Saturday 7 October 2017

MQA Core vs. Hi-Res Blind Test Part IV: Subjective Results & The Wrap Up...


After having gone through the "core" results and subgroup analysis from the MQA Core vs. Hi-Res PCM Blind Test, let's review and post some subjective comments gathered in the survey. This provides an opportunity for us to see how people described what they heard and since we know what the preferences were, put them in the context of whether the descriptions referred to a bias towards the MQA decode or standard PCM...

I. Subjective Comments and Preference


First, let's start with those who preferred MQA. Note that there were many other comments but for the sake of keeping a reasonable length for this post, let's just focus on those who rated their observations as "moderate" or "clear". Depending on the track, the MQA could be either sample A or B; what's important in these comments is that the MQA sample is said to be "better" by these respondents.
A is more nervous than B, be is more calm. More natural room on B. B has more ease and relax breath and open, A is more "stringent" even with a bit of more detail, a bit more flatter but wider and a bit "stress" compared to B
To my ears, the sample A seems a little bit harsher than the sample B. The sample B has a smoother sound, with softer highs. It may be a little less sparkly, a tad more compressed, but it is less fatiguing, and puts more emphasis on the delightful vocals. I'm hearing this difference both with my Sennheiser HD600 and my Beyerdynamic DT700 Pro.
Whole recording more open and detailed. Strings have more dynamics at beginning. Better focus on the singer, less blended with strings. Around 0:18 ’t’ consonant better on « et miserere ». Large soprano crescendo around 1:10 (some mike overload?) better, more ample. Child choir better staged, hear more the spreading.
A has more space, B sounds metallic and unclear
A Has more air and refinement
Track A felt more alive and open where B felt compressed
My preferred version sounded a little bit brighter, and therefore more 'open' in the mid to upper frequency range.  I'd guess above 4kHz or 5 kHz 
The vocals in sample B didn't sound as "clean".  Sample A seemed to have more air, purer sound. 
Sample A feels more excited, more plastic, more graphic, easy to tell apart.
I hear more the bell-tone (inharmonic) of the piano everywhere. Piano has more definition for each note, sounds closer and I hear more realistic hammers. The cello part is kept more separate. 
A Has more air and refinement
To my ears, Sample A contained more textural information: the piano sounded lifelike, whereas the piano in Sample B sounded like a digital facsimile. The higher registers in Sample B had a slightly hard timbre. Sample A also sounded more spacious, especially when the violin entered.
sample B sounded slightly warmer and not as clear but better than sample A.  sample A sounded slightly strident on certain notes
More natural acoustics around the instruments and natural timbre on Sample B.
A is more nervous than B, be is more calm. More natural room on B. B has more ease and relax breath and open, A is more "stringent" even with a bit of more detail, a bit more flatter but wider and a bit "stress" compared to B
The violin seemed far more strident (and 'screechier') in A
There is more dynamics, more subjective high frequency content. Better positioning of instrument and I hear more the room acoustics. At 0:43 the horns sound more delineated. When the solo violin starts I hear more the bowing. At 1:51 when the solo violin plays lower I hear more wood.
Massed strings were more natural sounding in sample B.
B has more space, A lacks nuance and the strings sounds chopped up, the solo violin sounds metallic
B Has more air and refinement. A Sounds harsh.
B's violin is natural, A is Brilliant.
The second sample had a better, clearer, sound quality, with the first sample seeming a little harsh. 
The major differences were with the soloist. The B version made her violin sound more expensive - the tone was silkier and a little recessed, as if some unknown added bits were no longer as present.

