MQA Users & Discussion Thread

[Doggie Howser]

Keep us posted :)

[mr.tangco]

Enjoying A-B-ing MQA on bluesound pulse - to these ears I'm enjoying the 2L tracks immensely and am looking forward to hearing more MQA asap. Why not try for yourself? Bluesound Node Gen 1 costs only $399 on sale - http://lenbrook.com.sg/bluesound/30-bluesound-node-n100.html

[Guest AndrewC]

Let me get straight to the point… this is going to be brief.

 

 

 

 

...

Just kidding… here’s lots more ;D

 

SOME TECHNICAL BITS

• While the Explorer2 uses an XMOS chip for digital signal processing, it has a  TI PCM5102 DAC chip for D/A conversion. This is a 384kHz/24bits capable DAC chip, but for whatever reason, Meridian restricts the Explorer2 to just 192k/24bits
  
• The current Explorer2 ships with the latest MQA capable firmware, which is version 17.17
  
   
  
• There are 3 LEDs on the top, the one closest to the printed label is multi-colour capable and shows whether you’re listening to normal PCM track (White), an MQA encoded track (Green), or an MQA Artist/Studio Authenticated track (Blue).
  

 

 

LISTENING TO STANDARD PCM TRACKS

 

I didn’t expect to spend a couple of hours with the Explorer2… but as it turned out, I did, and not because it was good. The Explorer2 sounds nicely full-bodied, but when first listened to with just standard PCM tracks, my initial thought was WTF?…Is there something wrong with my IEMs or Mac? :/

 

Right out of the box the Explorer2 sounds veiled and quite rolled-off, especially if you’re used to the Chord Mojo. You’ll think there’s something wrong with your IEMs, Laptop, or cables. I switched IEMs between my Westone and FitEar just in case it was some impedance compatibility issue, they both demonstrated the same issue.

 

I let the Explorer2 run-in for about an hour while I checked a couple of reviews online; no one describes anything remotely similar to what I heard, until I re-read Stereophile’s review from last month. Reading between the lines, Jim Austin does hint at it;

…With non-MQA music, the Explorer2 was very pleasant to listen to. It conveyed plenty of information in a relaxed way, which to me is important. It always worked, and it always made nice music. It was a big step up from the headphone jack of my MacBook Pro. Though the sound was a bit soft and rounded compared to that of much more expensive converters, it was good enough for an excellent home system.

 

I think he was just being polite ;D.

 

JA’s measurement section strangely doesn’t include a standalone frequency response chart (all other recent DAC reviews do)…  So after the hour of run-in, I did a frequency response measurement which didn’t look especially out of the ordinary (except for the interestingly different responses for the various sample rates). A closer look at the 44.1k sample-rate response though, compared to the Chord Mojo’s response, starts to hint at the issue…

 

 

The Explorer2’s roll-off from 20kHz (Red) is gentle slope which suggests it’s non-linear apodizing nature, all good, but it starts almost right at 20kHz which, depending on the filter type, could have repercussions at lower octaves. I also checked the impulse response (a one-sample 44.1k/16bit -3dBFS pulse), which shows a classic Meridian apodizing FIR filter response with no pre-ringing, but a very long ringing tail.

 

In JA’s measurement, he compared the Impulse Response of the original Explorer versus Explorer2 (Figure 1, vs. Figure 2 respectively). Strangely, my measurement shows my Explorer2 behaving like an original Explorer! Specifically, nearly a millisecond long post-impulse ringing, instead of about a-third of a millisecond (according to JA’s measurement anyway).

 

Have a look;

 

 

I wondered if it was something mistaken with my measurement/method, but I crossed checked against previous measurements taken on my other DACs at home, both linear and non-linear filters with shorter durations.

 

So, whatever firmware Stereophile received and JA measured with their Explorer2, seems like it’s not what the rest of us consumers are getting :o 

 

I didn’t measure how non-linear the phase-shift of the Explorer2’s filter is (not quite sure whats the best way to do that yet ;D).

 

Even though the Explorer2 came with updated MQA firmware, I re-flashed the firmware anyway using Meridian’s dedicated App.

 

There wasn’t any appreciable difference even after the update; an overwhelming sense of a veiled and rolled-off sound, and the impulse response wasn’t changed either. Something’s definitely off.

