Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Meridian Audio's new... sub-format called MQA. (Read 147134 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Re: Meridian Audio's new... sub-format called MQA.

Reply #275 –
Is that hostility I am sensing, Wombat?
ABX is a wonderful test I love, but a single,  so far never replicated test, of one person, with only ten trails, with a p-value not quite reaching the  .05 level? Perhaps we feel differently about the significance of this.

I also may be reading what he meant by being 'aware' of the noise floor differently than others, in his blog:

Quote
I can hear the musical residue but shifted to a higher pitch when playing the "difference" file. If I pump up the volume, the noise tonally is higher with a rising noise level above 5kHz as shown in the FFT. Perhaps this accounts for the tonality change I heard when ABX'ing.
[emphasis mine]
So he himself admits the noise may be the culprit! His notion that we don't always hear noise as noise, but rather it may, for example, alter perceived tonality, jives with my own experience.

and

Quote
Note that even though I was aware that the MQA decoded file had a higher noise floor, I did not purposely take advantage of this by listening to a quiet portion of the file with volume pumped up to listen to the hiss.

So he didn't focus on the hiss purposely. Good. But what assurance do we have that he didn't accidentally do so on a subconscious level?
----

Have you all listened to the tune? It's a great tune for hearing differences in noise floor:
It's quiet and yet dynamic, so when we say "the noise is 74 dB down [from 0dBFS]" it doesn't mean the same thing as with a loud and compressed song. If for example this quiet song has its (only occasional) loudest peaks 13dB down and the average RMS level is 26 dB down, then this noise floor is much closer to the music level than is normal/typical of compressed pop music.

Did he consciously listen for the noise? No. Might he have subconsciously though? Hmm....

Re: Meridian Audio's new... sub-format called MQA.

Reply #276 –
If he heard the difference in an ABX test well enough to pass the test, it doesn't matter one bit whether he heard it consciously or unconsciously. He still passed that one particular test, so it is very likely that there is a actual audible difference.

More testing is needed.
 

Re: Meridian Audio's new... sub-format called MQA.

Reply #277 –
But he didn't pass the test under the very common criteria of needing to reach a p-val = .05 or better. Heck, I happened to do even better by fluke, without even listening to the music in my test yesterday of other material and just randomly guessing. Single tests, by just one person, with a small number of trials, which don't even hit p-val .05 don't mean much in my book, but sure, let's do more tests, and with more people.

Plus,  if we were to later determine, through more testing [which I too encourage] the only reason MQA is discernably different from non-MQA music is because of it's added high frequency [but <20kHz] noise, as I suspect, is that something anyone would really want to purchase? Not me.
---

So today he has added a pic. Although, as he explains, he has offset the levels by 20 dB to make the graph easier to read, the relative noise floors of the different formats remain intact when compared against each other. At frequencies which matter, under 20kHz, there is clearly a lot more high frequency noise with MQA both decoded and undecoded. Look at 15 kHz, for example, a frequency some of you younger than me might still be able to hear: MQA is not just a little bit noisier, it is a whopping 12 dB noisier! [Not that I'm claiming it is necessarily always an audible difference.]


Even when not listening to a dead silent passage where at elevated playback this sticks out like a sore thumb as a jump in hiss level or tonality, added high frequency noise can add more "air", "openness", and "sheen" to certain instruments in certain recordings, in fact I believe he mentioned the sound he heard he described as "more sparkly".

Meridian is marketing nothing more than added high frequency noise* if you ask me. The emperor has no clothes!

*in the audible, < 20 kHz frequency band

Re: Meridian Audio's new... sub-format called MQA.

Reply #278 –
Quote from: mzil on
But he didn't pass the test under the very common criteria of needing to reach a p-val = .05 or better. Heck, I happened to do even better by fluke, without even listening to the music in my test yesterday of other material and just randomly guessing. Single tests, by just one person, with a small number of trials, which don't even hit p-val .05 don't mean much in my book, but sure, let's do more tests.