All that sounds good about MQA, right? And suppose if this test were not a blind test, and those people found out right there and then that they selected the MQA track, they could by right go online and happily express their "moderate" to "clear" preferences. But this is a blind test! So there is another side to the equation... Here then are the obverse comments - those who preferred Hi-Res PCM sound to a "moderate" or "clear" extent:
B just seemed easier on the ear and more pleasant to listen to.
Sample B preferred as there is much more seamless integration between the voice and the instrumentation and the voice is more expressive and natural sounding to my ears...
B is clear, brilliant.
01 deeper more resonant, wirey mid? 02 'hi fi' with emphasis on top end, less bass.  On 2nd listen this was open and relaxed without the strained mid.
Sample B sounds more open and dynamic
Sample B: Better definition of instruments, more clear piano sound.  Better "room" or "space". The female soprano or mezzo has problems with a disharmonic lack of overtones on both samples.
The B sample sounded "smoother", more enjoyable to listen to.
"t" and "s" sounds are more present in the sample B, B sounded cooler, I liked it more.
The sample B sounded more clear and relaxed.
A sounded thin. B had more bass and sounded warmer - fuller - more reverberant.
sample B was clearer, sample A sounded slightly veiled and warm
Again, B was easier on the ear, and I thought I heard some distortion on the piano in A
Sample B preferred as more lyricism and presence in the playing comes through. Very subtle but very profound too.
B has more space, A lacks nuance and is a bit wooden
"A" is harsh, shrill, thin
The lesser track lacked severely in the sense of being a piano
More natural, richer piano sound
In sample A, the pianist seemed to be just plonking down his fingers on the piano.  On sample B there was far more feeling in the playing.
More apparent difference between the tracks than on Arnesen tracks. On track A, low volume passages sound boxy at times. At times, track A very subtly sounded like an electric piano/synth/sampler. Track B sounds more open, dynamic and natural.
Sample A has much more rhythmic verve and drive...impossible not to conduct along with it and move one's body...
Sample B: more grainy, less transparent, less reverb, solo Violin slightly distorted
"B" is harsh, shrill, thin
Better spatiality, more 3dimensional. Sample B sounds a bit flat (stage wise) and opaque (I hate Audiophile speak).
Would it surprise anyone that the audiophile press covering industry events do not seem to mention that perhaps at least some people actually prefer the original hi-res PCM versions?

As subjective comments, I appreciate the sharing of the respondents and certainly I would not say anyone is wrong. In fact, that is the point of this exercise, to accept that for each person, we make a subjective choice and can justify our selection with words. But when viewed in totality given the 83 respondents, there was no significant evidence to show that MQA Core achieved sonic superiority.

II. Other Interesting Comments and Observations


Within the data set, I see some folks went quite far with ABX testing; like this:
My ABX test shows I was guessing half the time. If there was any difference, it was perceived as a slight image shift is the best I could describe it.
I certainly appreciate the time and effort! An example of how difficult the task can be... I know that some people were able to complete ABX testing by differentiating the slight differences in fade in/out between the samples but obviously that was not the intent of the survey. Even if one could ABX between Sample A or B, a decision still had to be made as to sonic preference which would be independent of where exactly a fade in/out happened.

Here are some more interesting comments I found in the data. Again, thank you to all respondents for taking the time to send these observations and thoughts.
Don't know if and what differences I heard during the test. What I heard was subtle and perhaps subjective. Can't say if MQA is better of not, not even when I would all choices made right. For me the current situation does not give me a reason to use MQA instead of FLAC.
I am a physician and hence a evidence based practicing man of science. Based on what i listened it's quite obvious that there is no "astounding difference" in MQA tracks. Even listening on my system which i think is one of the purest and most transparent, the tracks are almost identical and there is nothing outstanding about any one of them except a subtle personal likeness for "ARNESEN A" for some reason (ABX comparator gave me a 36% chance of guessing for it ). Based on my experience, i say that MQA doesnt offer significantly noticeable benefits for me and hence must be snake oil.
In my opinion, the MQA discussion seems to be more about selling than educating. Further, the proponents of MQA all seem to have something to gain by pushing MQA upon us. The closed shop nature of MQA and the need to invest in new technologies with little or no return means it is a non-starter for me. Even the lower bandwidth usage is irrelevant. If we can stream HD video over the Internet, where is the supposed problem with high quality audio? There are many who say they hear differences in quality when comparing high-resolution sources (44.1/16 and above), but few (none that I know of) who can prove their hearing ability in a controlled environment? The current CD standard seems to be more than our ears can hear, I see no reason to for investing in more.
I believe I can hear the effect of phase correction, having used extensively the DEQX active correction driver and room correction device. To be honest, my responses here are as much guesses as anything.
Thanks for this test, Archimago, though you already know my opinion on this :)Better use of one's time than to pry minute details out of recordings which are already made to exceed auditory capabilities of humans is to make our own music, or to start with OpenMPT and tracked sampling, where most hifi mumbo jumbo gets out of the window and linear interpolation is the HQ option for resampling and 44kHz samples are hifi :)Best of luck,OJ aka. Subit
With my 71-year-old ears, this required a lot of concentration. The differences were more subtle than comparisons of other files previously downloaded from 2L, such as between 44 KHz and 192 KHz files, where the differences were immediately apparent.
It's absurd to imagine that anyone could get worked up into a lather over such small differences in sound quality. If spending hours trying to distinguish such differences rocks your boat, it's time to re-think your life. 
You could have sold either one as MQA to me. Since I cannot complete the selection without making a choice, I selected B for all. I have no preference or my preference changes depending on the sequence. Will make another attempt with the $16K main system later.
I don't think MQA it's a good idea for audiophile, when I listen to TIDAL Master on my system I prefer always lossless standard FLAC version
Archimago - great work many thanks. Your effort - along with Mans - to deliver objective views and facts on audiophile issues is of great value and appreciated. It's a strong Counterpart to hyperbolic press reports (f.e. Lavorgna), seeming to be too gullible, less critical to what manufacturers are presenting them. Maybe sometimes there is also a great lack of true knowledge and understanding of digital audio technology.
How about a third file where the samples alternate every 10 seconds in a completely random way to test if one can sense the difference in the flow?
Like any new format MQA only offers something on old recordings, new recordings will not benefit from anything if produced and mastered well in the first place. 2L are nuts for giving there recordings to MQA, they are beautiful in the first place. I own a copy of the Hoff ensemble Quiet nights, a remarkable recording
It is very hard to distinguish the files and not practical to talk of an advantage of either file processing. Differences seem to be audible sometimes and on small passages only, which might be identified by concentrated (and exhausting) ABX testing. While I used calibrated headphones, those nuances would probably vanish in conventional/home listening environments, where reverberation etc. could mask them. Furthermore, I did NOT find one or the other sounding better [a choice was requested to complete the survey]; our 'musical brain' is adapting too fast to what it is processing (or, loosing musical memory too quickly). In real life, one would never switch gapless between those ABX files and thus enjoyment will be equal for both cases.Thanks for this test, that was fun!
Way to go! The samples, to my ears are very difficult to tell apart. There seems to be a difference, almost like an excess phase difference, but the ABX shows I was guessing half the time... I am going to try again, but use my Bose in-ear noise reduction headphones a try. All the best.