 

All of this suggest possible impact in the audible lower octaves, which I think is what I’m hearing; much of the “air” around good quality unencoded PCM recordings are missing in action. Technically, the long post-impulse ringing shouldn’t be a problem and should be filtered off naturally by our ears, but theres no clear way to know what other sonic impact this filter has. All I can tell by listening is that it doesn’t sound like a quality DAC at all. Interestingly, listening to tracks at various bit-rates, 176.4k/24bit tracks seem to sound the best, better than 192k/24.

 

 

LISTENING TO MQA ENCODED TRAKCS

 

I used a number of test files from 2L to test out MQA encoded tracks. They make available both the original DXD track and a MQA-encoded version of the DXD (both in FLAC), along with lower bit-rate unencoded versions.

 

With MQA encoded tracks (FLAC), the Explorer2 sounds significantly closer to the quality of the Mojo.

 

 

Here’s the rub;

MQA-encoded tracks decoded on the Explorer2 isn’t even on-par with Unencoded 192k versions on the Chord Mojo! It comes close, but not quite. And it gets completely smoked when compared to the DXD Unencoded version on the Chord DAC. 

 

In other words (“>” = “better than”);

DXD on Mojo > 192k on Mojo > DXD-MQA on Explorer2 > 96k on Mojo

And for just plain PCM, Mojo > Explorer2 at every bit rate

 

IMHO, if the Explorer2 is how most people are going to hear MQA, the battle’s lost. MQA can balik kampung ;D Ironically, the only attraction is the 10:1 file size reduction, which harkens back to Meridian’s MLP days (if that sort of thing is important to you rather than sound quality ;) )

 

There’s no way to turn-off MQA decoding on the Explorer2 and it doesn’t support DXD, so it’s a little challenging to measure anything thats apples-to-apples comparison based on currently available free MQA files (I don’t plan to buy any MQA encoded material). Achimago’s review compares DXD MQA-encoded tracks against unencoded 192k version, which IMHO is only marginally useful.

 

 

BOTTOM LINE

Like a few online commentary suggests (including on Meridian focused forums), MQA has the feel of The Emperor’s New Clothes about it.  I can’t shake the feeling that Meridian has intentionally crippled regular PCM playback on the Explorer2 in order to make MQA sound “so much better” (they probably assume that the average user won’t have another portable DAC to compare it to). If MQA becomes significantly more prevalent, like for example if all higher than Redbook streams from the likes of Tidal are exclusively MQA encoded, we’re fucked :-\

 

I haven’t heard another portable DAC at the same price point as the Explorer2, so, maybe it’s just my expectations being unreasonably high, but at about exactly half the price of the Chord Mojo, the Explorer2 in comparison, is IMHO significantly less than half as good. The Explorer2 is a very disappointing piece of kit from a traditionally strong vendor - I’ve owned multiple Meridian hardware in the past, so I’m not in any way biased against them.

 

The Explorer2 though, is highly NOT recommended, though YMMV :)

 

(Edit: Removed spoiler tag on request)

[Marc]

A facebook post from DAR brought this to my attention:

 

http://www.monoandstereo.com/2016/06/the-show-newport-2016-report-new.html

 

 

Sunil Merchant, owner of Sunny Components, sponsored several rooms, and in one of those rooms he was demonstrating the MQA digital playback system with Wilson Audio Sasha speakers and Audio Research electronics.  A friend of mine in the room asked Sunil to compare MQA with Redbook CD.  To my surprise Sunil said he was barred from performing an A/B comparison of MQA versus Redbook CD or any other digital format.

 

Sunil reported that MQA is forbidding dealers from conducting A/B comparisons.  I thought he was kidding, but he is not.  I would like to understand the business rationale for that consumer-unfriendly position.

 

Wow. I simply cannot understand this. Assuming it is true, which we do not know, why would it be so?

[Krispy Audio]

Yeah this is astounding hey. Surely if you had confidence in the technology, you'd welcome an A/B comparison. 

[Hilton]

No need to think too hard on that one I would think Why else?!?!?!??! Redbook will sound better.  It's meant as an alternative to AAC or MP3 streaming/encoding.

