Saturday, 15 April 2017

COMPARISON: MQA "Authentication" & Sound Quality? (Mytek Brooklyn & Meridian Explorer2)


You've probably heard or read the catch phrases from MQA over the years. "Revolutionary", "TAKE ME THERE"... "To the original performance..." Or how about using phrases like "end-to-end technology"?

As I have said over the last few months, I don't like talking about MQA based on my general impression of what they're trying to do and the way they try to convey supposed "value" to the audiophile world through their advertisements and sponsored articles in the audiophile press. Nonetheless, sometimes it's just necessary to comment and more importantly to put some of the rhetoric to the test. There appears to be a remarkable schism between those who advocate and praise MQA and those who have concerns. I'm pretty sure there are many wishing that MQA would just go away instead of complicating music playback with yet another questionable variant.

Last week, when I wrote about the idea of MQA CD, I brought up the Pono experience as another example of failure in the recent history of the industry. For Pono, the failure was perhaps rather obvious for those of us who have been listening to 24-bit and >44.1kHz music for awhile, especially those of us who have ever bothered to try an A/B-test. It does not take a genius to realise that audible differences are really quite subtle (if even there in most cases of mainstream music) and that differences do not translate to "benefit". Without clear audible benefits, there really was no way that the promise of the Pono music store could ever excite the music-buying public... Certainly not in the way Neil Young portrayed it (sure, the hardware PonoPlayer is unique but with its own quirks of course).

Part I: Intro / Rationale

When it comes to MQA, it's certainly a bit more complicated. For one, it brings into this world a new encoding system so it's not as easy to compare unless one had a decoder that could easily be switched on/off on-the-fly. But for me at least, right from the start, there were many ideas being floated that just seemed "fishy". How is it possible to honestly say this:


When you're packaging supposedly PCM 24/88+ worth of data into a 24/48 file that maintains compatibility with standard DACs? Obviously some form of lossy mechanism must be involved - is that not a form of sacrifice? And obviously in order to add the encoded data within the package to maintain standard playback compatibility, some potential lower level details will need to be discarded. I agree, it's not unreasonable in a 24-bit file because the lower bits are typically just noise, but still, a potential "sacrifice" of sorts depending on how many bits were affected - the higher the resolution of the original production, the more potential loss of low-level detail. In previous posts, I discussed digitally in software and using the Mytek Brooklyn DAC the effects of MQA decoding.

But there was one idea that always bothered me about the MQA claim worth thinking about... What does it mean above when MQA insists "And because it's fully authenticated, you can be sure you're hearing exactly what the artist approved in the studio"?

What exactly does it mean when they use the word authenticated? And how is this authentication related to the sound "in the studio"?

Notice that MQA is rather careful in the words they use on their website especially related to that second question linking sound quality. However, if you look at the older archives, we see Bob Stuart making a few assertions worth considering and the connotations they create. Here's one from a CEPro article in December 2014 when the press was first introduced to MQA at a company party:
Stuart says that with the development of MQA, listeners can experience their digital audio content at the same quality levels as the professional recording industry. 
“Music lovers need no longer be shortchanged; finally we can all hear exactly what the musicians recorded,” he explains.“MQA gives a clear, accurate and authentic path from the recording studio all the way to any listening environment—at home, in the car or on the go. And we didn’t sacrifice convenience.”

By around CES in January 2015, there was a Q&A posted on Meridian Unplugged and one of the answers went like this from Mr. Stuart:
1. The MQA syntax supports a hierarchy of authentication keys using strong encryption. The encryption protects the encoding/decoding instructions, various metadata and verification of both lossless digital transmission from studio to decoder and 'beyond digital lossless', it authenticates the analogue-to-analogue path -- which is a major step forward in sound quality.
And later in May 2016, we see this on AudioStream, in the "MQA Reviewed" article:
The MQA authentication ‘light’ indicates Provenance in the source for the file. The MQA display indicates that the unit is decoding and playing an MQA stream or file and denotes provenance that the sound is identical to that of the source material. MQA Studio indicates it is playing a file which has either been approved in the studio by the artist/producer or has been verified by the copyright owner.
And more recently here's a January 2017 video about MQA with various engineers that seem to be hyping this whole "assurance" and "authenticated" bit. For example, what does Morten Lindberg mean by "we have an assurance of how it's going to play back on the other end - that's something we haven't had before":