 

III. Concluding Remarks


I must say, although it is quite a bit of work doing these blind tests every once awhile, it is also great fun with the hopes of adding to the "data" out there beyond isolated opinions. It provides a behind-the-scenes look at a selection of experiences, opinions, motivations and glimpses into the participants in this hobby rather than depend on claims of members of the press or the promotional side of the industry. I see these 83 respondents as the people who actually fund the hobby. For that, I thank all who contributed to this study.

I totally accept that there are limitations to what we're doing here:

1. The 2L samples may not be the best to use. But they were released for testing purposes so I would hope the MQA company and 2L together agreed that these tracks were showcases of what this encoding technique could do. Some have suggested that these 2L tracks already sound "too good"... Well, it goes both ways. We're not just looking for MQA to make an improvement, but we're also looking for MQA to not cause deterioration to hi-res PCM. If I started with crappy hi-res PCM and all MQA did was "sweeten" the sound by some kind of EQ manipulation, I'm not sure if that would be a fair gauge nor should that be a ringing endorsement of MQA!

2. I can't control for the conditions used by listeners at home. I don't know if some ran the tracks through waveform editors, performed technical non-listening tests, etc... I also cannot know for sure that the equipments used were in good functioning shape, capable of high resolution, or that they were set up properly. But there are benefits to doing it this way. We have the potential of obtaining a naturalistic perspective with the gear people actually own, using their set-up, and it is after all in one's own home where benefits are the most meaningful. IMO, this kind of data is more useful than tests run in audio shows or even in a lab. While a lab can give us a better idea of thresholds of audibility, the numerous changes in variables would take us away from typical listening environments. It is also the home listener who coughs up the cash to buy a new MQA DAC or decides to consume / purchase MQA music.

3. The use of an MQA-like upsampling filter for all files can be controversial. Sure, I'm upsampling all the 24/96 or 24/88.2 tracks to 24/192 or 24/176.4 using iZotope RX 5 with its 64-bit DSP. The impulse response with this filter isn't exactly the same as what MQA uses but IMO it's darn close. The benefits of doing this is that with standard DACs, it will make the sound closer to what an actual MQA DAC will output by replicating the ultrasonic "leakage" inherent with the use of these gentle, minimum phase filters that MQA seems to prefer. This will allow home listeners to experience the sound of MQA closer to the result of the MQA "rendering" step. In any event, by controlling filtering and using the same upsampling settings, any differences heard would still be attributed to the difference between MQA Core and Hi-Res PCM source.