Edited June 17, 2016 by Hilton

[New Sensations]

Yeah this is astounding hey. Surely if you had confidence in the technology, you'd welcome an A/B comparison. 

 

I'm speaking with Bob Stuart again next week I think - no doubt this Mono&Stereo comment will be discussed.

Edited June 17, 2016 by J_o_h_n

[New Sensations]

 It's meant as an alternative to AAC or MP3 streaming/encoding.

 

Really?

[MusicEar]

Enjoying A-B-ing MQA on bluesound pulse - to these ears I'm enjoying the 2L tracks immensely and am looking forward to hearing more MQA asap. Why not try for yourself? Bluesound Node Gen 1 costs only $399 on sale - http://lenbrook.com.sg/bluesound/30-bluesound-node-n100.html

 

Can tell a difference between a 24/192 and MQA mastered on same track? Does it playback fully decoded?

[Hilton]

Well even that's being generous...

 

Let's see....

I can spend another few thousand on MQA equipment and MQA content (if it ever shows up) or...

Throw another 6TB+ into my storage for a couple hundred and have a huge library of redbook and/or HD downloads at my disposal.

 

As a storage medium MQA has no place.

 

As a streaming/DRM technology it has potential. However...

 

I can already stream my whole EXISTING music library in FLAC, OGG or 320K MP3 from my home server to my phone/tablet/PC using JRiver MediaCenter on the server and JRemote tablet/phone software using a 16GB a month mobile data plan for peanuts.

I can plug in my Sony MDR-1ADAC digital headphones with built in DAC directly into my iPhone lightening connector or Android Tablet for the best possible portable experience.

 

Why on earth would I need to buy a proprietary inferior encoded file to replace the perfect files I already have? (apart from the temporal ADC/DAC timing correction claims which I'm dubious on)

 

In any case there are tests on Computer Audiophile that show you can get better sound quality vs file size out of a FLAC file that's universally compatible.

 

So I again say, what's the point?

 

MQA also uses it's own noise shaping and dither to move the noise to a supposed less audible range. 

 

MQA is just trying to cash-in on the streaming market.  

 

I'll be listening to some MQA shortly for my own experience on a very high end system, but I'm still very sceptical.

 

PS. If apple launch hi-res streaming later this year with iPhone 7 as rumoured, they'll (MQA) be dead in the water.

Edited June 18, 2016 by Hilton

[Guest Eggcup The Daft]

Really?

It's meant to be a mass market product, and so it will now be primarily a streaming product because that is where the mass market is now going. Come to that, you'll find a lot of people streaming here, too (and if Australian Internet access was better there would be a lot more, I guess).

[davewantsmoore]

why would it be so?

We could only speculate ....    but given what we know about what MQA does, we should expect the differences to be relatively small, subtle, and need a very well setup system to hear.     Not sitting in the "sweet spot" of a system is going to be orders of magnitude more distortion than anything the MQA does.

 

So the real possibilities exist that at a show, with listers sitting and standing all over the place .... that people may say they couldn't hear any difference .... they may even say they preferred the "non mqa" better.

 

These potentially results say more about the listening / auditioning than they do about the merits (audibility) of what MQA is doing.

 

 

No need to think too hard on that one I would think Why else?!?!?!??! Redbook will sound better.  It's meant as an alternative to AAC or MP3 streaming/encoding.

 

No.  That's misunderstanding of what MQA does, and how it works.    In the very worst case possibility, MQA will be identical to the original redbook  (and so, it will also sound identical).

 

Otherwise, we know MQA has the scope to make an improvement to the audio   (whether or not that is practically audible is another question) 

 

If MQA decided to alter the encoding.....   it is possible that they could intentionally make the audio sound worse than the redbook version .... and you would only access the "better" version when you used a MQA decoder.    There's no evidence that has happened  (which is not surprising as they want it to be adopted) .... and only speculation that it might happen in future.

Edited June 18, 2016 by davewantsmoore

[davewantsmoore]

So I again say, what's the point?

• Get the highest quality version of the original audio
• Remove known errors from the recording (patented)
• Package it in a format which is, compatible with the lowest common denominator PCM player (16bit 44.1khz), and an appropriate size for streaming
• Render the audio ONCE to the optimum sampling rate for the playback device  (thus avoiding any sample rate conversion), and while doing this compensate for any known issues with the playback device (patented)
• Inform the listener that the above has happened  (in a way which can't be 'faked')

 

All of these things are improvements over what we have today.   However, whether they will be significantly audible or not, remains to be seen.

