the other fellow, not you, as I took the time - too much time, lost time - trying to digest all of the responses in this long comment thread for a shred of something that advanced my personal understanding of MQA.

It's been horrible trying to get any reliable, objective, or factual information on MQA. No doubt the responses have been horrible and inappropriate, but they are coming from people that are being asked to spend their money supporting an industry with a well-earned credibility issue. The latter only mens that the bar is a bit higher for the industry and, from my lay perspective, they flubbed the MQA roll out exasperating the whole issue.

I directed nothing personal at you, nothing was insinuated, and I certainly am not trying to amplify any of the other vitriol. I finished reading this thread, so I guess I am done...

JimAustin's picture

My apologies for misunderstanding.

seldomheard's picture

You kind of butchered the explanation of each - probably adding more confusion to the subject than clarity. Causal systems are linear, time invariant systems whose output only depends on current and past inputs.

Minimum phase systems are causal and stable systems whose inverse is also causal and stable.

Linear Phase systems are systems whose phase response in the pass band is linear with frequency (phase rate of change with frequency or derivatve is a constant - lacking group delay)

As for all the other mumbo jumbo hand waiving regarding impulse responses translated through each DAC - it's pretty obvious that the standard DAC (NOT MQA) which used a gradual roll off filter - produced the best impulse waveform. Whether or not a slight pre ringing existed in the output is totally irrelevant. We're talking about tiny voltage fluctuations that are a fraction of the fundamental impulse occurring over microseconds when the time it takes voltage to induce a magnetic field and move a coil in a loudspeaker is orders of magnitude longer. This is yet another "forest for the trees" hand waiving exercise designed to sell the uninformed on a solution to a problem that doesn't exist.

mcgilroy's picture

Hi Jim,

interesting stuff - could you point me to the sources you refer to in this passage:

"There are more recent, more mathematically sophisticated formulations. A more modern treatment rejects the assumption of frequency band-limitation and treats time and frequency domains symmetrically."

Thank you!

JimAustin's picture

I don't have a citation handy. Author to look for is Unser. Most accessible article is in IEEE's DSP journal. I'm thinking it was from around 2000, 20001, but I'm not sure.

mcgilroy's picture

Thx - quick Google-fu brings up Michael Unser and this article:

http://ieeexplore.ieee.org/document/843002/

Is that the author you mean?

mcgilroy's picture

Also: have you considered writing Unser and ask him if he feels his ideas properly represented in the MQA-literature?

Could be fun ;)

JimAustin's picture

That is the author--and I have read that paper. I found the one in the IEEE DSP journal a little more accessible, though still highly mathematical.

I wrote to Unser weeks ago. He didn't reply. I should try again.

SNI's picture

@dce22
Finally somebody got it right.
There is in fact no ringing in digital filters.
The pre- and post ringing seen og the impulseresponse from a digital filter is just the nature of a
bandlimited signal, they simply look like that.
So what you see is the reaction of a filter, presented for frequencies which are being filtered.
The music signal does not have any ringing at all, as this is in the passband of the filter.
Passband ripple has through the last 20 or more years been neglegtible. They are så small, that any analog equipment will present more ringing a lot actually.
This might be the most misunderstood measurement ever.
The only thing one can use this impulseresponsemeasurement for, is to se if the filter is phase linear or not.

CG's picture

Look at figures 5 and 6 above.

Now, open a new browser tab or window to here: https://www.stereophile.com/content/ayre-acoustics-qx-5-twenty-da-processor-measurements and look at figure 1.

If the scale in figure 1 is hard to visualize, try page 4 here: https://www.ayre.com/white_papers/Ayre_MP_White_Paper.pdf (dated February 4, 2009)

Who says that you can't learn a lot about audio by reading Stereophile?

I'm having a hard time determining the ratio of irony to tragedy in this one.

supamark's picture

I see nothing ironic or tragic, just that Jim found the same stuff the guy(s) at Ayre did and that Ayre's published measurements match JA's. so far, that's good stuff. It's not unlikely (I haven't looked, so I don't know) that Ayre and Mytek get their DAC chips from the same company (there ain't a whole lot of options). It's also not unlikely that Ayre and MQA have arrived at similar solutions to the same problem (ringing).

seldomheard's picture

If John Atkinson and his buddies at Ayre stopped measuring amps with straight resistance at the output, you might all discover that the ringing goes away pretty fast. The low pass filter that is your loudspeaker makes pretty quick work of signals high above 20khz - unless of course your amplifier is wide band and not very well protected from oscillation. In that case, your amp will do a little "ringing" on its own and quickly morph into a toaster.... Anyways, this foolish argument about impulse response, "ringing", "time smear" and the like is getting a little boring. Can we go back to discussing the benefits of things like power conditioners, placing amplifiers in tubs of beads for "isolation" and suspending speaker cables above the ground to avoid "electrical interference" from the floor? I found those ridiculous, pointless, exercises in crass commercial propaganda far more entertaining.....

jgossman's picture

You missed Steven Meijas' article/post about a return to cassettes simply for the enjoyment of using them. Anti-cassette people are a special kind of angry little man. Most the criticism I see (and an opinion I hold) is a perfectly acceptable one. Seems like a solution to a problem that doesn't exist.

spacehound's picture

"It's clear to me that MQA's developers see it as an idealistic venture designed to fix what digital broke....."

What's more, this 'Part 1' is totally wrong. Testing on 5 uS pulses
is meaningless nonsense.
As JA said, though I prefer 'nonsense' to 'illegal'.

And what you show IS NOT RINGING. (If it was the rule of 'cause and effect' would be wrong. I could start off for work today and arrive last week. Or wind is caused by trees waving their branches about)

PUT VERY SIMPLY:

ALL 'waves' whatever their shape can be shown to be a build up of sine waves.
What you incorrectly call 'ringing' is the waves building up enough for you to notice, hear, or measure. if you used a longer pulse you would eventually see the 'square' wave or whatever it happened to be.

Also see dce22's posts and read his links. You HAVE to understand (or at a minimum accept) the mathematics before you write about this sort of stuff or you just produce gibberish. And it's why you don't seem to understand why the Benchmark DAC sounds fine. It's not 'ringing' and it's not a 'fault'.
The Brooklyn DAC is not 'better', it just messes up the timing by adding a phase shift so you don't see the perfectly legitimate 'building up waves' on your scope, they get added to the 'reducing to nothing waves' at the end instead. Which is a silly thing to do but it fools some people.

tonykaz's picture

We have persons with "vested" interests defending their considerable libraries of investment grade Media being threatened by a "New" technology that promises "Master" Quality.

Media Collectors ( like me ) are gonna fight MQA to the death. We have our cherished ( life long ) Collection's $ value at stake.

What point is there in owning a $30,000 + record player if any iPhone owner, walking down the street or riding the Subway or sitting at Taco Bell can access Master Quality Level Performances from their cheapo iPhone 5.

Where is our Audiophile "Velvet Rope" of exclusivity?

Why have we been filling our basement Audiophile Cave with walls of shelving to store all those thousands of crappy Vinyls that we'll probably never listen to for the rest of our entire lives?

If MQA becomes pervasive ( which seems likely ) we Vinyl collectors will face even greater ridicule ( deservedly ).

MQA is more digital poison, along with that horrible RedBook 16/44.1 !

This digital toxicity is Satan's tool to destroy the Morality of good old Analog Vinyl. Anyone that says different has a "butt hole for a brain" ( an Analog Planet direct Quotation ).

Well, of-course;

I'm a Vinyl Apostate.

Vinyl died around 1985, as far as I'm concerned.

Analog vinyl's "True Believers" still cling to disparaging all attempts of the Audio Industry's advancement.

Mr. Austin should expect a barrage of flack raising up from the bunkers of old school "absolute sound perfection seekers".

MQA has a flaw,

somewhere,

doesn't everything?

21st Century Tony in Michigan

spacehound's picture

But Mr Austin simply doesn't understand what he is seeing on his scope so writes gibberish.

'Digital' is not "broke" and doesn't need fixing by MQA or anything else.

There is always room for improvement. Straight off the top of my head I would increase the sample rate above the 44.1 CD norm so we can move the filter out a little to make it further away from the audible range. You certainly do that with 96 but I personally think that's further than is needed.

DH's picture

Well known mastering engineer Brian Lucey has publicly stated that MQA versions of albums he mastered aren't authenticated by him (possibly by someone at the record label) and that he thinks their sound is both altered and inferior to the actual "master".
So MQA means:
"not the master, not the quality, and not authentic".
It's a remaster with different filtering applied to give it a certain sound. Either you like it or you don't.
My question is what for? Just sell/give us similar filters built into playback software and let us apply them ourselves if we like the way they sound.