I added the bold highlights. These are just three examples of the wording being used by MQA/Bob Stuart to suggest that somehow the MQA process retains the "sound" of the studio production. As if there is something special in these MQA files beyond what a standard HDtracks high-resolution download would be able to convey for example. Notice again the careful wording used.

The word authentication basically means "the process or action of proving or showing something to be true, genuine, or valid" - nothing more. In this regard, the little "light" on an MQA DAC would do the job and may be all that Stuart's talking about (whoopee do!). It's not hard to imagine how this is done of course - embed some MQA data in the stream and the firmware detects it. Perhaps something like a CRC may be embedded every so often to make sure the data isn't corrupt and the indicator light will go off or blink perhaps with each instance of corruption.

The words Stuart chooses coyly hints at another meaning; the idea that MQA authentication can perhaps bring the studio sound home when you play an MQA-encoded track. Perhaps hinting that somehow the technology is capable of making DACs sound closer to some kind of studio sonic standard. How are we supposed to interpret "which is a major step forward in sound quality"? Is there any evidence to this? When they speak of "end-to-end technology" ("analogue-to-analogue path"), is there some actual retained "authentic" sound that's different than say an equivalent HDtrack file's playback using that DAC?

One way to test this hypothesis is to check what happens to the sound coming out of two MQA-certified DAC devices. Is there any evidence that MQA processing with its claimed benefits (time domain accuracy, "better" filtering algorithm catering to the DAC, "accurate and authentic path", etc.) will result in sonic output more similar than just the DACs playing the same HDtracks tune? The idea being that if MQA processing makes the DACs sound more alike, then there's actually something being done to standardize the analogue output which ostensibly is a reflection of the "studio sound" or even the sound of the "original performance" when they say stuff like this on their website:



Part II: The Test

With the above in mind, my friend with the professional quality RME Fireface ADC graciously and meticulously recorded the output of two MQA DACs for me as he did with the previous test of hardware decoding from the Mytek Brooklyn (see that post for other details like procedure and settings used). This time, the comparison is between the Mytek Brooklyn DAC and Meridian's own Explorer2 with MQA firmware. With each DAC playing either the HDtracks download or streaming off "Master" TIDAL with "Passthrough MQA" for hardware decoding, in total, there were 4 recordings made in 24/192 resolution as below:

As you can imagine, with the different recordings, I can then run objective comparisons to determine how close the MQA decoded analogue output from the DAC is with the HDtracks download for each device. Plus I can compare the sound coming out of the DACs and see if there is a relative difference between whether the DAC was fed standard HDtracks data versus MQA.

Since we don't have an MQA-encoder available, let's try using a popular song for this comparison. We decided on using Led Zeppelin's "Your Time Is Gonna Come" from Led Zeppelin (1969, 2014 24/96 HDtracks, DR9). Note that I had already demonstrated with the track "Good Times Bad Times" previously that the TIDAL MQA stream is essentially the same mastering as the HDtracks download. We purposely picked another track from that album because the peak level for "Your Time Is Gonna Come" is the lowest on the album (-1.7dB peak) which provides some overhead protection to reduce intersample overloading when the DAC performs its usual digital antialiasing filtering whether natively or with the MQA parameters. Furthermore, this album lights up "blue" which means that it's "MQA Studio" authenticated, the highest level of authentication (the lower level has a "green" light).

So then, let's compare... First, comparing the HDtracks 24/96 download with MQA decoding to 24/96. As I did previously, I'll use Audio DiffMaker software to create the difference file from ~15 seconds of the song recorded at 24/192 starting at about 10 seconds into the tune.