4. Similar to #2 above, I cannot confirm if the listeners have adequate hearing acuity or whether they put full effort into the task. But even this is instructive because the impression we've been conveyed in the media is that MQA sounds obviously better with claims that it should be evident within "seconds". Remember though that because the procedure wasn't just a simple poll and required deliberate submission of results, these were not "casual" poll numbers and these respondents needed to do some work just to get the results to me.

Within these limitations, I think it is fair to conclude with the following:

As a group effect using MQA Core decoding compared to standard hi-res PCM downsampled to 24/96 or 24/88.2 in a home environment as tested by 83 audiophiles throughout the world in their own homes and using their own gear, there is no clear evidence that the MQA Core decoded version provides the kind of remarkable change to the sound claimed by many audiophile writers or by the company.

These listening test results correlate very well with previous demonstrations using TIDAL streaming showing high null depths with digital subtraction (ie. little residual difference). These objective results months ago predicted that the listening test would not be easy. As reported in Part III, I’m very impressed by the "golden ears" overall and particularly the individual who was able to successfully identify both the previous 24-bit vs. 16-bit blind test in 2014 as well as this MQA vs. standard hi-res PCM; this is certainly not the statistical norm!

Let's be honest folks, I know the MQA vs. Hi-Res PCM "debate" is but a tempest in a teapot (or storm in a teacup if you prefer). Suppose for a moment though that we actually have an important choice to make as a group of audiophiles. Suppose the "big decision" is that we have been charged with the power to decide whether MQA will become the de facto, hi-res encoding system going forward. When faced with a potentially important change, how do we rationally decide if it's actually worth it by the data that we have? One way is to look to the world of medicine for some perspective when making important choices using data that's not black or white...

In the medical world, where decisions made are potentially important for those with severe illnesses, the concept of “number needed to treat” (NNT) when assessing experimental procedures or medications could be instructive. For example, if we needed to treat 5 patients with a certain medication such that 1 patient would benefit significantly, then we could say that the expense to treat all afflicted patients with such a medication could be worth it. BUT only if the new treatment provides significantly better outcome for those it helps AND the new therapy doesn't do harm with side effects and the like (the converse concept also known as "number needed to harm", NNH). Those making the decision to adopt or reject the new therapy will have to balance these factors of benefit and harm, as well as magnitude of effect and severity of the affliction. For example, a government program like universal health care in most developed nations or perhaps Medicare in the US cannot fund every pharmaceutical company's product without decent evidence (not just company claims); it all depends on value based on results such as NNT/NNH.

The world of audio isn't managed like pharmaceuticals nor medical treatment modalities of course (thank goodness!). Rather, we do not have a regulated system where audio companies have to prove their worth before products can be introduced on the market (on balance, I see this as a good thing for consumer choice and marketplace competition). Analyses like the NNT and NNH aren't clearly defined nor expressed; the company is free to use any superlative it wants with no burden of proof although obviously if the consumer disagrees, it will look bad for the company to be overzealously advertising their products. As such, we expect the consumer to make these choices and a blind test like this one I think is an example of consumers doing due diligence. (IMO, it really should be an unbiased press aligned with consumers advocating for and doing much of the home work in helping the public understand products in an honest light.)

Within the medical analogy, we can hypothetically ask - should "we" as consumers and as the proverbial "funding agency" decide as a corporate body to spend money on this MQA "treatment" and its deals with the music labels (I hear the "big 3" of Warner, Sony, and Universal have signed on) based on the data? Certainly if we look at the data presented here, as a group, the effect of this MQA "therapy" isn’t particularly impressive with no evidence of "treatment superiority".

Does it make sense to promote a "therapy" where audiophiles spend money to buy a new MQA DAC when they already have DACs capable of better-than-CD specs already? Is the filesize reduction worth it on the whole? Does it make sense to spend money to convert much of the music out there to a partially lossy-encoded file type which actually can pose a potential sonic downgrade to some (a very real potential "side effect" as evidenced by all those who preferred Hi-Res PCM)? Does it make sense to pay more money to download the MQA version as opposed to a cheaper 24/48 or 24/96 or 24/192 downsample of an original DXD (based on 2L’s pricing system - check out the price for Magnificat for example)?

Also importantly, is it worth it to dispense with the freedom to access the full musical resolution contained in the data given MQA's proprietary encoding/decoding technique?

Is there any evidence empirically that these "costs" are worth the supposed rewards such that as a whole, audio lovers would benefit if we encouraged adoption of MQA en masse?

A few questions for "us" as consumers and audiophiles to reason out.