Plus,  if we were to later determine, through more testing [which I too encourage] the only reason MQA is discernably different from non-MQA music is because of it's added high frequency [but <20kHz] noise, as I suspect, is that something anyone would really want to purchase? Not me.
---

Not Archimago, either.

Quote
So today he has added a pic. Although, as he explains, he has offset the levels by 20 dB to make the graph easier to read, the relative noise floors of the different formats remain intact when compared against each other. At frequencies which matter, under 20kHz, there is clearly a lot more high frequency noise with MQA both decoded and undecoded. Look at 15 kHz, for example, a frequency some of you younger than me might still be able to hear. MQA is not just a little bit noisier, it is a whopping 12 dB noisier! [Not that I'm claiming it is necessarily always an audible difference.]


Even after that whopping increase.... it's still more than -72dB down in the audible band. 

Quote
Even when not listening to a dead silent passage where at elevated playback this sticks out like a sore thumb as hiss, added high frequency noise can add more "air", "openness", and "sheen" to certain instruments in certain recordings, in fact I believe he mentioned the sound he heard was "more sparkly".


Even when it's -72dB or more  down?  I rather expect such effect would be minor at best and difficult to ABX, at normal listening levels, ...which, not surprisingly, is what he reports. 

So I'm not sure what battle you think you are fighting here.

Re: Meridian Audio's new... sub-format called MQA.

Reply #279 –
I guess Meridian shapes the noisefloor this way by purpose. It may be one of the 'new found psycho acoustic elements' they use. In short: shaped noise
If you can hear that as easily as hiss as you claim there is no way to bit reduce with noise-shaped dither 24bit to 16bit. For that you can hear the noise when amplifying silent parts but not when music plays.
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

Re: Meridian Audio's new... sub-format called MQA.

Reply #280 –
Yes, maybe, krabapple. Remember the song he used is quiet and dynamic and the (only very occasional) peaks are far away [down] from 0dBFS from the get go (13dB), not to mention the typical, average RMS level which is even farther still, like around 26 dB down. Did he set his volume by listening to a loud track and then left it there and moved on to Blagutten without touching the volume knob? I bet not. He probably set the volume while listening to this quiet track itself [i.e. he cranked it up]. Also, he listens in an environment not all of us can easily replicate where hearing faint things is easier than for most of us:
Quote
very quiet room using a silent fanless computer setup

The reason we aren't usually troubled by noise in recordings during quiet passages, at such low levels, is not because it is too faint to detect, it is because it is almost always masked by typical environmental room noise [computer, HVAC, street noise, wind noise against the windows, etc.]. So his listening environment has some clear advantages for hearing faint noise.

In the right environment humans can hear sounds which are about 120 dB down from their pain threshold or 85 dB down from what this chart calls "average home hi-fi level".

I remember back in the days of cassette decks that even Dolby's stronger type C NR, which could provide a noise level of say -72 dBA wasn't good enough for me to eliminate noise under all conditions. That's why switching on dbx instead made a difference, albeit with incompatibility issues for playback in some scenarios.


Re: Meridian Audio's new... sub-format called MQA.

Reply #281 –
When I hastily recorded the Blagutten song yesterday from the web's sample [I did not download it because I am cheap] it was merely to assess what sort of a song it was. I didn't think I would be posting it or images of its dynamics so I did not attempt to record it at a level with fairly accurate absolute levels, in fact I didn't set anything I just hit the record button in Audacity (which I barely know how to use). So don't make any judgements about the absolute levels from this image. Still, I think this image says a lot about the dynamic nature of this song compared to typical compressed music these days, so I thought to post it for all to see:

The song is quiet, especially the opening 17 seconds, yet gets very dynamic too, with peaks which are markedly higher than the average level. There is also a moment or two within this first minute of the song where it fades down to almost complete silence, albeit only briefly.

Re: Meridian Audio's new... sub-format called MQA.

Reply #282 –
You would hope that decoded MQA vs undecoded MQA would be a little more different. Maybe it is, in the time domain. Maybe it is for other content. Looking at that spectral view, taking it at face value, decoding seems a little pointless.

It would be interesting to send some impulses through a complete chain and see what comes out.