JimAustin's picture

Lucie's argument is that anything that's altered relative to his master is by definition inferior. Then he goes on to further justify his viewpoint by expressing negative observations about the sound--which don't jibe with measurements, or with what I or anyone else who's ears I trust hears. That's not a commentary on MQA, but on his aural observations.

If Lucie didn't sign off, that must mean that he doesn't own the IP. Hardly surprising; despite being the last step in the production process, most mastering engineers do NOT own the IP. The content owner can, if s/he wants to, go off and get it remastered again and again--which is pretty much what's happening with MQA. I understand that he's not happy about that, but then life isn't perfect.

I find Lucie's position a little precious--a little self-important. But whatever--yes, MQA changes the sound, just a little. I'm pretty sure Ben Webster (e.g.) didn't sign off on the MQA version of Soulville; too bad, can't be helped. Presumably he's not complaining.

In future, for new music, artists will indeed get to sign off on it, and mastering engineers will--already do in fact, in some cases--have the opportunity to hear what their master will sound like in MQA.

AJ's picture
Quote:

Lucie's argument is that anything that's altered relative to his master is by definition inferior.

Actually, he nicely exposes the fraudulent claims about MQA being "authentic", rather than pure marketing hyperbole. Guess not all mastering engineers have drank the koolaid.

His actual real comments can be read here http://fairhedon.com/2017/11/05/an-interview-with-mastering-engineer-brian-lucey/

Quote:

Then he goes on to further justify his viewpoint by expressing negative observations about the sound--which don't jibe with measurements, or with what I or anyone else who's ears I trust hears. That's not a commentary on MQA, but on his aural observations.

Please do explain how him not liking the added aliasing distortion, the only thing MQA actually does is at odds with the measurements, vs those who swoon over it, "jibe".

JimAustin's picture

I mean no disrespect to Lucey--sorry I misspelled it. He seems to be a good mastering engineer. I've enjoyed some of the recordings he's worked on. He said, in that interview:

MQA brightens the high-mids in the Mid section while thinning the low-mids on the Sides. There’s also some harmonic distortion which some people could find pleasing,

There's no evidence of any consistent alteration of the frequency response in MQA. The key word here is >>consistent.<< It is possible to observe (usually small) changes in the frequency response, mostly due to changes in the background noise level, or so it appears. Oh, and aliasing distortion isn't harmonic.

Here's the money-shot:

When a record is first tracked, then rough mixed, mixed, revised, mastered, revised in mastering and finally approved … there is no fixing it. Anything that changes violates 5-20 people who have all signed off. Distortion artifacts are musically incorporated in to all music production, there is no perfection in music.

Here's one more quote along the same lines

* "If I want that distortion in the master I would’ve put it there in the first place."

He just doesn't like anyone messing with the sound that he produced, in collaboration with all the artists and engineers that worked on the recording before him. I get that. I followed his comments on Gearslutz before he did that interview. That's his main point; the rest came later. The bit about "perfection" is a strawman, as "perfection" has nothing to do with MQA. And of course it's also true that other mastering engineers are quite impressed with MQA: Katz, Ludwig, Silverman. A the recent AES meetingI heard--but cannot confirm--that Jimmy Douglass (mixing, not mastering) expressed admiration for the improved sound of 4:44, Jay-Z's latest. But who really knows.

Please do explain how him not liking the added aliasing distortion, the only thing MQA actually does is at odds with the measurements, vs those who swoon over it, "jibe".

Can you back up your claim that aliasing distortion is "the only thing MQA actually does"? Can you support the assertion that MQA's aliasing distortion is even audible? Or are you merely expressing an opinion, informed or not?

By the way, I think your subject line--"fabricated authenticity"--is interesting and raises an important point. If you consider the digital object to be the "original"--that's Lucey's perspective, and it's reasonable--then, philosophically, there is something artificial about the idea of making it more original (by, as MQA would claim, getting rid of artifacts introduced during the A-D conversion). I'll write more about this in a future article.

AJ's picture
Quote:

Then he goes on to further justify his viewpoint by expressing negative observations about the sound--which don't jibe with measurements, or with what I or anyone else who's ears I trust hears. That's not a commentary on MQA, but on his aural observations.

There's no evidence of any consistent alteration of the frequency response in MQA. The key word here is >>consistent.<< It is possible to observe (usually small) changes in the frequency response, mostly due to changes in the background noise level, or so it appears. Oh, and aliasing distortion isn't harmonic.

So by your own admission, it's possible to observe (measure) changes in FR...but *his* observation of FR changes, don't "jibe" with what tiny sample of measurements you have performed???

Oh BTW that's really nitpicking, yes the distortion is anharmonic, so he misspoke as a mastering engineer, not a EE.
To get a full picture of what MQA actually does, I would suggest you read a real DSP experts view, rather than MQA marketing mumbo-jumbo for audiophiles nonsense:
http://www.audiomisc.co.uk/MQA/origami/ThereAndBack.html

JimAustin's picture

So by your own admission, it's possible to observe (measure) changes in FR...but *his* observation of FR changes, don't "jibe" with what tiny sample of measurements you have performed???

A tiny fraction of all that's available, yes, but not a small number. I've probably studied more MQA spectra--the analog output of a MQA-enabled converters, compared, when possible to non-MQA versions from (apparently) the same master--than anyone outside MQA, the company. Hundreds, surely.

jca

AJ's picture

Unless you have measured the specific tracks that Lucey listened to, I have no idea how you could correlate/jibe measurements vs what he subjectively perceived.

This may all be a moot point anyway, as I posted in JAs thread, there are news articles that don't bode well for Tidal music.
https://www.engadget.com/2017/12/13/tidal-jay-z-financial-trouble/
If that happens, I don't see MQA surviving.

JimAustin's picture

Unless you have measured the *specific tracks that Lucey
listened to*, I have no idea how you could correlate/jibe measurements vs
what he subjectively perceived.

Unless his observations are general--unless they apply to more than those specific tracks, his observations have little merit even if they're accurate.

Yeah, I've got no idea whether MQA will survive. Only time will tell.

jca

AJ's picture
Quote:

Can you back up your claim that aliasing distortion is "the only thing MQA actually does"?

I meant that in the sense of the MQA processing itself, the "leaky filter", etc. The "Time smear" is marketing mumbo-jumbo for adolescent believers in Bogey men who think digital is evil due to "time" issues, there is zero evidence that old men can hear anything above 15-16k, much less "time" >20k. No, the "Hypersonic" feel good effect nonsense doesn't count.

For the full "MQA "authentic" effect", yes, there is more than just adding aliasing distortion and Re-EQ:
http://www.superbestaudiofriends.org/index.php?threads/mqa-op-ed.3817/page-3#post-166101

JimAustin's picture

I meant that in the sense of the MQA processing itself, the "leaky filter",
etc. The "Time smear" is marketing mumbo-jumbo for adolescent believers in
Bogey men who think digital is evil due to "time" issues, there is zero
evidence that old men can hear anything above 15-16k, much less "time" >20k.
No, the "Hypersonic" feel good effect nonsense doesn't count.

I'm sure you're aware that there are people in the world--smart people--who disagree with you on this point. There also are smart people who agree. This is, of course, a crucial point: If you're right, MQA is nonsense, even if it's sincere, even if its elegant. That's one of the things we aim to test. Yes, I'm ambitious.

For the full "MQA "authentic" effect", yes, there is more than just adding aliasing distortion and Re-EQ:

Yes, I discovered long ago that MQA versions are often made from different (almost always better) masters (than equivalent CD-res recordings). I found it weird that so many critics used this as another cudgel to beat on MQA--as if, by releasing MQA versions of the best possible masters, their goal was to fool the public. In fact, I think it had just that effect, since many people (including writers) initially made comparisons without knowing what they were comparing. And usually, when a high-res PCM is available, it was made from (or perhaps is) the master the MQA version was made from. It's only when you compare to CD-res that you typically find different masters at the root. (In my case, my early comparisons were mostly decoded versus undecoded. That's an unfair comparison, but at least you know with a high degree of confidence that both are from the same master.) But even if it had that effect, it's silly to suggest that it was an attempt to deceive.

AJ's picture
Quote:

This is, of course, a crucial point: If you're right, MQA is nonsense, even if it's sincere, even if its elegant. That's one of the things we aim to test. Yes, I'm ambitious.

I will be truly fascinated by how you test whether this is really a time ("smear") vs a frequency (aliasing) issue in isolation, given their interdependency.
Party on Jim ;-)

JimAustin's picture

As you say, frequency and time-domain are equivalent. So, assuming MQA can be demonstrated to shorten impulse response, then--you're point not mine--they really are one and the same. If you accept that the frequency domain changes alter the sound, then you admit that the time-domain changes alter the sound. The only question is conceptual--what's the best way to think about it?