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

With this post, I think I've reviewed the blind test survey data thoroughly enough and expressed my impressions to an adequate degree. I have one more post after this to address a couple of interesting articles on MQA which I hope would be the last of my comments about MQA for awhile.

Again, to the 83 respondents, thank you for taking the time and all the efforts! It has been fun...

Hope you're all enjoying the music...

23 comments:

  1. Hi Archimago

    My last comment on this subject. After some thought, one cannot dismiss MQA, as that would be unfair. I have heard it with full decoding from Tidal and it does sound good. But I think the points raised in various places around the internet support my decision not to fully support MQA.

    What I have come to realise, as I stated before, is that MQA is not significantly improving the quality of the experience of listening to music and therefore does not warrant further investment from me. Contradictory to this is that I own a Meridian Explorer 2 DAC and very good it is too. However having used it with MQA and non-MQA files I haven't had a eureka moment that changed my thinking. I will continue to use it as my portable DAC and reap the benefit of better than the tablet/laptop output it gives.

    Like any product a business can sell it as the leading technology in it's field and we the consumers can decide for ourselves whether it is worth it and ultimately how successful it will be. As always we can challenge the claims, as you have in an adult and measured fashion. Conversely consumers can just go with their experiences and buy into the product.

    I am glad there has been a lot of discussion around MQA, it has been enjoyable to read both the backers and the non believers. The only downside to this sort of discussion is that those we seek to bring in to the world of better quality of sound will have been turned off especially by the technical side apart from the bickering (your pieces on MQA have not done this in my view).

    Whilst I understand there are many who would disagree, I still feel that the music industry should move to a 24/96 recording quality. Whilst I do not know about cost, to drive that down means it needs to be in a lot of devices (and is) and for it to become the de facto standard. 2L may not be mainstream but their production process shows what can be achieved and I am sure they are not the only ones doing it. As I read his website, Mark Waldrep's work should not be lost but handed on to the next generation, let's hope some of his students take up the mantle.

    The manufacturers of playback equipment have done a lot to produce good quality hardware at affordable prices, so the content should match it. I quote as an example the pair of KEF Egg wireless speakers I purchased. I can clearly hear the difference in recordings using these, they are small, relatively powerful, certainly enough for the average UK front room, and I added a very cheap 50 watt sub woofer to them for the princely sum of £65 and get a whole lot of enjoyment when I use them. At a total outlay of just over £400 I would imagine this would please a lot of people and satisfy them. Best of all is that these speakers can handle 24/96.

    Anyway, thanks for this test, I look forward to what you have to entertain us with next.
    Definitely enjoying the music regardless of the resolution (higher is better for me)

    ReplyDelete
    Replies
    1. Thank you Gordon for the a great discussion.

      Glad that you're enjoying the Explorer2. Although I don't own it, having explored the output measured by a friend, it's certainly a very capable device regardless of the MQA capabilities.

      While bickering could of course turn folks off, I do hope that we as a hobby can achieve a more refined state of discussion and debate that elevates and achieves honest true advancement without pitfalls and compromises. This is audio engineering after all so I do believe it is achievable!

      There is also no shame in acknowledging that maybe after all these years of refinement, hi-res audio like 24/96 as you discussed is more than likely all we ever need. What we truly *desire* are recordings worthy of that "hi-res" title... To get musicians, producers, engineer up to speed with the best of what can be achieved by the likes of 2L, AIX, Telarc, Chesky or sensibilities of what sounds good by audiophile-level sound engineers like Mark Waldrep, Bob Katz, Steve Hoffman, Kevin Gray, Barry Diament... (No, I do not ascribe to all that these guys might believe in, but what matter is that they make good recordings!)

      Hardware is indeed getting less expensive for amazing quality. That too is a natural consequence of technological maturity and incremental change to the point of diminishing cost-benefits. I suspect that some in the "high end" might not like this, but we can't deny the reality.

      Delete
    2. Just wanted to also acknowledge the works of "Prof" Keith O. Johnson (Reference Recordings), Jared Sacks (Channel Classics), Morten Lindberg (2L), Peter McGrath (Wilson).

      Like I said, I might not agree with all these guys on what they believe about stuff like MQA, but when it comes to *results*, they deliver!

      Delete
  2. "Hardware is indeed getting less expensive for amazing quality. That too is a natural consequence of technological maturity and incremental change to the point of diminishing cost-benefits. I suspect that some in the "high end" might not like this, but we can't deny the reality."