Don't TOS8 me. I'm not talking about what it sounds like. I'm talking about seeing what the algorithm is doing.

Cheers,
David.

Re: Meridian Audio's new... sub-format called MQA.

Reply #283 –
The whole MQA paradigm makes no sense;  and I don't think it's really supposed to make sense. 
the more people are confused they less they can criticize it technically for all it's LOSSY flaws. 
It's pretty much just a ploy to create demand, infrastructure, support, and money for a product nobody needed. 
Be a false negative of yourself!

Re: Meridian Audio's new... sub-format called MQA.

Reply #284 –
The same is true for anything greater than 48kHz, 18-bit* as a delivery format.  This is just another effort to convince the masses who will always "want the best" that they need to have "more."

That it can be ABX-ed is just short of proof that the entire thing is smoke and mirrors since at least some of the difference between the original source and the undecoded output can be captured at 16/44.1.  That some of the difference between the original source and the decoded output can also be captured at 16/44.1 really demonstrates the absurdity of the whole thing!

It shouldn't come as a surprise that the sample presented by mzil was obviously not subjected to the standard dynamic range mutilation, either.

(*) 44.1kHz, 16-bit if we're talking about something other than specialized signals presented to people with >20 kHz sensitivity and/or highly dynamic content played in the quietest of environments which is likely combined with gain-riding.

Re: Meridian Audio's new... sub-format called MQA.

Reply #285 –
Quote from: Nystagmus on
The whole MQA paradigm makes no sense;  and I don't think it's really supposed to make sense.
I think that's a bit unfair. Clearly it's supposed to make money, but it's also supposed to solve a (just about ABXable under pretty extreme conditions) problem with filtering at 44.1kHz, without having to transmit higher sample rate content.

If you look back at the AES papers from Bob Stuart, Peter Craven - and also some papers from Malcolm Hawksford and even Michael Gerzon - you can see some of these ideas coming for a very long time. Burying "correction" data within 16/44.1 audio isn't a new idea, and neither is the suggestion that the filters used with CD quality audio need some work.

I'm sure the technology is supposed to make sense*, and work. I'm surprised at that spectral plot.

* ignoring the level of audibility, and other non-proprietary ways of achieving a similar effect.

Cheers,
David.

Re: Meridian Audio's new... sub-format called MQA.

Reply #286 –
Quote from: 2Bdecided on
I think that's a bit unfair. Clearly it's supposed to make money, but it's also supposed to solve a (just about ABXable under pretty extreme conditions) problem with filtering at 44.1kHz, without having to transmit higher sample rate content.
It is not even clear that the "problem" still exists at 48 kHz, which is what MQA uses.

Quote
If you look back at the AES papers from Bob Stuart, Peter Craven - and also some papers from Malcolm Hawksford and even Michael Gerzon - you can see some of these ideas coming for a very long time. Burying "correction" data within 16/44.1 audio isn't a new idea, and neither is the suggestion that the filters used with CD quality audio need some work.
It is all well to look into these filtering issues, I don't think anyone would have a problem with that. If we understand exactly what matters in the design of reconstruction filters, we can all benefit. If there are mistakes that can be made with potentially audible effects, we can learn how to avoid them.

I was and still am prepared to go along with this apodizing filtering thing, and was hoping for some convincing arguments and blind tests to confirm that there's really a valid point. I'm not yet convinced that there is, but I'm prepared to change my mind given sufficient evidence.

I am rather less impressed by the timing-related arguments, which are supposed to point towards the need for higher sampling rates. They are either hopelessly vague or in ignorance of basic properties of sampling. The scientific papers by Stuart et.al. which I've seen, tiptoe around the problem in a way that makes me believe that they know what they are talking about, but studiously shape their prose to make the reader suspect that there is something substantial, without giving any tangible evidence. I have to say that I have a rather dim view of that.

Re: Meridian Audio's new... sub-format called MQA.

Reply #287 –
Quote from: 2Bdecided on
Clearly it's supposed to make money
Ok, we agree on one thing.