AJ's picture

"Time smear"
Now if they had said, "Hey we added some nice(?) frequency "spicing" to the mix in the form of the aliasing with our silly time filter...and you may or may not hear/like it." I'd be ok with that. There's all sorts of ways to add distortions that some find pleasing (Vinyl, certain tube design, etc).
Lets call a spade a spade. "Authentic"???...please, no.
Worse, I don't want that spicing *to be the master*. A version, sure, knock yourselves out. But make no mistake, having a "spiced" master is the goal.
You may have noticed a bit of consternation about that goal...

spacehound's picture

It was Lucey's preciousness and huge self-importance (usually expressed by being rudely dismissive of any view different from his own, and also being totally dismissive of ALL music 'consumers') that got him booted off Computer Audiophile.

The second one amused me. I wonder who he thinks recorded music is for?
And if it is live people like him are not needed :)

Brian Lucey's picture

I have great respect for Audiophiles, who are not assholes about Dynamic Range, and I said so. Try and not follow MQA on the lying train off the cliff of self importance.

Brian Lucey's picture

Skipping your personal attack I have to ask ... how do you misspell a name you've seen so often and pretend to be intelligent? P.S. What do you do for a living? I make VERY FINE adjustments to CLIENT APPROVED masters. These are being BUTCHERED by MQA, in BATCH PROCESSING, because the MQA deal with labels forces the back catalog of my work to all be released MQA.

How many lies does it take for you to smell a stinky fish?

As far as having the codec here, I passed because they demanded a NDA, and I already heard the best they could do on some of my work, and it was high mids forward, sides thin in the low mids, and harmonically distorted. Blurry in a subtle way some might like. Yet I do some very specific things with harmonic distortion post clipping already in my work and it's DESTROYED as they batch process or when it was done with care. RIP MQA, it's bad for the world. Bad for science, bad for music. As if a better filter and codec system for PCM will never be invented? Stupidity. Greed. Lazy.

JimAustin's picture

OK, I'll fess up about the name thing. I actually didn't remember the exact spelling--just hadn't dedicated the brain space to it--but knew that I could scroll up 6 inches and see what was likely the correct spelling, since the previous poster had used it. So I guess both things are true: a spelling lapse plus a small gesture of disrespect.

But it was true what I said about your mastering. I own and enjoy the Lucinda Williams vinyl--did you do the vinyl or only the CD? My son played Arctic Monkeys for me the other night. I thought it was pretty good. That's you, right?

Anyway, welcome to Stereophile.

Brian Lucey's picture

AM I did. Lucinda for both, last 2 records. www.magicgardenmastering.com/

JimAustin's picture

Great, thanks. yeah, it was Ghosts of Highway 20 I was referring to, on vinyl.

Brian Lucey's picture

All my vinyl work is cut flat by Chris Bellman and Grundmans, and has around 3db more DR than the CD, plus other adjustments for the medium.

JimAustin's picture

OK, now I'm learning stuff. Why more DR on the vinyl, when the CD is capable of so much more. Is it just that clients require more dynamic compression on the CD or what?

Brian Lucey's picture

What you hear on CD/digital is EXACTLY what the client wanted you to hear. Mastering is good or bad at HOW it does, not WHAT it does. Vinyl cannot take the same levels, side low end, Essing, subs, or high end ... it's a physical medium and requires a more conservative premaster file. And luckily ... the vinyl audience is not volume knob averse.

JimAustin's picture

Thanks. That corroborates something I've been thinking for a while--not that I'm the first one to have it.
(it's original to me though.. Vinyl sounds better (when it does) because the physical requirements of the medium impose better sound, in a way--or what we audiophiles would consider better sound--even if the clients don't want it.

I appreciate the feedback.

jca

adamdea's picture

"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"
Jim Austin where did you get this from- it seems to me to be an entirely inaccurate statement. If you know what you are saying then it is absolutely misleading. An impulse response is simply a time domain statement of a system's behaviour and is entirely interchangeable with a complete frequency domain statement. But no no no -you can't just look at how short the spike is and decide that that makes the overall system more faithful.

Consider a NOS dac whose impulse response is one sampling interval wide. It is not accurate in the time domain at any point between the sampling instants. It's not faithful at all. Having a "wider" response is absolutely necessary in order to reproduce the signal between the sampling instants.

Impulse responses are not a shortcut way of measuring accuracy in the time domain (whatever that means).
If you are going to claim to be a reporter asking the real questions then start asking the real questions. If you want to learn something go and talk to a real expert like Jim Lesurf (who posts on Pinkfish). Start asking critically what exactly time domain optimisation might mean and how it could/should be demonstrated on a real world music signals.
I would greatly respect Stereophile if it decides to start (or you could say resume) engaging in proper genuine thought leadership.

I have noted a habit in the audiophile press of deriding internet forums, but I can say with absolute conviction that the standard of discussion and thought on many internet forums is far far higher than the audiophile press seems to be prepared to engage in. I have had the privilege of engaging in discussion on those forums with real experts who have shared their time to give real thoughtful and illuminating answers. (all of which began when I was trygin to understand the sampling theorem and the fuss about minimum phase filters around 2005 or so) If the audiophile press does not want to engage in rigorous thought then it will end up confining its readership to those who basically don't really care what the blurb means.

JimAustin's picture

Jim Austin where did you get this from- it seems to me to be an entirely inaccurate statement.

I'm pretty sure that the first time I encountered this idea was in a sophomore-level physics class at Swarthmore, where I earned a bachelor's degree (in physics), before going on to grad school and completing a PhD (in physics)

MQA folks have focused on impulse response as an illustration of their technology's time-domain performance. But of course it is much more than that. If you know the impulse response of a system, you know how it will respond to any input. Or, as Wikipedia puts it, "Since the impulse function contains all frequencies, the impulse response defines the response of a linear time-invariant system for all frequencies."

So what's the problem?

Jim

adamdea's picture

Jim
I'm not an expert, but I have spent a lot of time thinking about this stuff, and I've been asking hard question of the smartest people I could find for some time with a view to getting to the bottom of this stuff. If there are any professors of information theory, metrologists, medical physicists or the like reading, I will be happy to be corrected because I actually want to know the answer.

I think the answer is as follows:-

The problem is not in considering impulse responses as mathematically interchangeable with complete frequency domain (with phase) responses as a complete statement of the response of a system. That is just maths.
The problem is in thinking that you can evaluate a system by visually comparing its response with an ideal. "Looks a bit like" is not a mathematical function Jim.
They did not teach you that in your physics PhD.
If you consider the concrete example I used you will see that the superficially "good"NOS dac impulse response does not yield good time domain response (look at the staircase voltage/time plot).

Moving to a deeper level of understanding as to why all this is the case involves going beyond my lowly paygrade, but one clue is that we are talking about sampled systems not continuous time systems. This requires a frequency band limit. Thus the response for a theoretical signal having unlimited bandwidth is particularity misleading. It's a mathematical tool not a visual one.

If you have an impulse response consisting of an impulse for the reconstruction filter in a sampled system then you can't be reproducing the moments between the sampling instants, now can you?

An impulse response is not the literal response to a real world signal, and it is not something which can generally be visually compared to see how much it looks like a perfect response. That said you can draw some conclusions. eg that anything with an asymmetric impulse response must mangle phase (ie mess things up in the time domain).

So what is a perfect impulse response. Strictly in a band limited system it is a sinc function with infinitely long extension in time. If it has an infinitely sharp filter slope then the central lobe will be infinitely narrow (I think). Is this good time domain behaviour? Rob Watts would say yes.
Perhaps he is not right about that, but otherwise it is difficult to see what a perfect impulse response is. And even then, deciding whether something looked a bit like it would not be a useful piece of analysis.

Considering the supposedly bad (according to MQA) impulse response. Why is pre-ringing supposed to be a problem?
You can't get away from the point that only energy in the transition band of the filter can pre-ring. How in god's name could that be a problem when
a) its at 48 khz for 24/96 recording
b) there virtually no signal at that level anyway so all you would be distributing is noise.
c) the filter is sharp so the transition band is small.

This shows that merely looking at the shape of the impulse response can be misleading

So is a really narrow impulse like thing with a teeny bit of post ringing better than an infinitely long linear phase filter with an infinitely small transition band which cuts in at 1MHz? How does your visual inspection help here?

Just messing around around with impulse responses is not analysing the behaviour of an audio system in the time domain.

Let's go back to this "time domain" analysis- where is the evidence of a real world signal behaving better in the time domain with the MQA filter. I have never seen anyone attempt to do this.