    Once upon a time high end was truly high end with respect to high fidelity. That is generally not the case today. It really started with the introduction of CD. It was a perfect storm as prices for quality hardware came down and more evidence of expectation biases and marketing influence being problematic - for example, Carver in the 1980s proved beyond any reasonable doubt that all amplifiers can be made to sound the same and transparency was well within reach of the average consumer.

    High end today is a rather sad state of affairs. For example, given that a fully transparent amplifier made with quality components inclusive of a reasonable profit margin for manufacturers, distributors and retailers can be had for less than $2k, what is the point of paying $20k plus for an amp? Generally these high end amps deviate from transparency to produce a "signature" sound, which may be fine if that is the sound one prefers, but it deviates from the goal of high fidelity. The same can be said about high end DACs and every other component in the reproduction chain except perhaps for speakers (but even they have improved with technology - eg price of good active speakers) and to some extent turntables.

    The problem with the current state of affairs, and the compromised subjective publications and their reviewers, the focus is turned away on what is important, ie quality of recordings. Given that the quality of recordings, speakers and room acoustics are the three areas where one can achieve better sound, this is what audiophiles and publications should be focusing on. Hi res, MQA etc are just diversions and even if they did make a difference, we are talking about one per centers rather than significant improvements which can be achieved with the three areas.

    ReplyDelete
  3. Hi Guys,
    Many thanks for this enlightening discussion! Very very interesting.
    I have a question!
    Using Tidal on my mac, set to master quality and using a Echo Audiofire4 interface.
    I cannot get Tidal to force the bit rate on my soundcard.
    I have selected 'exclusive' setting in Tidal as recommended.
    When listening to masters, what should I set my Audiofire to? Options are: 44.1khz,48,88.2 or 96khz.
    Not sure if I can hear any difference. When I do, I am not convinced I am imagining it!
    thank you in advance.

    Rich.

    ReplyDelete
    Replies
    1. Exclusive Mode
      In Tidal, got the settings, than at the right end of your sound device name "Audiofire" go deeper into the settings and activate "Exclusive Mode". With this, Tidal has control of the sample rate of your Audiofire hardware and does chose automatically the right sample rate.
      Juergen

      Delete
    2. Hi Juergan,
      Thanks for the reply. Very much appreciated.
      I have been through all the settings in Tidal and the console program for the Echo but there is no way to force the sample rate. I have activated 'exclusive mode' but no luck unfortunately :-(
      I check the sample rate on the console of the Echo but it doesn't change.

      Delete
  4. *I am not convinced that I am not just imagining a difference!
    sorry, thanks again!

    ReplyDelete
    Replies
    1. Hi there Rich,
      Not sure I can answer the specific hardware question and settings but I'll try... Looking at the Sound-on-Sound review, it's a Firewire interface, right? I also don't use a Mac for Tidal, but if the software decoding is like Windows, then all it'll do is perform the "unfolding" either in 88.2 or 96kHz depending on which timebase the original track came from. For example, 2L tracks originating in DXD will be 88.2, but if the original came from 192 or 96kHz, then you should set it to 96kHz.

      Personally, I don't think there's much of an audible difference these days with upsampling, so I'd be happy with setting it to 96kHz and listen that way...

      Good luck!

      Delete
    2. Hi Archimago,
      Thank you for the quick response! Really appreciated.
      Yep, the Echo is a firewire interface.
      I was listening and bouncing between the different settings last night and had come to the conclusion there was no noticeable difference. I will run with 96kHz for sure after your advice.

      Also just to add, The Diana Krall album on Tidal "turn up the quiet" has been the most revealing test material I have tried so far. Going between the Master version and the non master version is a great way to hear subtle nuances as it is so well recorded. Have a try with 'blue skies'

      Thanks again for taking the time to advise, really really appreciate it. You have a new regular reader of your blog. Keep up the good work :-)

      Delete
    3. Thanks for the recommendation Rich.

      Hmmm, I wonder if anyone has checked the Diana Krall track to see if the mastering could be different between standard CD and the hi-res (which presumably the MQA encode comes from)?

      Delete
  5. Part of my response is a function of age, I know, but, really?

    Ever since the advent of the CD there have been a camp of those who insisted, with a straight face, that vinyl was audibly superior, and another camp convinced the CD could be audibly bettered. As to this second camp, I always applauded their determination and mission. We're human, so we believe everything can be made better, right?