 

Except the first point.... which I don't think anyone doubts has a lot of potential.   Although nothing stopping record labels letting other people have the best versions of their content either, if they decided to.

[Guest Eggcup The Daft]

A facebook post from DAR brought this to my attention:

 

http://www.monoandstereo.com/2016/06/the-show-newport-2016-report-new.html

 

 

Wow. I simply cannot understand this. Assuming it is true, which we do not know, why would it be so?

 

There's a kind of schizophrenia in what has come out to date, and I'm not sure whether they have little idea of what they are doing, are desperate, or are obfuscating.

 

The Meridian products that have been out there for a while with MQA include their very expensive active speakers, and the Explorer DAC, which is a 192/24 product. Now, you might think that they would get 192/24 based MQA files out there to work with the Explorer - but no, the little that is out there is DXD (354kHz) from 2L, and one or two other albums. You might also think that they would work to a launch date for software, but no, we have had a trickle.

 

They added to this by denying Auralic, a genuine hifi company, the right to send decoded MQA through digital outputs (it could have output the 2L files properly and shown off the format) but seem to have allowed Bluesound to, as documented in this thread  (thanks @@Sime for your work on that) when the Node2 can't actually play the high quality 2L files.

 

Selected media outlets seem to have been provided with the wherewithal to do subjective listening tests. However, anyone who would be likely to measure or perhaps has sounded sceptical - no. Even then, nobody claims a major improvement, and the releases have been controlled. MQA have conducted some public demos, and again they have either been against low bitrate lossy format files, or any benefit has been disputable.

 

So, what are they doing? MQA the company is a hybrid, the tech guys are major figures on the UK audio scene but the other half is Richemont, a marketing led Swiss watch company who couldn't give a damn about the small high fidelity market. The end game for MQA is to get their decoder into phones, lifestyle products and anywhere where they believe the mass market will compare their product to lossy compressed files. That explains the Bluesound/Auralic business - the tech guys want end-to-end, but the marketing people NEED to get onto Bluesound and Sonos, so MQA will compromise without necessarily admitting it.

 

In the meantime, the concensus among audiophiles who have heard it, could best be put as saying that MQA (in the form on the 2L testbench) isn't proving inferior to other formats, when played back on MQA enabled devices. If there's no night and day improvement, then that may be what they are aiming for with us. If nobody is screaming "foul" then they can get on to marketing to everyone else, and pick us up as they go along.

 

 

So, to answer @@Marc's question. Third party demos would blow the story of MQA being better: because while it might be technically so, a big difference isn't being heard by those audiophiles with MQA equipped DACs reporting on the forums and blogs. There's no reason that it would do any better in a show demo, unless they cheated.  I can't see Bob Stuart tolerating that.

 

One more thing to consider - much of the "informed non-audiophile" market already "knows" that 192/24 is indistinguishable from the best AAC and MP3 files as listened to by the mass market on Sonos speakers and through consumer grade headphones. If audiophiles turn up en masse, up in arms, shouting that MQA is no better than 192/24, then why would that group consider it? And aren't they the early adoptors for MQA's apparent strategy to work?

[davewantsmoore]

Now, you might think that they would get 192/24 based MQA files out there to work with the Explorer - but no, the little that is out there is DXD (354kHz) from 2L, and one or two other albums. 

 

John, this comment seems to indicate you misunderstand how MQA works.    There is no advantage to them making "192 based MQA files" for the Explorer (a 192khz and below DAC).

 

 

MQA is distributed at a base rate  (eg. 44.1khz).    At playback the MQA decoder retrieves the remaining audio, in your example up to 352.8khz.    Then it computes the base rate, plus however many samples per second extra above the base rate the target device wants  (for the explorerDAC this would be 176.4khz) ....    and it then renders the audio at this rate to the DAC.

 

The whole idea of MQA is that very high rate audio can be distributed in a form which is optimal for all DAC .... without going through a resampling stage to achieve the desired rate for playback.