JimAustin's picture

I much prefer this (mostly) respectful mode of communication. So much better if we don't just yell at each other and consider each other idiots from the start. Despite your "lowly paygrade" comment, it's clear that whoever you are, you know some things.

There's a lot of sensible stuff in this post. I don't think I have time to deal with all of it. I'll start by admitting that in calling a short time-domain response "good," I was glossing over a lot of controversies. My approach--and I may not have made this clear enough--was to try and determine whether MQA was doing what it claims. Then again, I only had two pages to work with, and there's more to come.

It's important for me to note that I am not trying to defend MQA--except against it's rabid and uninformed detractors. A few specific points:

... but one clue is that we are talking about sampled systems not continuous time systems. This requires a frequency band limit.

Well, not exactly--and I mention this because it's relevant to a later point you made. This certainly is true in the conventional view of sampling theory. But did you ever notice the inherent asymmetry in this approach, when in most respects the time/frequency symmetry seems essentially perfect? Turns out there's a more recent generalization of Shannon's work that does NOT assume strict band-limitation. I mentioned this in a response to another comment. This remains to be investigated, but it appears to be connected to MQA--rather, there's some evidence that MQA is an application to music/audio of this work, which was developed abstractly and then in other fields.

So what is a perfect impulse response. Strictly in a band limited system it is a sinc function with infinitely long extension in time. If it has an infinitely sharp filter slope then the central lobe will be infinitely narrow (I think). Is this good time domain behaviour? Rob Watts would say yes.

I'm not sure about the details either. I agree that someone like Watts would probably suggest that a filter like the one you've sketched out would be ideal. He may be right. Ultimately though, this is a psychoacoustic question--or, as MQA people suggest, maybe a neuroscientific question. Charles Hansen, with whom I often quarreled--and who was among the most vocal MQA critics--surely thought such a filter less than ideal; he thought you could hear the pre-ringing. I own one of his CD players. I think it sounds good. I also think the Benchmark DAC3, and the PS Audio Directstream (which I also own)--both strictly linear-phase--sound very good.

As for your point about energy in the transition band--that's a damn good point, and one a noted digital designer made to me in an email correspondence. I'll discuss it in a future MQA article.

Now time for a drink and dinner.

Best Regards,
Jim

adamdea's picture

I'm not sure about this generalisations of Shannon to which you refer. I am aware that there are situation in which non legal sampling might be worthwhile, think there is an example where you are trying to identify when an event occurs using quite low sample rates.It's also worth bearing in mind that (i am told) in optics people rarely use anti alias and anti imaging filters.Hence aliasing.

In my searches I have only really found one text dealing with time domain encoding and how it applies to sampling. It contains more information than the whole history of stereophile and all the MQA patents.
http://www.dspguide.com/CH3.PDF
figure 3-15
If you want to know when an event occurred it is possible that having no anti alias filter might be better in some circs with a pretty low sample rate. However the problem is that I can;t see how this applies to audio events. The problem is they are not sub sampling interval events. They always take some time and 44khz sampling is pretty fast relative to the attack and decay of a note. is this why no one ever uses real audio events to illustrate "time domain benefits"
An inquiring mind could not avoid asking that question.

Now turning the subject to psycho acoustics, we do indeed have some questions to ask about this time domain. An honest an inquiring mind would then have to look more generally at what we know about the ear. Is there a separate mechanism outside the cochlea for detecting events in the time domain? What does this tell us about the possibility that detection of events in the time domain involves detecting frequencies you can't hear because your basilar membrane is too stiff and or the hairs at the relevant place are not causing neurons to fire.
What are the limits of time resolution of human hearing (gap detection etc- there is no single measurement, and no not interaural Mr Kunchur) Is it measured in picoseocnds? microseconds?

Does human hearing deteriorate in terms of time resolution just as it does in terms of frequency resolution?
The person I would like to ask abiout this is Brian Moore the author of the leading text (at least in the UK) Introduction to the Psychology of hearing.

Now lets turn to the question of what neurophysiology tell us about the reliability of the human/ear brain as a sound quality detection device in sighted tests?

And why is it that in Archimago's blind distributed file tests it seems that people did not prefer MQA. (just as they did not prefer minimum phase filters and did not prefer flac over Mp3).

Gee does psychoacoustics tell us anything about the spooky disparity between what people actually prefer and the strange tendency of people writing articles to claim that they prefer the new shiny thing in a sighted test?

Now where were we?

JimAustin's picture

Lots of good (and open) questions here, to be addressed in future articles.

volvic's picture

I have no horse in this MQA race, no interest in streaming or subscribing to Tidal. With hundreds of CD's and vinyl waiting to be listened, I have no time to spend on other sources, Fricsay and Furtwangler sound just fine through my vinyl rigs. But I do appreciate the interesting articles and responses. In a way some of the comments and reviews from the press and readers, Stereophile included, remind me of the early days of CD which had its detractors and supporters. In the end sometimes even the best technology doesn't always win out, the best I can say is that the market and manufacturers will decide whether or not MQA succeeds. I appreciate reading your responses as well as readers' comments....enlightening.

Fokus's picture
Quote:

If you know the impulse response of a system, you know how it will respond to any input. Or, as Wikipedia puts it, "Since the impulse function contains all frequencies, the impulse response defines the response of a linear time-invariant system for all frequencies."

So what's the problem?

An impulse introduced to a sampled-data system as a sequence of zeroes, then full scale, then again zeroes, is in fact an aliased signal. And once a sampled-data system aliases, it is no longer linear and time-invariant.

You can learn a lot from impulses in digital audio systems, up to a limit, and if you know how to look at them. But ultimately, these signals violate the sampling theorem's very first sentence...

seldomheard's picture

Same old salesmen hiding behind stupid fake names making comments to support their so called "technical articles". The arguments are so ridiculous, even a total novice idiot can tell you that if your loudspeaker's response starts to roll off dramatically above 18khz, there's not a lot of reason to buy yet another black box that generates flat signal response out to 50 or 100 khz - putting aside the FACT that essentially nobody can hear a darn thing above 20khz.

spacehound's picture

If you look at the specs of most DACs, including the much vaunted dCS ones, you will find that the analog output rolls off rapidly around 20 KHz regardless of the sample rate.

So even if there was 'music' information beyond 20KHz in the original signal, it doesn't even get to the DAC's output sockets, let alone the speakers or your 'golden ears' :)

JimAustin's picture

"Give 'em enough rope?" No, not the Tubes song. Anyway, this post is where you did that.* It's completely, and obviously, wrong.

Below is a plot of the output of my preamp--DAC first, then preamp--with, as input, a 96kHz (sampling rate) flat frequency sweep. The sweep, then, goes to 48kHz. This is the DAC output, via the preamp (so you can see that the preamp preserves the signal, too).

* hang yourself

Frequency Response DAC

or maybe you just need a new DAC?

spacehound's picture

I didn't say ALL dacs, I said "many".
And what I said about their frequency responses is FACT - it's in their own MEASURED and published specifications.

Some examples -
PS Audio Direct Stream dac (a Stereophile 'product of the year')
All other PS Audio dacs.
All dCS dacs (prices up to $135,000).
All Chord dacs (prices up to $20,000).
All Cambridge Audio dacs (prices up to $1,500)

And there are MANY others. In fact dacs that roll off rapidly at about 20 KHz are greatly in the majority.

So YES - nothing much above 20KHz comes out of the majority of dacs regardless of sample rates (and prices :)).

JimAustin's picture

You're definitely wrong about the PS Audio; I own one. I'll post its output spectrum later today. I suspect you're wrong about all the rest, but I don't have them on hand to test.

spacehound's picture

Are you saying their own measurements are incorrect?

If they can't even measure their own stuff I'm surprised it does anything at all, except maybe blow up.

I own examples of the other three I mentioned, though none of the ones I own are their top model. But they are all 'current' not obsolete ones.
Being retired I have no way of measuring any of this.

Cambridge and Chord actually show frequency response charts, and the drop after 20KHz is quite rapid.

dCS merely said it drops off "considerably" above that so they don't think it's worth bothering with.
BTW: Their background is in military digital communication and 'studio' equipment, including the BBC, so they do have a clue.

JimAustin's picture

I think you may be misinterpreting the specifications. Stereophile has measured several of these products. Here are the frequency response graphs.

Chord DAVE:

PS Audio Directstream:

dCS Rossini:

Meaning no disrespect, if you need help interpreting the measurements, let me know.

JimAustin's picture

Isn't it interesting, then, that so many people get bent out of shape over the fact that MQA is lossy above 48kHz.

Cheers,
Jim

seldomheard's picture

I couldn't say how many people are bent out of shape in connection with anything that has to do with this latest digital gizmo nothingburger. Perhaps to [flame deleted] an extraordinarily tiny niche market segment might actually contain "so many" people that are indeed bent out of shape. What's Stereophile's circulation these days? Would that number qualify as "so many" to someone like Sandy Gross?