Mytek Brooklyn DAC - HDtracks playback vs. MQA hardware decode off TIDAL:


Meridian Explorer2 DAC - HDtracks playback vs. MQA hardware decode off TIDAL:



Click on the waveform composite images to have a look at the details like the frequency spectral display plot and the averaged FFT of the "difference" file. Basically what we see in both the Mytek and Meridian DAC outputs is that indeed there is very little difference between MQA and HDtracks playback. The "Waveform display" is essentially a flat line, the "Frequency Spectral display" looks like low level background noise in the ultrasonic range, and the FFT spectrum shows a few noise peaks down at -80dB or so. As I noted in my previous examination of MQA, this is good I suppose... It basically tells us that MQA is generally able to represent a 24/96 HDtracks file compressed into the 24/48 "bit bucket" (remember, this isn't a true high-res album so any slight loss in noise floor due to MQA encoding below 16-bits or so is not an issue).

Now, let's see if we can do some inter-DAC comparisons!

HDtracks 24/96 playback - difference between Mytek Brooklyn vs. Meridian Explorer2







MQA hardware decoding - difference between Mytek Brooklyn vs. Meridian Explorer2




Whoa! Look at the differences now between the DACs. We see that this is comparatively much different compared to playing back an HDtracks file vs. MQA on the same DAC. No surprise, right? After all, we can't expect a US$2000 Mytek Brooklyn to objectively measure identical to a US$200 Meridian Explorer2. Remember that the Audio DiffMaker program does try to compensate for small sample rate drifts and will also compensate for differences in gain (average amplitudes were only around 0.1dB between samples). Clearly the timing differences in playback are beyond the default level of compensation I'm using. This is likely why you see that Moiré pattern in the spectral frequency plot; the timing between the DACs are slightly different, more than likely undetectable in a listening A/B test but when you overlay them, a "beat frequency" can be detected.

Before anyone freaks out about the magnitude of difference seen here, realize that this is the result of objective analysis using a very sensitive 24-bit ADC capable of detecting very minute frequency response differences, changes in low-level noise, and distortion - capabilities beyond the human ear/mind. Though the difference is measurably obvious, if I were to listen to the two samples in a volume matched A/B test, I would not bet much money on passing a blind test :-). For the record, unblinded on my main sound system (TEAC UD-501 to Emotiva XSP-1 to Emotiva XPA-1L monoblocks to Paradigm Signature S8 speakers), I thought the Explorer2 sounded a little brighter and vocals slightly forward compared to the Brooklyn with this track - which one is "better" can only be judged by the ear/mind of the beholder.

However, what is interesting is that the difference between the DACs playing back either the HDtracks or MQA data has remarkably similar RMS power. This observation tells us that despite whatever customizations MQA is doing for each DAC as it implements its "end-to-end technology", the playback isn't any closer between the Brooklyn and Explorer2 than as if they're playing the HDtracks file. Based on this observation, we can take one more step and try to compare the difference between the MQA vs. HDtrack differences! If the observation is correct, then there should be very little signal left over...



Yup. Very little sonic variance between the DACs between HDtracks and MQA playback. At a total RMS power level in the -60's dB, I can turn my headphone amp up to 100% while playing this and barely hear much of a signal above the noise.

Part III: Conclusion

In this example with an "MQA Studio authenticated" (blue light, supposedly "approved" in studio by the artist/rights holder) playback of Led Zeppelin's "Your Time Is Gonna Come", we see that:

1. The TIDAL MQA stream is clearly based on the HDtracks mastering. Very little difference with total RMS power down at -70dB over 15 seconds of music which is inaudible using both the Mytek Brooklyn and Meridian Explorer2 DACs. This is basically a confirmation of what was found previously but this time using actual hardware decoded analogue output rather than digital extraction from software-decoded TIDAL output.

2. Whether the MQA DACs played an HDtracks file or MQA, the sonic variation between them was about the same. There is no evidence that these two MQA DACs sounded any "more similar" or "more different" to each other when playing straight high-resolution PCM compared to hardware decoded MQA.