 

 

Conversely, what happens without MQA?     The file is distributed at a rate of 352.8 .....  It doesn't work if we just play it  (the DAC can't accept that input format from the player).    So the player must resample the audio into a new rate.     MQAs stance is that this resampling is detrimental to quality.    In theory, it certainly is, but how audible is it really?

 

In some situations, it may be that the user is left to choose the resampling rate themselves.    In this case, they might make the poor choose of 192khz, thinking the "highest rate is best" .... or they might not know how to enable/disable the resampling settings in a player, and just give up.

[Marc]

Official statement regarding the alleged MQA restriction of A/B demos, incoming. Sorry - can't add more atm.

Sent from my iPhone using Tapatalk

[Hilton]

We could only speculate ....    but given what we know about what MQA does, we should expect the differences to be relatively small, subtle, and need a very well setup system to hear.     Not sitting in the "sweet spot" of a system is going to be orders of magnitude more distortion than anything the MQA does.

 

So the real possibilities exist that at a show, with listers sitting and standing all over the place .... that people may say they couldn't hear any difference .... they may even say they preferred the "non mqa" better.

 

These potentially results say more about the listening / auditioning than they do about the merits (audibility) of what MQA is doing.

 

 

 

No.  That's misunderstanding of what MQA does, and how it works.    In the very worst case possibility, MQA will be identical to the original redbook  (and so, it will also sound identical).

 

Otherwise, we know MQA has the scope to make an improvement to the audio   (whether or not that is practically audible is another question) 

 

If MQA decided to alter the encoding.....   it is possible that they could intentionally make the audio sound worse than the redbook version .... and you would only access the "better" version when you used a MQA decoder.    There's no evidence that has happened  (which is not surprising as they want it to be adopted) .... and only speculation that it might happen in future.

 

I don't see how this can be the case.  Spectral analysis has clearly shown that MQA at redbook sampling rates is inferior when compared to native redbook. It alters the audio. (fourth image down - look at that high frequency noise!> http://www.computeraudiophile.com/blogs/miska/some-analysis-and-comparison-mqa-encoded-flac-vs-normal-optimized-hires-flac-674/

Edited June 18, 2016 by Hilton

[Guest Eggcup The Daft]

John, this comment seems to indicate you misunderstand how MQA works.    There is no advantage to them making "192 based MQA files" for the Explorer (a 192khz and below DAC).

 

 

MQA is distributed at a base rate  (eg. 44.1khz).    At playback the MQA decoder retrieves the remaining audio, in your example up to 352.8khz.    Then it computes the base rate, plus however many samples per second extra above the base rate the target device wants  (for the explorerDAC this would be 176.4khz) ....    and it then renders the audio at this rate to the DAC.

 

The whole idea of MQA is that very high rate audio can be distributed in a form which is optimal for all DAC .... without going through a resampling stage to achieve the desired rate for playback.

 

 

Conversely, what happens without MQA?     The file is distributed at a rate of 352.8 .....  It doesn't work if we just play it  (the DAC can't accept that input format from the player).    So the player must resample the audio into a new rate.     MQAs stance is that this resampling is detrimental to quality.    In theory, it certainly is, but how audible is it really?

 

In some situations, it may be that the user is left to choose the resampling rate themselves.    In this case, they might make the poor choose of 192khz, thinking the "highest rate is best" .... or they might not know how to enable/disable the resampling settings in a player, and just give up.

Consider the Bluesound DAC that Sime was testing with. This is the real world example of MQA that someone has documented for us. That real world example is not doing what you have described. It works, for our purposes, in the same way as I've been told the Explorer 2 does. So, what happens?

 

MQA does NOT up- or over-sample, or at least no description indicates that it does that. What we have seen with the lower resolution 2L files is that the original rate is reconstructed by the MQA decoding. Sime included a graphic of the display showing that the 352.8 file is recognised at the higher rate, although it was transferred to his Chord DACs at the lower rate.  In fact, it appears from the descriptions that the original rate has to be decoded. You can't throw away the folded data in the bottom eight bits, you have to decode that  So, if you start with 352.8, that is what has to come out. If you start with 44.1, or 96, that is how it comes out. Now, if you feed that into a DAC stage that can't use it, you don't get anything. So, the data is subsequently downsampled.