John Atkinson's picture
seldomheard wrote:
What's Stereophile's circulation these days?

As you can read in my "As We See It" essay this month, I have been criticized for stating that Stereophile is the "numero uno in hifi mags." But this happens to be true. Stereophile is the highest-circulation English-language magazine devoted to high-end audio reproduction in the world. Its circulation has remained consistent around 71,000 for several years now, and to put that number into context, it is greater than the combined circulations of The Absolute Sound, HiFi News, and HiFi+.

Stereophile's website has also more unique visitors and page views each month than other high-end audio magazine websites and webzines.

seldomheard wrote:
Would that number qualify as "so many" to someone like Sandy Gross?

From my discussions with him, I certainly believe so.

John Atkinson
Editor, Stereophile

spacehound's picture

It is a demonstration that WILFULLY technically illiterate and 'anti-science' people are easily suckered.

As I said, it doesn't come out of most dacs and also adult humans over about 45 years old don't hear anything above 14-15 KHz anyway. Not even 'golden eared' humans.

What it of course means is that:
ANYTHING THAT COMES OUT OF MQA ABOVE 20KHz IS TOTALLY POINTLESS. (That's true of non-MQA stuff as well of course :))

Les's picture

Some relevant info on physics of sound here (i.e. Fourier Transform):
https://www.youtube.com/watch?v=izqaWyZsEtY

It will shed some light on the validity of impulse testing.

btostenson's picture

If they cant get Amazon, Apple, and Spotify on board, does it really matter? It will go the way of DVD-A...nice but irrelevant.

JimAustin's picture

As Yogi Berra said, it's tough to make predictions, especially about the future.

I don't have any idea what the future holds for MQA. I do know that they've got more traction than most people thought they would, with, apparently, the bulk of all recorded music destined for the MQA format. Obviously the record companies think there's something to it, and they own the rights to most of the music.

But like I said, I don't know what the future holds. I just think it's our responsibility to consider the issues MQA raises.

Best,
Jim

btostenson's picture

It seems that the most valuable technology to the streaming services and users are the algorithms that compile playlists based upon user preferences. Audio quality from these services seems to be adequate to the majority of users.

JimAustin's picture

https://thebaffler.com/salvos/the-problem-with-muzak-pelly?src=longreads

btostenson's picture

And Apple is buying Shazam and rumored to become streaming only in 2019. I can’t see them using a format that they don’t own.

JimAustin's picture

Yeah, I think I saw "valuable" in your post and thought, "from a music-listener's perspective." Yeah, those things do appear to be the most valuable part of a not-very-profitable industry.

I've got no idea what will happen. MQA has some traction. It may not be enough. It's still an uphill climb. We'll see.

spacehound's picture

Spotify is useful for finding music that is entirely new to you but that's about all.
Finding stuff that you actually enjoy enough to listen to more than once can be a very long and tedious process :)

Brian Lucey's picture

Not a news flash to you I hope?

That British accent and soft tone of voice fools a lot of laymen also. Mostly, MQA is going after the low hanging fruit. Because there is no there there. Just a lossy codec wrapped in marketing and patents with manipulative language.

David Harper's picture

so, if I hear no SQ difference between a well mastered CD and high-res, would this mean that I can confidently write off and forget MQA, as far as my own listening is concerned?

JimAustin's picture

Mostly--although I do note "well-mastered" in your question. Many MQA recordings appear to be derived from superior sources--superior "masters". In those cases--and this is quite common--the sound quality is obviously better, with little or nothing to do with the MQA format. It's possible, even likely, that you would find those difference worthwhile.

On the other hand, those differences are MOSTLY apparent even in undecoded MQA files--so you could enjoy this aspect of MQA's benefit without investing in an MQA-enabled DAC.

Best,
Jim

seldomheard's picture

Snake oil and its purveyors pretty much killed the recording and playback hobby. "Audiophiles" are essentially extinct - displaced by an angry mob of ignorant, obnoxious salesmen who neither individually nor as a group possess the slightest hint of a sense of humor, self awareness, or humility. It's no wonder there are usually more exhibitor personnel than public attendees at the various audio shows around the country. The snake oil illegitimacy hasn't quite invaded CES yet, though. If and when it does, I'm sure that venue and market segment will suffer the same fate.

JimAustin's picture

"Audiophiles" are essentially extinct - displaced by an angry mob of ignorant, obnoxious salesmen ...

That's as good a description as I've seen of the angry mob of MQA-bashers on social media. I'm not angry or ignorant, and I don't think I'm obnoxious. I'm just a guy trying to get to the bottom of things.

adamdea's picture

Jim
If you are a guy trying to get to the bottom of things, [and I speak as a guy who has probably wasted £100k of his otherwise billable time reading books on LTI systems, fourier analyis and digital signal processing, just trying to get to the bottom of this] then the easy way to show it would be by asking some difficult questions of industry figures.
You can't just rehash MQA press releases. Ask why any of this matters. ASk for real evidence (not OOhashi). Ask JJ.

JimAustin's picture

If you are a guy trying to get to the bottom of things, [and I speak as a guy who has probably wasted £100k of his otherwise billable time reading books on LTI systems, fourier analyis and digital signal processing, just trying to get to the bottom of this] then the easy way to show it would be by asking some difficult questions of industry figures.

I'm doing plenty of that. And I don't think I'm rehashing MQA press releases.

seldomheard's picture

My comment wasn't directed at you - rather a poster who stated in another related thread to this one that I should STFU. He's one of the main supporters posting here besides you. Regardless, snake oil salesmen (they know who they are) have for years thrown out ridiculous subjective, unproven assessments of various products "reviewed" here and elsewhere. They hate it when someone challenges them to produce "provable" evidence supporting their findings. They get "nasty" and wonder why the hobby has dwindled and practically died. There are educated people that read these pages from time to time. Instead of insulting them with nasty rude comments, perhaps it would be better to say "sorry, I don't know or I don't have the kind of proof you're looking for. Listen for yourself when you get the chance and see if you hear what I hear".
Tolerance and civility is a two way street. Give a little and you might get a little in return.

JimAustin's picture

Sorry I misunderstood. Anyway, https://www.stereophile.com/asweseeit/505awsi/index.html

spacehound's picture

They are just ****** off by constant MQA shilling.

And Brian Lucey is correct.
1) If it is NOT 'signed off' by the studio it isn't what the artists/studio intends the public to hear.
2) If it has been 'changed' in any way it has been 'downgraded' as High Fidelity means 'as accurate to the original as possible' by definition.

But the MQA people went off and messed with his recordings anyway, so they are therefore claimed as 'authenticated' even though they NOT.
Further, he actually TOLD them he did not want his recordings to go through the MQA process.

Remember:
High Fidelity is NOT subjective.

John Atkinson's picture
spacehound wrote:
Brian Lucey is correct.

1) If it is NOT 'signed off' by the studio it isn't what the artists/studio intends the public to hear.

This is not quite correct. I have produced, engineered, mastered, and played on more 40 commercially released recordings. While the input of studio engineers, mixers, and mastering engineers is always welcome and important, it is primarily the producer who has ultimate responsibility, working closely with the artist whose name is on the project. A mastering engineer doesn't have final sign off any more than the musicians booked for the sessions.

If the producer (who represents the label's involvement) and the artist sign off on an MQA-encoded release, then all is kosher, surely.

John Atkinson
Editor, Stereophile

music or sound's picture

It is sad to hear it is not as the artist intended but as the label(producer) intended but that is no surprise for me as I see a lot music sounds like homogenized industrial products. (of course there are labels which behave differently and respect artists)

John Atkinson's picture
music or sound wrote:
It is sad to hear it is not as the artist intended but as the label(producer) intended...

With respect, you are being too cynical. Yes, as the label is generally financing the recording project, they have the final sign-off. But the producer's role is to balance the label's interests with the needs of the artist and the engineering team. The goal is for all involved to be satisfied with the end product.

Of all the hats I have worn during a 47-year involvement with making records, the producer's is by far the most difficult. Yes, you have to have a deep knowledge of both music and audio engineering, but you also have to be a politician, a psychiatrist, a shoulder for someone to cry on, and an unforgiving taskmaster.

John Atkinson
Editor, Stereophile

spacehound's picture

And it is claimed by MQA that Warner has "MQA'd" its entire catalog.

Many of the producers and artists will have moved, left, retired, or even died. And the living ones aren't going to go back and 'sign off' all that old stuff are they?

So it's BS.