Ultimately, I think when we read about the "authentication" function of MQA, all they're saying is that the DAC "light" ensures that the signal is what was expected from the studio and presented for MQA encoding. It's like a CRC embedded in a ZIP file to check for errors. Realise that this is not the same as "bit-perfect" because a decoded MQA 24/96 output will not be exactly bit-perfect with the studio's 24/96 "master" PCM fed into the MQA encoder (the bit-perfect master would likely be the HDtracks 24/96 download). That's all authentication means from what I have seen.

I see no evidence that there is any special "authentic studio sound" that MQA decoding actually preserves or ensures. As you know, MQA's algorithms are proprietary and whatever optimizations they might have included in the firmware to make the Brooklyn and Explorer2 "certified" DACs (beyond money the companies paid to get the firmware done) did not seem to change the nature of the sound in a detectable way with the professional ADC. Putting it another way, there was no special "MQA voicing" found.

As far as I can tell, the more I examine this whole MQA "thing", the less I see evidence of any actual technology that improves overall sonic fidelity. It's an encoding technique that "encapsulates" some data for ultrasonic reconstruction (not really significant IMO) in the lower bits typically in the noise floor, throws in the indicator light for error detection ("authentication"), and uses a sort of upsampling filter. Claims of time-domain accuracy appear meaningless, and there's nothing so far to suggest that it brings us any closer to some concept of sonic standardization. This is not surprising I suppose, last year when I spoke of DSP room correction, I already suggested the meaningless of MQA's claims of "end-to-end" authentication because even if they did somehow ensure that MQA-certified DACs were calibrated to have similar tonal quality and accuracy (perhaps an analogy might be color calibration of TV sets), they could never ensure that the final sound is of a certain fidelity! MQA has no dominion over the effects of speakers and the sound room which are of course where the vast majority of distortions and tonal irregularities arise.

Note that I'm basing these observations on a single track (although "Studio MQA" authenticated). Although I only showed the results from one comparison point, I did have a look at another spot about 1 minute in to find similar results. Perhaps this Led Zeppelin album isn't fully representative of what MQA is capable of... Who knows. Maybe Meridian/MQA could clarify this with some objective results of what "major step forward in sound quality" they're talking about rather than just more words and innuendo. There are many obvious ways they can do this if they honestly want to "walk the talk".

Thanks again to my friend for "virtually" making available his gear and valuable time :-).

---------------------------------

In other news, I was reading this analysis of MQA from SoundStage! the other day. A few comments with quotations from the article as applicable:

1. "One is the claimed lossless compression of files of resolutions as high as 24-bit/768kHz to 24/44.1 or 24/48 (depending on the original sampling frequency), the results being files roughly 50% bigger than a 16/44.1 file."

Yes! Thank you for reminding people that the resulting file is actually bigger than a similar 16/44 FLAC lossless compressed file. Going from 16-bits to 24-bits with compression like FLAC usually results in around 50% increase, but in many cases, these MQA files are minimally compressible in those lower 8-bits where MQA data resides and results in >50% size increase.

2. "Since the process is claimed to be lossless, there should be no reduction in sound quality".

Remember, this is not lossless (ie. bit-perfect) as we normally think of lossless FLAC or ALAC. It's "partially lossy" as I noted last year - that's as certain as the Earth revolving around the Sun. So even though it's "claimed", we need to remember that at best this is "perceptually lossless" assuming that for the piece of music, the change in noise floor due to the embedded data isn't noticeable. We need to end the "claimed" provision and call a spade a spade.

3. Nice to see the audiophile press actually acting more like journalists and expressing independent thought! Kudos. However I must say that criticism of MQA began for me back in January 2015, way before the writer's article in April 2016 when Meridian started acting dodgy with their lack of A/B comparisons at audio shows.