    So we had a hardware explosion (DACs, CD transports, ludicrously expensive cables, etc.) and something of a tech escalation. Remember HDCD? How about Super Audio CD vs. DVD-Audio? How many gazillions of dollars were spent (by me, among others, I admit it) on various formats and components in search of digital Nirvana. And what have we learned, most recently in these tests? They pretty much got it right in the first go. I have seen no tests that have shown me people can reliably tell 16/44.1 from 24/384. Or CD from SACD. Or SACD from DVD-Audio. They say they can in Stereophile or the Absolute Sound, but when the blindfolds come out everyone goes quiet.

    Future improvements will continue in the hardware. It won't help the sound, but it will further the evolution foretold by Tom Nousaine - that modern audio equipment is just another appliance, something to be plugged in and used and not obsessed over. The only thing capable of actual forward progress on the audio quality front is transducers. Otherwise, we've pretty much got this down. If these results, which were obtained after great time and effort were expended, and required tremendous concentration, are indicative of the actual audible "differences" we can realistically expect, it's time to move on to something actually important.

    ReplyDelete
  6. I forgot to add a heartfelt thanks. This must have been a bear to put together. You are a better man than I am.

    ReplyDelete
    Replies
    1. Hey Jeff,
      Yeah, a good amount of work to put together, but overall, not too bad and worth it IMO to get data out there for people to think about... As I noted above, I think the work should have been done by a non-biased press that puts the interests of consumers first. At the risk of becoming (even more?) "politically incorrect" I feel that what we have these days is an emotionally stunted, intellectually deficient, and morally unprincipled mainstream "audiophile press" catering to their self interests only (heck, much of the online press is even worse!). There's little actual content, critique, or education. Might as well be cheerleaders. There are exceptions and I appreciate when I see the occasional "minority report" out there.

      On the whole, yes, I believe Nousaine was right. Over time, we will refine the understanding of the limits that we as humans are capable of. But the refining process and better understanding of the thresholds do not mean that there is wholesale significant improvements to be had; typically, further improvements become essentially meaningless. It's not like the science changed *that much* over the last few decades! Hats off to those who brought us the CD 30+ years ago. They did their homework well and even though we can argue about the margins of improvement in terms of bit depth, "brick wall" filters and whether 22.05kHz was perhaps too close to the typical 20kHz limit of human hearing - 99.9% of the time, nobody's going to know, care, or bother.

      As audio nerds, we can certainly keep pushing those boundaries and wanting the best (like say lossless hi-res). But as time goes on, other than making sure that we buy reasonably high quality hardware, there will be fewer and fewer nerds left willing to care about a jump from DSD128 --> 256 --> 512, or PCM 96 --> 192 --> 384 --> 768... We intuitively know that this all becomes unimportant and there's absolutely no need to put $$$ down. IMO, this is why the hobby is diminishing in people interested even though there will be spurts of interest with mobile gear, headphones, and LP.

      Who knows, maybe in time there could be new "formats" worth exploring but I doubt there's much science left in the tank for new ways to handle PCM/DSD. We might need to hack the human neural network system first :-).

      Delete
    2. The only (not so) new format really worth exploring is: 5.1 channels and beyond. Those differences are *real*. ;>

      Delete
    3. Steven,

      I agree that 5.1 and beyond represents huge potential in really exploiting psychoacoustics. Two-channel will persist for some time as long as we have portable devices, and for people like me who have neither the money nor space to do "5.1 and beyond" right. And, boy, can the average guy (including me, probably) screw up the setup for anything more than two channels (and even that, come to think of it).

      But I don't think there's much more to be done with digital decoding, at least as far as resolution, etc., are concerned. We know how to do that, how many different ways do you want to complicate matters in order to get essentially the same result?

      My own admittedly amateur opinion is that multi-channel techniques will grow more and more sophisticated psychoacoustically (Is that a word? It is now!). At the same time, I agree with the late Peter Aczel: We have to abandon the "monkey coffin" school of speaker building and start moving in the direction of Siegfried Linkwitz and others (many, many others) who are revolutionizing transducers. We can do much better than we are when it comes to speakers. Outside of the physical listening room, speakers are the most important part of the reproduction chain. And almost everyone, myself included, has boxes with drivers in them, maybe floorstanders, maybe on stands, suffering limitations we don't accept from other components. The best way to listen at home is sitting down in front of a playback system worthy of the room it's in. We have to give people systems that make it impossible to get up and leave, the sound is so good.