[off-topic text deleted by John Atkinson]

Brian Lucey's picture

So they are sausage grinding them out. This is a 100% integrity break on the sales pitch about authentication and the sound of the mastering studio. Alterations in the digital are never better, it's impossible. The only TRUE MASTER is the one at the native sample rate and bit depth from the mastering session. 24 to 16 is much less damaging than MQA.

Mastering is not a whore for MQA to use to sell, we do real thing with real data and it's being F'd up in bulk processing as we speak to make $$$.

spacehound's picture

We don't get as much snake oil.
We have both a 'Trading Standards Authority' (prevents selling of products) and an 'Advertising Standards Authority' (prevents advertising of products).

Some 'high end' cable suppliers have already fallen foul of them.

I suspect MQA will fall foul of them soon.

JimAustin's picture

We have both a 'Trading Standards Authority' (prevents selling of products) and an 'Advertising Standards Authority' (prevents advertising of products).

Some 'high end' cable suppliers have already fallen foul of them.

As an American who believes in free speech as a core value, I'm not sure how I feel about this. I'm torn. Generally though, I think the record shows (for anyone who cares to follow it) that I'm generally against audiophile bullshit. I've written several essays along those lines. However, I do recall my earliest experiences with so-called objectivity in reviews: Consumer Reports magazine. They purported to be objective, and consistently botched the job. It's an example of what has been called "greedy reductionism". JA has used the word "scientism". It looks like science, it tastes like science, but ultimately it's just closed-minded posturing. As a scientist, I'm skeptical of everything, including excessive skepticism. I've got a book of record reviews published by Consumer Reports. It's hilarious.

My worst Consumer Reports experience: In an article comparing shampoos, they dismissed dandruff shampoos, to which I was practically addicted, with this statement: "Any high quality shampoo should remove dandruff flakes." (Translation: Dandruff shampoo is snake oil.) Wow! I don't need dandruff shampoo! So I ran off to the grocery and bought their highest-rated non-dandruff shampoo. Two days later my shoulders were covered in snow.

spacehound's picture

And most western societies believe in free speech too. Unless your free speech is against 'minorities' or other PC stuff of course :):):)

Trading standards is the easier of the two. It's basically "Is it fit for purpose?"
And a car that breaks down once a month or needs its engine replaced after two years isn't.
And a lamb kebab who's meat content is not 100% lamb isn't a lamb kebab. (Lamb is twice the price of any other meat in the UK and that one was on TV only yesterday. I was very bored at the time is my excuse for watching such stuff :))

Advertising is more complicated. Despite the existence of the Advertiaing Authority "Let the buyer beware" remains the main factor in the UK.
But is advertising genuinely 'speech'? is another difficulty. REMEMBER, a business corporation is NOT legally a 'person' so may not be able to say things an individual can.

Anyway:
MQA
Does it do what it claims? Basically yes.
But:
It slightly attenuates higher frequencies that are below 20KHz so nominally in the 'audible' area. These have to be boosted back in a process that is not exact.
In 'deblurring', (which it does actually do), it causes artefacts within the 0-20KHz range.

So it WILL be different from the original. Some make 'like' this difference, some may not, and some won't notice the difference. Which explains the varying opinions.
But either way, they are not hearing the 'real thing'.

JimAustin's picture

REMEMBER, a business corporation is NOT legally a 'person' so may not be able to say things an individual can.

Regrettably--not true in the U.S. anymore, following some unfortunate Supreme Court decisions. This is why corporations can now give unlimited money to candidates in electoral campaigns. Money is speech, and corporations have the same rights as people. Apparently.

spacehound's picture

The MQA 'objectors' are NOT "salesmen".

The MQA salesmen are the MQA employees, selling to the record labels. And hopefully to 'us'. Because if 'us' are not interested the record labels/Apple/Spotify/etc. will not buy into MQA.

All the rest of us, both pro and anti MQA, are buyers. Or not.

fetuso's picture

"The primary way people listen to music shifted dramatically in 2016. Streamimg, and not downloads of physical album sales, is now king..."

That's a quote from Nate Rau of USA Today.

Isn't that the bottom line in all this? The record companies, which want to continue monetizing their catalogues (fair enough), have jumped on MQA because it decreases file streaming size. My personal opinion is that we're being sold MQA as a sound quality improvement, when in fact the real reason for its existence is to boost streaming revenue. There has already been a lot of marketing money spent on "high res" and MQA is just another tool to get people to buy into the high res hype. The current rush of harware companies adding MQA reminds me of the AV Receiver spec wars. Everyone just wants to check that box for marketing purposes, not because they believe in it.

spacehound's picture

But MQA is a less efficient compression method than 'straight' FLAC so the files are in fact larger.

Why?
Because the 'folding' idea compresses the data but not as efficiently as FLAC.
Then they compress the resulting 'lossy' file with FLAC and that's what you receive.
But FLAC doesn't expect ever to see 'lossy' files so doesn't compress them efficiently. The result is a file larger than a FLAC compressed WAV file of the same original sample rate.

So basically, although MQA can be said to 'work as designed' the result is not the same as the original, that degraded result is also lossy, and larger than a FLAC compressed WAV.

So it's all a bit pointless for the end user. It makes everything worse.

Which should show you quite clearly who's side this magazine is on. And it's the same for many of the other magazines too. We all know why. If magazines of any sort depended on the money from people buying them they would go bust.

fetuso's picture

I freely confess to only having a very basic understanding of the physics involved in all this, but i do have common sense. That leads me to ask why record companies are all over this? The answer to me is that they feel they can make money this way. I know they don't prioritize sound quality, so it can't be that. I don't feel like i have a stake in this, anyway, because i don't stream. I had a Tidal subscription for a couple of months, but suffered from paralysis by analysis...too much choice. I did, however, get to sample some MQA tracks via my AQ Dragonfly Red and Oppo PM3 phones and i honestly couldn't tell a difference compared to the same tracks (maybe not the same masters) i had stored on my hard drive in 16/44 flac.

spacehound's picture

....see it as a chance to sell much the same stuff over again. That's why so much of it gets 'remastered' - it's not just an ego trip by some nonentity who thinks he can do better than (for example) the original Beatles records, though that IS part of it.

Tidal. Yes. I tried the same with Spotify. I like 'classical' music but other than Bach, Mozart, Handel, and the rest of the well-known crowd I don't know much about it.
I found it immensely tedious wading though all sorts of stuff I have never previously heard of so I have given up.

MQA? The difference can't be immediately obvious because if it was there would not be so much arguing about it :)
And if it isn't obvious then it isn't worth the bother.
BUT - all those 'machinations, making it lossy, getting added noise as a side effect, etc. CAN'T make it more accurate, which is the very definition of 'HiFi' and the reason why some of us spend all that money on equipment.

Some people may prefer the result of those 'machinations' but it AIN'T an improvement, it's just a personal preference, no different in principle from turning up the 'tone' knob on old radios :)

ok's picture

I’m not an MQA proponent (nor do I feel it as a personal threat either) but I consider unfair some folks’ attempts to pose existential questions regarding certain measurements – or, worse, to dispense with them altogether – in case they don’t like the outcome. One could always debunk any measurement after all. No need for wide bandwidth, since only bats can hear it; no need for low electronic distortion, since transducers distort in spades; no need for high damping factor, since cable/speaker systems pose a bottleneck; and so on. It’s true that measurements don’t always deliver what we’d like to expect, but they are the only ones we’ve got for the time being. Moreover extremely good figures admittedly come at a price, sometimes revealed by other measurements, sometimes hidden from sight (and detected by ear), sometimes purely financial. But numbers and graphs correctly applied do show something after all, even if in many a case no one knows what exactly that is.

JimAustin's picture

I think this is a pretty wise take. My goal is 1. to figure out whether MQA's thinking makes sense, and 2. determine whether MQA does what it says it does. The final step is to figure out whether, and how much, it matters. But as you say, unanticipated consequences should be considered, too.

spacehound's picture

The base problem is that we can never know how it is SUPPOSED to sound. So we have no reference. Which makes our ears totally useless so "trust your ears" is a nonsense. Always. It just displays your personal preference and has ZERO to do with 'goodness'.

So we measure.
We measure everything we know how to measure as accurately as we can.

If it all measures 'good' (for example a flat frequency response is one 'good thing') we can reasonably assume we ARE hearing what the studio/record label intended even though we lack any reference.

Does MQA make sense?
Basically yes.
But:
It introduces audible artefacts (almost an MQA 'signature') that the above measurements will demonstrate will NOT have been in the NON-MQA'd original even though we do not have that original.

[off-topic text deleted by John Atkinson]

tonykaz's picture

"how it's SUPPOSED to sound" ( oh dear, is there purpose to that yelling ? ) or is it simply a rant ? ( I , for one , rather enjoy a good rant, occasionally )

Of course you are at the very intersection of our dilemma, aren't you ?