Seriously folks, wasn't it obvious that what they were proposing had a strong ring of improbability to it even with the initial announcement back in late 2014? Are audio technology writers that out of touch with how things work that questions weren't asked right from the beginning? Maybe it's the lack of courage to swim against the tide of other publications? Or perhaps they were all under the influence of Meridian's fancy public relations party (check out this ridiculous video; "bought and paid for" comes to mind) at the The Shard in December 2014?

4. "I can’t confidently say that I’ve heard any differences. As a result, I’m sure that if I did a pseudo-blind test, as John Atkinson did, my results would probably be similar to his -- basically, the equivalent of guessing."

Nice. Again, about time an audiophile writer admits to this given the presence of obvious objective evidence. So what does one make of J. V. Serinus' claim in December 2014 that "I literally laughed at the difference when the MQA version began. Not only did it feel as though a veil had been lifted, with far more color to the sound, but instruments also possessed more body. With more meat on dem bones, I also noticed less of a digital edge on the violin. I've heard Hahn in concert several times, and this was the closest to real I've ever heard her violin sound on recording"? What kind of faith are we to have in subjective impressions like that? Did the company present a fair comparison for him? Does his hearing need to be checked? Was he caught up in the hype and excitement? Ultimately, would or should this kind of testimony have much affect on audiophiles at large?

Well folks, after years of pure subjective "analysis" and opinion-making with little facts, I honestly do hope that the audiophile hobby can come to terms with the importance of being rational. I believe it all has to become more objective and more rational in the days ahead for the sake of credibility compared to other technologically based hobbies. Perhaps we'll talk more about this another time soon...

Hope you're all having a great time enjoying the music! Happy Easter.

19 comments:

  1. I was an audiophile back in the day and recently picked up the hobby again in 2016. I am appalled at the rise of subjectivism and, even more so, the clear lack of engineering knowledge on the part of the audiophile press (I talkin' about you Mr. Fremer; less so w/ Atkinson, who perhaps knows but chooses not to say). It's really quite simple: the more something costs, the more the audiophile press goes ga ga over it. (I actually don't think these people are bought off by the ad revenue--I think they've gone full Stockholm to the extent that even if there were no ads, they'd still believe their nonsense.

    One might ascribe audiophiles' belief in their ability to discern differences where none is there (I can "hear" those ultrasonic frequencies, which don't exist in the source recording, through my mastoid bone), but then what is one to make of the whole cables thing? Let's just say it: people who think a power cord makes a difference in the sound are idiots.

    In conclusion, as your excellent post alludes to but doesn't outright say (I'll say it for you), MQA is essentially DRM by another name.

    ReplyDelete
    Replies
    1. Hey there aonetech,
      I really like the "full Stockholm" phrase :-).

      And yes, no need to mince words. MQA is a form of DRM which not only insults end users, but also pisses off manufacturers. Linn is right IMO to call it an "outright land grab":
      https://www.linn.co.uk/blog/mqa-is-bad-for-music

      I am truly blown away by the audiophile press' commission in this whole business. Seriously, are they that desperate to report on "news" and that easily bamboozled?

      I dunno about Fremer, I like my turntable and LP collection but the sonic limits are obvious. When he insists about vinyl having higher resolution than CDs, that it betters 24/96 and all the while turning a blind eye to all the timing irregularities, I don't bother reading much of what he says these days. As for JA, I do believe he knows what he's talking about but has to play the part. I'm not surprised if he struggles with this...

      Delete
  2. As always, excellent article Archimago. Objective measurements showing MQA is much ado about nothing. Unless, of course, it’s your tech ;)

    When I purchased HDTracks 24/96 deluxe version of Damn The Torpedoes (DR13), it came with a with a message from Producer/Engineer, Ryan Ulyate, which said in part:

    "We're committed to finding the highest quality way to get music from Tom Petty and the Heartbreakers to you. We want you to hear it at home the same way we hear it in the studio.