 

I haven't seen an Explorer 2 working with MQA, but I understand that if you feed it 352.8 based MQA, you get 88.2 out, just like the Bluesound, because that rate won't stutter. We can also assume that the MQA part of the process is subsequently downsampling on the Bluesound, because the original "DXD" file Sime used played at a higher rate and stuttered.

 

You may not be able to tell the difference when listening, of course, but MQA's graphs show that the claimed improvement in impulse response is only fully realised at 96kHz. If we use a 96k, 176.4k or 192k based file, these DACs would give that optimum impulse response according to MQA's description, as I understand it.

 

192 would be best by Bob Stuart's indications, that he wants people to use the highest bitrate they can to create studio masters to be fed into the MQA encoder, so I would expect that to be the standard we should be using to experience MQA properly on the Explorer 2, or indeed the Bluesound and any other DAC that uses only 24/192 and MQA.

So, 88.2kHz output does not give the "full benefit" of MQA on the Bluesound or the Explorer 2, and is what these DACs are putting out when they are fed the 2L "DXD" files.

 

I can't see how the decoding can work as you describe. The encoded data is in effect a description of how to create an "audibly improved" version of the original bitrate file. The decoder has to follow those instructions or it introduces the chance of error. The change of rate has to be subsequent to the decoding.

 

If I've got the Explorer 2 wrong and it outputs a higher rate, then I apologise.

[Guest Eggcup The Daft]

I don't see how this can be the case.  Spectral analysis has clearly shown that MQA at redbook sampling rates is inferior when compared to native redbook. It alters the audio. (fourth image down - look at that high frequency noise!> http://www.computeraudiophile.com/blogs/miska/some-analysis-and-comparison-mqa-encoded-flac-vs-normal-optimized-hires-flac-674/

The specification for MQA states that the file when played back on a non-MQA device has undergone the deblurring process for the ADC used, part of the MQA process. So it would be a surprise if it was identical to a redbook equivalent, wouldn't it? They would claim that the file is improved because they have removed the ADC's effect on the signal.

 

I guess that in this case, we would have to submit the results to blind preference testing to see if it is audibly inferior. More than one person has claimed that to be the case, anecdotally. I felt it to be so with the lower resolution sources, not at all sure with the DXD based files comparing to the lower bitrate versions on the testbench. Using a DOS routine to randomise ABX,  I could only pick the Neilsen as different reliably.

[New Sensations]

...given what we know about what MQA does, we should expect the differences to be relatively small, subtle, and need a very well setup system to hear.     Not sitting in the "sweet spot" of a system is going to be orders of magnitude more distortion than anything the MQA does.

 

So the real possibilities exist that at a show, with listers sitting and standing all over the place .... that people may say they couldn't hear any difference .... they may even say they preferred the "non mqa" better.

 

These potentially results say more about the listening / auditioning than they do about the merits (audibility) of what MQA is doing.

 

 

Dave - nicely put. That's precisely how Bob Stuart explained it to me the other week.

[powerav]

I also people look for the negatives so they can justify to themselves that they do not have to buy all new equipment again.

[New Sensations]

I also people look for the negatives so they can justify to themselves that they do not have to buy all new equipment again.

 

I agree - those who would dismiss something as 'snake oil' are just as financially motivated as the company selling the damn stuff. Rubbishing something before you've even heard it keeps any potential down-the-line audition as the remotest of possibilities and the money you might spend if you indulged such an audition therefore stays in your wallet.

[davewantsmoore]

I don't see how this can be the case.  Spectral analysis has clearly shown that MQA at redbook sampling rates is inferior when compared to native redbook. It alters the audio. (fourth image down - look at that high frequency noise!> http://www.computeraudiophile.com/blogs/miska/some-analysis-and-comparison-mqa-encoded-flac-vs-normal-optimized-hires-flac-674/

 

What Jussi (Miska) shows there is that it is different ... and that is no surprise given how MQA works.     "Inferior" is much more complicated concept.

 

 

2L have been quoted as saying that most of the recordings there do not have 16bits of information in them  (the noise floor begins before 16bits) .... and this is why the MQA encoded file has the noise floor where it does.

 

Jussi infers that the difference is audible.    I would very much not expect it to be.

 

MQA infers that what Jussi is showing is not audible (which seems sane to me) .... and also other things that Jussi has not examined that are audible (he has only looked at the noise in the concept of 'resolution').