How many of us have any sort of knowledge of how any recording should actually sound ?

I'll answer that : the guys that were there when the musicians recorded the performance.

Who are those "experienced" people ?

They include 'our' John Atkinson, don't they ? ( I love thinking that he's ours now )

We can also include 'our' Bob Katz and his little group of Recording Professionals ( Bob is a sometimes contributor to Stereophile Mag. ) and frequent contributor to all-things Pro-Audio.

But, "Live" music doesn't ever sound quite-like the stuff coming out of our hifi gear. Should it ?

Summing things up : MQA is just another compression system , can't we simply accept it for what it is ? and then, move onto important stuff like Cramolin treating our connectors and German Audiophile Fuses.

21st Century Tony in Michigan

ps. Meridian is everywhere you look, in the USA. I guess they go where the money is. ( just kidding )

Glotz's picture

And Spacehound- Let it Goooooooo... We get it, you hate and always will hate Meridian. Horrible lot, all of them!

What about the very question of what is fidelity and MQA? Strangely enough, JA put up the Absolute Phase article from way back when, just a few days ago. It's a great analogical piece to contemplate MQA and its complexities.

And firmly remind everyone in an uproar, MQA is not going to kill you. (I wish it would sometimes... lol.) It really is the political meme atmosphere that has infected every dang person on this planet and MQA's hate is proof.

Bob Stuart for President!! Lmao...

spacehound's picture

BIG LETTERS because I don't know how to do bold or italic. Or smillies :)

How's it supposed to sound? We can never know. So all we can do is buy equipment that from its measured responses (one example of which is frequency response, but of course there are others) doesn't mess with the 'signal' its getting.

And for sure the original performers didn't intend for it to have audible aliasing, increased noise, and parts of it being left out.

All of which happens with MQA.

tonykaz's picture

I'm just trying to have some fun outa this MQA controversy ( I'll admit to not having any sort of dog in the fight other than once being a Meridian Importer, Owner, Lover, Admirer and probably future owner of a new DSP Meridian system ). I can admit that I never had any level of success in selling Meridian stuff, leaving me with the feeling that; Meridian is Great Dog Food that the dogs won't eat ! ( at least from my 1985 hand )

Is there any Consumer Audio Company with an equally long commitment to digital formats ? I'll answer that : Yes, PS Audio.

I also Imported PS Audio Gear ( from California to Michigan ) in the early 1980s, before they kinda went bust. I loved both Paul McGowan and Stan Warren !

I went bust too !

I returned to the Transportation Industry, where people can earn a living and keep wondering if anyone can cut-out an existence in the Consumer Audio segment selling to those/us "hair-shirt", "lunatic fringe", vinyl based Audiophiles ?

Well, I just learned that B.Stuart followed my path to Jaguar and makes significant money from them. Hmm........

I like hanging out at Stereophile, a place to learn about the gear that is pleasant to listen to Annnnnnnd find kindred spirits who value sharing reasonable thoughts about our shared love of recorded music. ( and blind testing various Wire s )

Stereophile is run by a wonderful Professor type person worthy of our Trusting instincts and has a fascinating character writer ( much like Donald Sutherland ) in Herb Reichert ( a must read ).

I have every reason to say & think that we are in "Good Hands" here.

21st Century Tony in Michigan ( the home of Rock & Roll )

ps. I'm a lifelong Classical that just discovered Elton John. I ordered all his albums which I'm told have arrived and are ready for delivery on Friday. An Old dog like me still learns tricks.

spacehound's picture

...stuff or all sorts of reasons.
Elton John is quite good but I don't buy his records because I don't like his sexual proclivities and the blatant way he proclaims them. Also because PC says I should not think or say what I just posted.

I won't join Tidal because I think Jay Z is a jerk and rap is not music at all, just a sort of 'protest'. They don't even have to sing, just talk fast like a horse race commentator. So I refuse to support anything even remotely connected with rap, which is the ultimate in 'dumbing down' something which should require ability.

I think Stuart is a con man. And he's clueless about how to design audio, it's just good looking 'designer' cases. And his Jaguar/Land Rover audio is mediocre, as is everything he has ever produced.

I buy Harley-Davidsons because they are made by 'people like us' who are actually paid real wages, the same as I am, and they appear to show interest in what they make.

I buy Naim audio because way back in 1984 they had an open day at their factory (which is near where I live) and I was impressed. They provided free beer too. But I don't buy their DACs.

I bought a Technics SL1200 simply because it wasn't a Linn. Nor was it some 'specialist' junk made by two guys in a small garden shed.

I like the modern Quad electrostatic speakers but haven't got the nerve to actually put my money where my mouth is on those.

Cheers :):)

tonykaz's picture

Geez,

I had a quick 'Google Images', of your Elton John. Phew, what a dresser ! We had a piano guy named Liberaci, my mother loved, that was a 1950s version.

I can't understand Rap but I do understand those Black Protest movements annnnnd the Women protest movements. In my Industrial World, people loose what they abuse. Here in the States, those Black & Women movements converted into political horsepower. ( Senatorial Level Horsepower )

Is Stuart that bad ? , Boothroyd did the designs, I'd thought.

Around here, Harley riders will say that they'd let their sister work in a whore house before they'd let her ride a "rice burner". The United States is where Honda makes their motorcycles & cars. The Honda people deserve engineering applause. The same people that build my GM cars build the Honda stuff.

I like the Naim stuff but never owned any of it. I love the round knob controls and simplicity. If I'm wrong, I don't want to be right. But, I got addicted to Elecrocompaniet and never looked back.

I've heard plenty of praise for that Technics turntable but..... I made my Audio mark by Importing every turntable England made. I especially liked the "Pink" Triangle design. ( I didn't understand the symbolism of the name )

The Quad 57 original is still turning audiophile heads, I've owned the ESL 63 ( long gone now and forgotten ). I'm a dynamic driver person now-a-days so I'll pass on all those exotics.

Well, I appreciate your choices and reasons.

Tony in snowy Michigan

spacehound's picture

is all very well but why to they have to pretend it's music? Which rap certainly isn't - it's just bad poetry with a supposedly 'musical' background noise.

I've had Harleys, BMWs, Ducatis, Guzzis, and a 1970's MV Agusta, one of the 'real' ones, a Magni tuned and chain drive converted 750 'America', not the modern ones, which aren't related to those at all. (I foolishly sold it and it's now worth at least 10 times what I sold it for).
To me Japanese bikes are all the same, just like small hatchback cars.

Electrocompaniet appear for a year or so on the UK market and then vanish for several years. Then they reappear and vanish again. So I've never paid them any attention.

I like the SL1200 simply because it ISN'T one of the 'classical' British turntables. Which only use belt drive because they don't have the competence or the equipment to make anything more advanced. So they 'knock' Technics and try and fool people that belt drive is better. And they are more 'exclusive, and we know what many 'HiFi enthusiasts are like, don't we?. If what they distainfully call the 'masses' are aware of it they won't buy it, totally forgetting that we are ALL the 'masses' 90% of the time :):)

'Pink' Triangle? It was simply because the two guys who owned it had the same 'sexual orientation' as Elton John :)

tonykaz's picture

I have no idea about Rap, I certainly don't consume any of the stuff, I'm not into the head banging stuff either. But, the people that produce Rap are vastly successful ( enviously ? ) .

Transportation stuff is my World.

We, ( GM ) have a proving grounds in Milford, Michigan where we engineering types have a good close look at those Japanese products we compete with. I'll report that any Honda or Toyota ( in the 1995 era ) would have higher mechanical integrity ( at the 100,000 mile mark ) than a brand New GM Car. Phew. I'm told that the Japanese Motorcycles are equally impressive. Now-a-days, the typical Japanese Car will still be getting licensed as it clocks over the 300,000 mile mark. Our GM Cars will be recycled around the 150,000 mile mark. Our F150 Pick-up Truck types will go 500,000 miles.

None of us GM guys were surprised by Top Gear's inability to destroy a little Toyota Pick-up Truck ( the one the 3rd World Armies use ). We tried to find out what it would take to knock-out 'any' Toyota.

I grew up around motorcycles ( I'm from Wisconsin where moms will own some sort of Harley for running around ). In Wisc. Harley's are everyday transportation devices, not Cultural Statements. Every Harley leaks.