    We think the best audio option for your computer or media server is FLAC. FLAC is a high quality file that is the best sounding format for downloads. Unlike mp3, which discards elements of the audio to make the file size smaller, FLAC is a "lossless" format, which sounds the same as the source files it was created from. We made the FLAC files from the same high-resolution uncompressed 24-bit 96K master stereo files we used for the vinyl and Blu-Ray versions of Damn The Torpedoes: Deluxe Edition. When we compared those files to the FLAC's, the waveforms tested out to be virtually identical.

    FLAC captures the full dynamic range of the music from the quietest to the loudest sounds. Because of this (and because we are adding no digital compression) it will not sound as "loud" as a standard CD or mp3. To compensate for this, turn up the volume! With the right system, you’ll be as close to being there as we can get you."

    You see folks, we are already hearing exactly what the artist approved in the studio. In the case above, direct from the Producer/Engineer himself.

    And with MP3 and Dolby AC-3 patents expiring, we should see more widespread use in the consumer space. Which means consumers have access to both free lossless and lossy de/encoders, which work perfectly fine.

    Given that we have free lossless and lossy de/encoders and already being delivered to "as close to being there as we can get to you", says the Producer/Engineer, then what purpose is MQA?

    The number one consumer music "production" issue for the past +20 years has been wimpy loud sound.
    The number one consumer music "reproduction" issue since forever is loudspeakers and rooms.

    I don’t see MQA addressing either of those issues. Nor do I see MQA addressing timing issues with so called de-blurring algorithms. Have a look at loudspeaker timing issues, this is where the audible problems are, and can be addressed with Digital Signal Processing.

    So what value is MQA to the consumer then? There can be only one logical conclusion, as aonetech has come to in the comment above…

    ReplyDelete
    Replies
    1. Hi Mitch. Thanks for the great comment.

      Great quote from the "Damn The Torpedoes" album!

      Nice distillation of "number one" issues as well.

      I wasn't aware this was such a big year for MP3 and AC3... I see the AC3 Wiki refer to patent expiration "between 2008 and 2017". As for MP3 I see that they've all expired as of April 9, 2017. Calls for a beer tonight :-).

      Delete
  3. "When we compared those files to the FLAC's, the waveforms tested out to be virtually identical."

    Virtually identical? Not exactly identical? :-)

    ReplyDelete
    Replies
    1. Good catch Don... Yeah, what's up with that?

      Maybe he's referring to the vinyl master with low bass chopped off and mono'ed and before the RIAA curve was applied :-).

      Delete
  4. Maybe this is off topic, but maybe not. Last night I read through a very worthy topic on Computer Audiophile. It was by Mitchco, and it was offering his insights into recording and mastering, and particularly the effects of dynamic range: https://www.computeraudiophile.com/ca/ca-academy/dynamic-range-no-quiet-no-loud-r643/ . You, Archimago, offered some insightful thoughts too. It really is a masterful piece of writing on the subject of good sounds.

    It's very clear, that until whatever comes out of the studio is done right, everything else (if it has any effect at all) is really just tinkering around the edges, MQA and all. Should we really spend so much time chasing the 0.1%, when 50% gets discarded in the studio right off the bat?

    ReplyDelete
    Replies
    1. Hi Bob,
      Yes, indeed the article by Mitch really does highlight the truth about this whole music-loving-audiophile hobby endeavour. When "hardware audiophiles" debate at length about what DACs we use, whether features like DSD and MQA are worthwhile, how high the samplerate should be, and God forbid engaging in Holy Wars around whether cables make a difference, the world of (non-classical) music production is a mess.

      It's ironic I find that these days, as high-definition movies and 4K TVs are moving towards "high dynamic range" (http://archimago.blogspot.com/2016/09/musings-4k-uhd-tv-hfr-3d-hdr-etc.html), higher resolution and overall more realistic imagery, we in the audio world continue to languish in the shadow of the "Loudness Wars" of dynamic compression.

      Perhaps even sadder is the fact that those who *should* be concerned to voice opposition to the state of affairs stay silent and incapable of using their media platform to speak out... Here's looking at Stereophile, The Absolute Sound, HiFi+, What Hi-Fi, etc. Merrily diddling with more cable announcements and boutique brands starting at 5-figures while literally a generation of new music passes with truly terrible fidelity.