Quote from: 2Bdecided on
but it's also supposed to solve a (just about ABXable under pretty extreme conditions) problem with filtering at 44.1kHz, without having to transmit higher sample rate content.
Right. The BS paper apparently showed pathological 44.1k is detectable. Hurray.
Meridians own manual states non-pathological 44.1k is "transparent".
What's it gonna be mate?
How's MiracleQA going to fix the zillion CDs out there, imagined to have embedded innumerable filter "problems"?
Wait, remasters you say? Why not with per Meridian manual, "transparent" 44.1k filtering this time? Or are you suggesting their manual is wrong, no such thing as transparent 44.1k?

cheers,

AJ
Loudspeaker manufacturer

Re: Meridian Audio's new... sub-format called MQA.

Reply #289 –
Having skimmed through it, I couldn't avoid some blood pressure increase. Stuart really has converted bullshitting into an art form.

To pick just one example of his style of arguing, here's how he deals with Archimago's lengthy blog post from January (the "Editor" of Stuart's response seems to have missed that the URL has changed, see Q82):
  • He cryptically suggests in footnote 13 that Archimago prevented him from posting corrections. Yet, there's a comment section following Archimago's article, and there's no reason given why he couldn't have used this means to post his objections. Archimago obviously looks at those comments, and he appends new information to his article in the form of updates. I regard Archimago's behavior as exemplary, and if Stuart has any issue with it, he ought to make clear what his accusations are, and provide some evidence.
  • He paraphrases the assertions such that he can more easily dismiss them and make the asserter look silly. Try reading Archimago's blog post and find where he writes what Stuart has paraphrased.
  • He criticises Archimago for assuming that the Nielsen example was from a 16-bit master (A82 part iii). This is a particularly telling example. The Nielsen piece was listed among the free examples for comparing MQA to the original, which was described to be a 44/16 DAT master. As it turns out, you needed to read the narrative given on the detailed page for this release to realize that we are actually dealing with two separate masterings. This was far from obvious, yet Archimago became aware of it not long after he wrote his article, as evidenced by his Addendum 3 at the bottom of the page. That's more than two months ago, but Stuart doesn't seem to have noticed it, yet he has no problem leaving Archimago in a bad light for not immediately becoming aware of the issue. Anyway, note that comparing two separate masterings with each other is not particularly useful when one wants to see what MQA does, so this example puts a shadow over the "HiRes Download - test bench" list page provided by 2L.
  • Even when the Nielsen piece was remastered to 24 bits before MQA encoding, it still obviously originates from a 16-bit source. Such remastering can't magically remove the noise that was there before. They allegedly did some processing derived from the measured behavior of the converter that was originally used for recording, in order to change the temporal behavior (their time-smear thing), but Stuart doesn't explain how this makes the lower 8 bits not noise. The result, as Archimago points out, compresses less well with FLAC, in addition to being more data to begin with. I don't see anything wrong with Archimago's comments.

There's a lot more (don't get me started on Stuart's diagrams), but what I described already suffices to demonstrate the ethics in force here.

Re: Meridian Audio's new... sub-format called MQA.

Reply #290 –
Quote from: pelmazo on
Bob Stuart has answered the questions that were collected by the ComputerAudiophile forum:

http://www.computeraudiophile.com/content/694-comprehensive-q-mqa-s-bob-stuart/
That's an awful lot of words to simply say "You're too stupid to understand our great format, and even if you weren't I'm not going to tell you what it does".

Re: Meridian Audio's new... sub-format called MQA.

Reply #291 –
Too mindlessly moronic to read it through to the end
Loudspeaker manufacturer

Re: Meridian Audio's new... sub-format called MQA.

Reply #292 –
Quote from: ajinfla on
Too mindlessly moronic to read it through to the end

As I mentioned earlier in the thread, this is par for the course. I am surprised that others haven't really laid into the technical sections of their papers, which consist of the most outrageous half-truths, insinuations and semi-citations in an effort to push this ridiculous "revolution in psychoacoustics proves audiophiles were right all along" nonsense. It's like nailing snake oil jelly to a wall.