I'm a bit taken back by Audiophiles seeming to like Direct Drive Record Players, they seem to be "Digital" turntables, somehow. I shouldn't be bothered, should I ? because I gave up on 2nd Generation Media ( 30 years ago ). Now I'm a committed pursuer of 4th Generation Media technologies. ( like MQA )

21st Century Tony in Michigan

ps. I lived in England for a while, I owned a Daimler 250 V8 in Gold, it was the slooooooooooist Car I've ever owned. I love England and the English. I'm Swede & Rus

music or sound's picture

I think this is a good discussion but I would like get more clarity what are the different aspects of MQA and how related are they (like partial unfold through software). Here part 1 was more or less only about digital filters for the final DA conversion. How essential is that for MQA? What is time deburring? How are MQA claims to time integrity related to mastering i.e. independent of playback? I hope the next installments will explain more the different aspects of MQA and which part of it is "essential"" for what.
In general it would be great to have some new articles form Stereophile about time domain (not only in DA conversion but the whole system including recording systems to the listening room).

JimAustin's picture

There's a lot to cover. We'll do our best. Thanks.

spacehound's picture

....recorded music.
A 'stream' may not be, it might be live or a recording. FM and AM radio are also streams.
MQA is for recordings.

As the originally claimed 'checking of the studio equipment' is not usually happening, What does MQA do?

1) It takes a file whether created on 'MQA checked' equipment or not.
2) It 'folds' it, creating compression, creating losses, aliasing that wasn't in the original, and a small treble loss that has to be boosted, increasing noise.
3) It further compresses it using FLAC. Unfortunately FLAC doesn't work very well with already compressed files so it doesn't compress them very much.

Result:
1) A file larger than a FLAC compressed WAV of the same resolution would be.
2) A file that is lossy.
3) A file that contains aliasing that wasn't there to begin with.
4) A file that contains more noise than the original.

So what's the point?
(We all know the answer - it's to preserve the existence of Meridian and the income of Bob Stuart.)

Note: I'm British. I live in the UK. Unlike many USA citizens we are not trained from birth to buy stuff without questioning it. There are many other cultural differences too. Despite our common language the UK is not a 'little USA'.
What is more, unlike reviewers, audio magazine owners, etc. etc. no part of my income is derived from the 'music' business and its continued existence.

seldomheard's picture

And having looked again at the impulse responses, it's pretty clear now that the author of this "article" didn't have the capacity to realize that what he was proffering as superior was actually inferior. "Pre ringing", as I noted above is a FEATURE of the special, artificially generated impulse test signal. Mr. Austin said this at the outset:

"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.
Read more at https://www.stereophile.com/content/mqa-tested-part-1#8QmR24PRXQcZf2kM.99" -Jim Austin

This tells me he is absolutely clueless about the information someone in industry has been feeding him. His further statements about "deblurring", linear phase, and time coherence are more proof that he is completely over his head on the subject of digital audio and digital to analog conversion.
The only reason the MQA impulse rendering is without the pre and post ringing is likely because the noise floor of the output has been raised. These examples where MQA has "been enabled" in figures 4,5, and 6 actually show a distortion of the original impulse presented in figure 1. THEY ARE INFERIOR. Mr. Austin is so clueless, he seems to think that "cleaning up the ripple" is a good thing - apparently because in his mind, all input signals - not just discontinuity/impulse signals, have pre and post ringing. This "pimposium" has to be one of the most epic fails in audiophile snake oil salesmanship ever. Mr. Austin should stick with studies in optical physics. This is just plain embarrassing.

Fokus's picture
Quote:

The only reason the MQA impulse rendering is without the pre and post ringing is likely because the noise floor of the output has been raised.

Certainly not. Noise has nothing to do with. MQA's intentionally leaky filters are the reason.

I am not defending MQA, nor Austin, but you seem a bit confused, too.

prof's picture

Per John Atkinson's footnote, it would be a great service to your readers, none of whom can do Fourier-transforms in their heads (I certainly can't), to feed the output waveforms of these impulse-response tests through a spectrum analyzer, so that we can see how much distortion/nonlinearity/"leakage" is introduced by these various filters.

kiranps's picture

Dear JA,

Have you noticed this artifact of MQA

https://www.youtube.com/watch?v=7JZm3VppCfQ

Thanks
PSRK.

spacehound's picture

Was shown in an article by Jim Lesurf about a year ago, though he didn't comment on it. But I can't find the article now.

It won't matter, at 21-22 KHz it won't be audible by any member of the human race, no matter how young they are. Nor, of course, will anything above it.

seldomheard's picture

This entire exercise has been an embarrassment - blatant proof that what Brian Lucey said is absolutely, spot on correct - a few record company and "audiophile" industry insiders are pimping useless technology that degrades sound quality of digitally rendered music. Mr. Atkinson should think long and hard about allowing this farce to continue with more "lectures" by an individual who is so clueless, I feel embarrassed for him.

As noted earlier, Phillips solved the problems associated with anti aliasing about 37 years ago - simply digitally interpolate or upsample the signal before converting to final analog and filtering at a more gradual rate. To suggest that the system is now - almost 4 decades later - "broken" is a complete farce. And if Mr. Atkinson chooses to continue this charade - his credibility and the credibility of the publication he works for will be destroyed.

JimAustin's picture

nt

spacehound's picture

We are not stupid either. Most of us are fully aware of why Stereophile and its rivals are 'pushing' MQA.

To continue to exist magazines HAVE to 'keep the pot boiling' or interest (and magazine purchases) will decline and advertising revenue will decline correspondingly.
That is true of all 'specialist' magazines and commercial websites, it's not just Stereophile and it's not just audio.

It's not 'wrong' and it's not 'actively dishonest'. But we all need to keep it in mind

JimAustin's picture

against what I perceive to be ill-considered (and excessively personal) criticisms, but it would be a mistake to consider me an MQA advocate. Neither JA nor anyone else has pressured me in any way to advocate any particular point of view. You'll notice that when people have made GOOD, well-informed critical comments (or raised good, well-informed questions) in this discussion, I have acknowledged that.

My goal is not to stir up controversy, 'though I realize that's inevitable. My goal is to get to the bottom of thing. We'll see where that takes me.

jca

Glotz's picture

and keep PUSHING!

MQA deserves a hard look and discussion.

While it may seem like forced marketing to some, no one screamed bloody murder like this when HDCD hit players 20 plus years ago.

spacehound's picture

You were not even MENTIONED in my post.

What is more, you are far from the only one who has written about MQA in Stereophile.

But like it or not, that is how 'special interest' magazines/commercial websites, such as Stereophile itself and this web 'offshoot' survive nowadays - on advertising revenue.

And to keep the readers interested enough to buy the magazine or look at any 'supporting' website AND SO SEE THE ADVERTS there have to be articles.
And 'new developments' are a good source of these articles. And MQA has caused sufficient controversy that the editor and/or the owners can keep it going for a long time - look how many 'clicks' your one article has generated - the advertisers (or their agencies) are aware of these numbers, though probably in total, not for any one article.

All of which is PRECISELY why you were asked to write this series of articles, be they favourably to MQA or not - that doesn't matter to any advertisers except the MQA company, who won't be a big advertiser anyway.

Surely you are 'business aware' enough to see all that?

JimAustin's picture

When I wrote "personal", 1. I didn't mean me, or not mostly. I was thinking mainly of those who have suggested that Stuart is some kind of charlatan--although, yeah, there have been plenty of personal attacks on me. But 2. I definitely wasn't suggesting that your post, the one I replied to, was a personal criticism of me.

As for the rest, I'm quite comfortable with the old-fashioned arrangement among readers, publishers, writers, and advertisers. My job is to write interesting, factually correct stuff that people will read and enjoy. If I do that job reasonably well, the other three groups all benefit: readers get informed, advertisers get exposure (from the accompanying ads), and the magazine makes money. I'm OK with that.

jca

spacehound's picture

I'm ok with it too.

But I think that the 'naive' should be aware of the 'ins and outs' of the magazine business.

Not that I feel I have any duty to 'save' them. My interest is in the 'difficult' questions that the MQA people avoid answering.

As for Bob Stuart, as I have said, I am fully aware of his entire history in the audio business. As a result the equipment sold by the organisation(s) he manages would never be my first choice :):)

Fokus's picture
Quote:

Philips solved the problems associated with anti aliasing about 37 years ago - simply digitally interpolate or upsample the signal before converting to final analog and filtering at a more gradual rate.

And again wrong. Moving to an oversampled system does not relax the demands on the filter at all. What it really does is to allow you to move the bulk of the filter action to the digital domain, where high orders, linear phase, and accuracy are infinitely cheaper than in the analogue domain.

David Hyman's picture

are we supposed to hear what the mastering engineer intends us to hear or something we hope is better? if the mastering engineer is adjusting his dials to account for ringing or ANYTHING else he/she hears, and we then use mqa to adjust for some ringing, didn't the mastering engineer already account for that in his work? doesn't that just make MQA a crap shoot? who gets to play audio god here? me, the mastering engineer or MQA?

Pages

X