      Sad, but true.

      Delete
  5. Perhaps another post off-topic, but I wonder if you have heard about the new audio analyzer "soundcard" being offered here: http://www.diyaudio.com/forums/group-buys/306322-gb-rtx6001-audio-analyzer-ak5394a-ak4490.html

    I've been following your blog for years and really like your blend of objectivism and subjectivism. Since you are measuring high end audio gear you should preferably have some state-of-the-art measurement equipment as well. Here is one offered at a very reasonable price, and it will work with all the test-software you are already using.

    ReplyDelete
    Replies
    1. Thanks for the suggestion Ole. I'll look into it more.

      I'm sure the precision of the device is great! There are however a couple of very important factors I've greatly appreciated over the years when using an ADC for these measurements that seem to be a limitation with the project:

      1. I've really appreciated the Focusrite Forte's ability to function as a USB powered device! Very portable and small. Just bring a charged up laptop and the device and I can measure anywhere.

      2. High precision pre-amps are important for calibration and to make the most of measurements. The EMU 0404USB was good in the past and the Forte is even better with greater gain/attenuation range. I don't see this function for the DIY project nor does it have built in +48V phantom if needed. I saw someone recommending getting a Behringer pre-amp (yikes!) which I suspect would deteriorate the performance.

      These factors above are important and even if the measurement device isn't the absolute most accurate, so long as I know how it performs and have a good understanding of the limitations, I can work around them and discuss them in my write-ups. Obviously nothing beats a pro AP unit but as a hobbyist, that's insane overkill and won't really add much more!

      BTW I do like that the project uses the AK5394A ADC though which does not show the noise shaping ultrasonic noise shown in many ADC chips. In fact, this week I'll be posting on a device that uses this :-). Stay tuned...

      Delete
  6. If i see, the Diff-maker difference between Mytek Brooklin and Meridian Explorer 2, i am now interested in a Diffmaker measurement between dac´s. But i see, the difference are probably modulated because slightly different clocks. The Frequency spectral display in comparison to the Waveform Display has the same similarity`s at 7 to 8 and 13 to 14 hms. (whatever hms means). But if you have multiple dacs with a spdif and coaxial input, you could in theory give them the same clock signal if you power them from the same spdif source, like your cm6631a to spdif box from breeze audio. Because Spdif has the advantage and disadvantage of transmitting the clockspeed directly. Modern Spdif receiver have a pll with a low pass filter, to filter high frequency jitter, but they still try to archive in average the same clock as the source, because this is what the pll does. So maybe this would make a comparison with different dacs kinda more feasible. Because if big difference between dacs occur, it would mean audio dacs are not that accurate, or the low-pass oversampling filter is not that accurate. Would be also cool to have a multi-bit dac for this comparison. Well just in Theory at least. Sorry for my bad English im German.

    ReplyDelete
    Replies
    1. Hello "Unknown"!
      Thanks for that suggestion. Yes, you're very correct, I could indeed try to use the DiffMaker to examine different DACs by driving it with the Breeze Audio SPDIF. This would make a fantastic little experiment in the not-too-distant future. Now, unfortunately I don't have the MQA DACs in house here, but I can at least look at what I do have with SPDIF input.

      I'll be getting another DAC here pretty soon which will provide an opportunity to try this out!

      Delete
    2. Thank you for replying. Im David. the only question remaining is how is it possible to don`t get slightly different clocks on the ADC side. If the ADC has very low crosstalk, you could in theory record one DAC at the left ADC channel, and the other DAC on the right ADC channel, so a clock should be the same on the ADC side. Thank you for reading what i have to say :)

      Delete
  7. Hi Archimago and "Unknown"

    As for comparing 2 DACs with Digital Audio Bi-Phase Inputs, you should aware of two issues:

    1. For ADC, you need then a 4 Channel synched ADC (hard synched and not Master Clock synched), if you would like to compare two stereo outputs (or doing just one channel of one DAC for comparison, 1 stereo ADC is sufficient).