Re: Meridian Audio's new... sub-format called MQA.

Reply #293 –
Much as I respect the technical expertise of members of this forum, my issue is so much more low-end, naive, newbee, etc.  When Robert Hartley of Absolute Sound goes so far out on a limb in favor of MQA, I feel I need to judge the sound for myself.  I'm sure there are better ways, but I bought a Meridian Explorer2, hooked it up, downloaded a host of MQA-encoded tracks, and let it rip.  The sound was OK, but the lights on my Explorer2 told me that the output of the DAC was CD-level PCM.  I called Meridian in England, and they recommended Foobar 2000, which would not convert the MQA signal to PCM (or similar evil) before sending it to the Explorer2.  I followed the few instructions they gave me, and the result was:  failure.  The three little lights remained stark white, indicating a PCM signal with no MQA encoding.

As always, I'm eager to take the blame.  I must have set up Foobar 2000 incorrectly.  Can anyone tell me whether there's a method, even a trick, for configuring Foobar 2000 to send the MQA signal from my computer hard disk to the DAC?  I would be extremely grateful.

Jeffrey Steingarten
Food Critic, VOGUE magazine, NYC

Re: Meridian Audio's new... sub-format called MQA.

Reply #294 –
I am not familiar with their instructions.
All you should need is bit-perfect playback. The easiest way to ensure that is by using the wasapi output plugin. On foobar's preferences - output page select "WASAPI (event): your device" with highest supported output bit depth supported by the device.
Make sure you have no active DSPs on the DSP manager preferences page and the volume slider in the main windows set to 0 dB (100%).

Depending on how "great" the Explorer2 is you may even have to set the system volume to 100%.

If that doesn't work you could also try ASIO output ...
"I hear it when I see it."

Re: Meridian Audio's new... sub-format called MQA.

Reply #295 –
Quote from: jeffrey@vog.com on
Much as I respect the technical expertise of members of this forum, my issue is so much more low-end, naive, newbee, etc.  When Robert Hartley of Absolute Sound goes so far out on a limb in favor of MQA, I feel I need to judge the sound for myself. 

That's an admirable sentiment, but: Robert Harley is not a credible authority on the nuts and bolts of audio, or on performing listening tests.  He's a pseudoscientist at best and a huckster at least.  I wouldn't embark on any investigations of sound quality on the basis of his recommendation (or disrecommendation).

Re: Meridian Audio's new... sub-format called MQA.

Reply #296 –
Quote from: xnor on
I am not familiar with their instructions.
All you should need is bit-perfect playback. The easiest way to ensure that is by using the wasapi output plugin. On foobar's preferences - output page select "WASAPI (event): your device" with highest supported output bit depth supported by the device.
Make sure you have no active DSPs on the DSP manager preferences page and the volume slider in the main windows set to 0 dB (100%).


He might have to install the WASAPI plugin first... if so here it is:

https://www.foobar2000.org/components/view/foo_out_wasapi

Re: Meridian Audio's new... sub-format called MQA.

Reply #297 –
That's MQA doing one of the things it was designed to do: proving that you have a bit perfect copy of what was created in the studio arriving at your DAC (or proving that you don't, in this case!).

For all the rest of it, I like that idea.

(Yep, I know it can be a bit perfect copy of a nasty brickwalled (or otherwise hopeless) recording ;)  )

Cheers,
David.

Re: Meridian Audio's new... sub-format called MQA.

Reply #298 –
Quote from: 2Bdecided on
That's MQA doing one of the things it was designed to do: proving that you have a bit perfect copy of what was created in the studio arriving at your DAC (or proving that you don't, in this case!).
As does any other proprietary, patented, closed, or obscure format or even just watermark. Heck, a simple checksum that a studio publishes would be enough and wouldn't require tampering with the audio data at all.
Bit-perfect playback is also easy to set up.
"I hear it when I see it."

Re: Meridian Audio's new... sub-format called MQA.

Reply #299 –
In the end this also means all MQA DACs, always and ever, no matter what vendor must sound absolutely the same.
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!