 

 

My comment about MQA being identical to redbook, was if you encoded a redbook source with the MQA encoder....  the output would be identical to the input, aside from extra noise out to 24bits.

[davewantsmoore]

Consider the Bluesound DAC that Sime was testing with. This is the real world example of MQA that someone has documented for us. That real world example is not doing what you have described. It works, for our purposes, in the same way as I've been told the Explorer 2 does. So, what happens?

 

The MQA decoder chooses the rate at which to render the audio....    as opposed to a player doing a sample rate conversion from BigRate to SmallerRate.

 

Regardless of what is actually happening the in the BlueSound .....   the comment about them needing to "make 192khz files for a 192khz DAC", isn't correct.    There is no benefit for them to do that  (in fact, there is a potential downside if the source audio is 352.8 - as they will need to use a samplerate converter - which they say is bad)

 

 

MQA does NOT up- or over-sample

 

Yes, that's correct.   It's one of the (supposed) core benefits.

 

 

Now, if you feed that into a DAC stage that can't use it, you don't get anything. So, the data is subsequently downsampled.

 

That may be a fault with how the BlueSound is working with an external DAC.     The intention is that the MQA decoder chooses the optimal rate to render the audio at for the DAC.

 

 

I haven't seen an Explorer 2 working with MQA, but I understand that if you feed it 352.8 based MQA, you get 88.2 out, just like the Bluesound, because that rate won't stutter. 

 

If that's the case, then it is simply a hardware/software issue.

 

Given that it is trivial to make a device which doesn't "stutter" processing audio at these rates.... and that price of the Explorer2.  I would be extremely surprised.

 

 

You may not be able to tell the difference when listening, of course, but MQA's graphs show that the claimed improvement in impulse response is only fully realised at 96kHz. If we use a 96k, 176.4k or 192k based file, these DACs would give that optimum impulse response according to MQA's description, as I understand it.

 

It still appears that you misunderstand.

 

If I receive MQA encoded audio in a 44.1khz 24bit container....   which is encoded from a source that is 352.8khz.     The decoder renders audio in any rate it wants to....  and as long as this rate is a 2x rate or higher  (eg.  88.2 or 96kz), then they can get "the best" impulse response in the result.

 

The whole point of MQA is we don't need to use a container that has  88.2/96 or higher rates to get the benefit.

 

 

192 would be best by Bob Stuart's indications

 

They're already doing higher than this  (it's mostly 352.8 or 384khz sampling rate audio)....  and as long as the MQA decoder is rendering at 88.2/96  (or maybe 174.6/192), then everything is peachy.

 

So, 88.2kHz output does not give the "full benefit" of MQA on the Bluesound or the Explorer 2, and is what these DACs are putting out when they are fed the 2L "DXD" files.

 

Sure.   It might be better if rendered out at 174.6 / 192.   <shrug>

 

 

 

 

I can't see how the decoding can work as you describe. The encoded data is in effect a description of how to create an "audibly improved" version of the original bitrate file. The decoder has to follow those instructions or it introduces the chance of error. The change of rate has to be subsequent to the decoding.

 

I don't really know what you mean by the underline parts, but it seems to indicate you don't understand.

 

The decoder retrieves all the extra audio information.    In our examples, that means a rate up to DXD / 352.8.    Now, it renders new audio  (not using a simple sample rate converter), at whatever rate is best for the DAC.

 

If I've got the Explorer 2 wrong and it outputs a higher rate, then I apologise.

 

I don't know.    Talking about specific DACs just confused the broad issue.    MQA can render the audio at whatever rate is best for the DAC.

Edited June 19, 2016 by davewantsmoore

[davewantsmoore]

The specification for MQA states that the file when played back on a non-MQA device has undergone the deblurring process for the ADC used, part of the MQA process. So it would be a surprise if it was identical to a redbook equivalent, wouldn't it? 

 

What Jussi was showing over on computeraudiophile didn't show, or not show, the above.     The extra noise he shown was completely to be expected, given the comments from 2L about the files (and they way the MQA encoder works)

 

I think the assumption that it is audible (when played without a MQA decoder), is shaky.

Edited June 19, 2016 by davewantsmoore