    2. But every PLL measures also different, on the DAC output. With "slaving" SPDIF Input receiver, they will have the same "average" sample rate, but as I am doing also long term sample accuracy measurements, depending on the resolution of your ADC hard and software, you can / will measure different analog low pass filters in the PLL (see also some Julian Dunn work on that) and will measure also the different behavior of digital PLLs.

    What I want to say is, that you could end up in more differences between two DACs "slaved" via SPDIF Input to the same average Sample Rate, as when you do compare to asynchronous USB DACs, having so the slightly different average Sample Rates, but less "short time" variations of the clock.

    Juergen

    PS: As I was measuring mostly analog PLLs until 10 years ago, and they do differ with filter slope and corner frequency of the analog PLL filter, I was surprised to see, that digital PLL does measure totally different, that what are the weak points with analog PLL came to strong points in digital PLL and vice versus (and so a combination of both is / could be a good idea).

    ReplyDelete
  8. Thanks for the great Objective analysis. While much of the tech is beyond my audiophile understanding, it seems that digital analysis, using ADC has demonstrated how little difference exists between the MQA files and the original HDtracks downloads. In this era of Hyperbole in advertising and marketing, most of mid level audiophile consumers have grown skeptical, of all the hype. We rely on blogs such as yours to sort though the crap often precipitated by the Hi end sponsored, Audio Publishing industry.

    Many of us have long ago ripped our library's into the best formats of that era. Sold all our albums and CD's and are moving forward with the technology. I think that while there is has been much Hype from the Meridian MQA camp, as you mention, carefully worded, but I think focused to the Masses of AudioBud, home and auto streamers, who have not heard true Hi Rez audio 24/96 and above. The fact that you can buy a 3-$400 device and get a Tidal HiFi scrip, for $20 a month and even be in the ball park of HDtrack download listening is an accomplishment in itself. The shear convenience, ease of use, ability to browse and create music playlists from the Millions of tracks out there is no-brainer from my Mid level audio experience. As bandwidth costs decreases, I am sure full resolution streaming will be available in the near future. Just like most of us are sitting on the sidelines waiting for substantial content before we go out and buy a 4k compatible TV. When that happens, streaming 24/96 and beyond music to our Audio system will be a drop in the Home Digital download bucket. MQA may be a passing fad, but I think it is a step in the progression to improving the listening experience for the masses. To all those that have a full HiRez library, and hi end DAC's, your golden ears have nothing to fear/hear.

    ReplyDelete
  9. Here's a 44-page technical paper on MQA, maybe you've seen it:

    https://www.xivero.com/blog/hypothesis-paper-to-support-a-deeper-technical-analysis-of-mqa-by-mqa-limited/

    It seems to concur with some of Archimago's findings.

    Marco

    ReplyDelete
    Replies
    1. Thanks for the note Marco and the link.

      Yeah, MQA... So many claims, so little proof :-). I'll have more to say about MQA in the weeks ahead as I explore the Dragonfly Black MQA "rendering".

      Delete
  10. When you take a flac file and it's pcm original, and generate a pcm version from the flac again it is my experience this "flac" generated pcm file is bit identical to the original version. There seems to be no need for MQA at all in my opinion.

    ReplyDelete
    Replies
    1. Yes, FLAC is lossless. No problem.

      The claims / "promise" of MQA basically distills down to 2 things the more I look into it:
      1. File compression - basically being able to decompress some high frequency material about equivalent to what one has in a 96kHz file in a 48kHz stream. As discussed, this is basically due to the partially lossy compression technique.

      2. MQA can apply some DSP processing to change the sound. This can make the sound different but not necessarily better. I can also say for a fact that the MQA stream has embedded within it parameters for depth of dithering, type of noise shaping used, and playback digital filter settings. In a way similar to how HDCD embeds playback settings. Effect of these changes would be variable and more likely than not highly subtle.

      As above, will discuss more in future article(s)...

      Delete