      Delete
    4. Dr. Mark Waldrep has some of the demos at
      http://www.realhd-audio.com/

      Delete
  7. "Is the filesize reduction worth it on the whole?". The words "nail" and "head" come to mind. Fast download speeds and affordable computer storage make the data reduction part of MQA's aims redundant. So we'd better have a huge increase in sound quality then? Hhmmm.

    ReplyDelete
  8. Dears,
    just to understand something more... I am not a big expert and the docs in Interent are confusing me.
    Let's take a premium label selling high-res liquid music: http://www.2l.no/
    Thay say that everything they sell is recorded in DXD format (aka: 24 bit, 352 kHz PCM), mixed, dubbed, etc in DXD format than put on the shelf for customers in DXD native or any other format: lower resolution PCM, DSD, MQA.
    Seems that while few record in DSD, no one is recording in MQA and so MQA (and any other format) comes from the DXD.

    If the above is correct, then the question:
    MQA is intented to reduce the size of the file and keep quality versus the original, but how can MQA be better that a PCM 24\352 ?
    One can say, maybe, that a DAC can work better in decoding MQA so the final quality is better, but there no DAC (chip) in the market capable of decoding MQA and everything is managed from MQA to pure PCM later on.

    Again, how can MQA be better than original DXD \ PCM 24\352 ?

    ReplyDelete
    Replies
    1. Good day sonus.faber,

      Yes, you are correct. MQA is a derivative of standard PCM whether it was originated as DXD, 24/192, 24/96, or even lower-res stuff like 24/44 (for example, TIDAL's Bruno Mars or Beyonce albums come to mind).

      One does not record in MQA. It's simply a final encoding system to compress the original "studio master" high resolution data into something that resembles 24/44 or 24/48 supposedly for easy streaming with subjective sound quality maintained (hence bringing the "studio sound" to you).

      As you know, they claim the sound can be "better". I can claim that I'm an awesome chef and make the best tuna casserole in the world :-). Neither of us seem to be able to prove this "fact", unfortunately... You'll have to take it on faith.

      Cheers!

      Delete
  9. Speaking just about technical aspects, I would say that a MQA encoded file can be better that a pure PCM file with same resolution \ file size, because MQA can pack more information, but I would say that it can not be better than the original file because MQA for sure stripped something with its lossy compression.

    Moreover MQA is decoded to pure PCM before reaching the PCM DAC chip: MQA is always decoded by a standard PCM DAC.

    Of course one can say that MQA sounds better or not, but these are subjective feedbacks...

    From a pure technical perspective, MQA can be better and of course it is bettere for streaming and low bandwidth scenario.

    ReplyDelete
    Replies
    1. Standard FLAC actually compresses better than MQA at similar quality.

      Delete
  10. I know, I'm late to the party, but...

    Taken as a whole, the subjective comments leave me with the impression that the PCM files were more detailed with a more clinical/analytical sound, while the MQA files were warmer and less detailed, perhaps with some additional emphasis on the midrange at the expense of treble. It reminds me of comparisons between between tube and solid state amps - solid state has unequivocally lower measured distortion, yet many listeners prefer the warmer sound of tubes.

    I think that CD audio (PCM 16/44.1) is the point of diminishing returns in digital audio - the engineers who developed this format did their job very well indeed. Modern technology has drastically reduced the cost premium for better than CD quality, so we might as well indulge.

    If one wants to reduce streaming bandwidth, while offering better than CD quality, why use a the proprietary lossy format like MQA? I would prefer any of the following options using open codecs widely supported by modern hardware and software.

    If one wants a lossless format, one could use FLAC encoded PCM at intermediate bit depths and sample rates: 24/48, 16/88.2, 18/96, etc.. Any of the common sample rates could be combined with any bit depth between 16 and 24 to get the desired balance of audio characteristics and file size, and the resultant file could be played by most (all?) reasonably recent DAC's. Previous studies indicate that the difference between CD and high res audio is barely audible, and the difference between an intermediate format and and high res would be even less noticable.

    If a lossy format is OK, they why not downsample the master to 24/96 and apply MP3, Ogg Vorbis, etc. compression to reduce the file size? While lossy formats are bad for archiving music (transcoding them to another compression format often creates audible artifacts), previous blind listening tests show little if any difference between lossless and high bit rate lossy compression.

    For archival master copies, I'd like to see studios using the highest possible bit depth and sample rates with lossless compression, as this offers maximum flexibility for transcoding to other formats in the future (including future formats that don't exist today). I'd also like to have the option to buy the master file and transcode it myself to suit whatever devices I have today and might buy in the future.

    ReplyDelete