Stereophile Series on MQA Technology

[mansr]

Just now, GUTB said:

Can you state that definitively?

Yes. It's simple maths.

[Rt66indierock]

15 hours ago, Indydan said:

 

Was Jason Victor Serinus serious when he said that? Or, was it just a joke?

 

From what you are saying, some of the Stereophile people know you. Did Stereophile know your true identity when they called you a troll? If they did, that would make it kind of personal...

 

You seem to be in the audio business, probably at the studio level, engineering, etc. But, I admit I don't know your true identity. I don't know if it is common knowledge for other posters. I am not trying to make you identify yourself; that is your business if you do or don't. 

 

 

 

Jason was serious. Certainly John Atkinson, Jason Victor Serinus and Herb Reichert of Stereophile know me. I knew it would become personal when facts that MQA Ltd didn’t want to be discussed in public started to be. The supporters of MQA have nothing left other than to attack opponents of MQA.

 

I’m not in the audio business. I did consult in the broadcasting industry mainly radio for 15 years. Today I spend my time musical professional time time with emerging acts and try and match people to get the best recording environment. Many people know who I am but I use Rt66indierock for a reason. I don’t like people to appeal to authority in hi-end audio so I don’t want people to appeal to me. My words, reasoning and analysis should be able stand on their own.  

[barrows]

Here we go!

 

My concerns with MQA:

 

1.  If it gains wide acceptance it may reduce the availability of real, uncompressed hi res audio files, as record producers very

     well may prefer to release DRMed MQA versions instead.

 

2. MQA almost entirely requires consumers to purchase new hardware to take full advantage of it, when the complete "unfold"

    could be offered as software to be implemented in playback softwares.

 

3. MQA marketing is clearly disingenuous, the name itself screams BS loud and clear: who is "authenticating" what exactly?

 

4. DAC manufacturers who want to implement MQA are forced to accept MQA's world view on digital filtering as being "best",

    despite their own experience designing their own filter approaches, this reduces creativity and additional potential advances

    in audio playback performance.

 

Potential benefits of MQA:

 

1. If MQA is widely adopted and replaces MP3 for average non-audiophile people, then those people will get better SQ, and will

    very likely engage more deeply with music because of it.

 

2. Smaller files (although still lossy), I do not consider this that meaningful in our age where storage is not much a problem, but 

    I can see how this is an advantage for streaming from cloud based sources.

 

Anyone who is testing MQA needs to use the following approach, anything less will not be a valid test:

 

1.  The music files must be from the same master, testing MQA vs. a rebook source is absurd in the extreme. Example: use 

     a 24/192 native file vs. a MQA version of the exact same file for the comparison.

 

2.  The same DAC must be used, and this DAC's standard (non-MQA) filters need to at least approximate the MQA filter.  Or,

     alternative filters could be used which might be better than MQA, as long as this is disclosed by the tester; analysis of the

     filter responses needs to be included for full transparency.  There are still potential problems, as the DAC in question may be

     optimized to sound best with MQA (re analog reconstruction filter, etc). 

      

[psjug]

1 hour ago, mansr said:

Yes. It's simple maths.

Explained here:  http://science-of-sound.net/2016/02/time-resolution-in-digital-audio/

"By the way, the samplerate doesn’t even show up in the equations!"

Bob Stuart even says this is not what he means by temporal blur, although I'm not sure what he does mean.

[mansr]

34 minutes ago, barrows said:

2. Smaller files (although still lossy), I do not consider this that meaningful in our age where storage is not much a problem, but I can see how this is an advantage for streaming from cloud based sources.

But MQA files are not smaller than standard FLAC at equivalent resolution. In fact, they are quite a bit larger.

[#Yoda#]

59 minutes ago, barrows said:

1.  The music files must be from the same master, testing MQA vs. a rebook source is absurd in the extreme. Example: use 

     a 24/192 native file vs. a MQA version of the exact same file for the comparison.

The only largely reliable source to get MQA and HiRes from the same Master are the records published by the label 2L.

 

All MQA files from the major labels are not reliable from the same master as the HiRes, even if it is most likely that MQA Ltd. used the existing HiRes files as source for the conversion to MQA (is it still really MQA with the supposed benefits then?), maybe except the view albums that enjoyed a special mastering and are usually used for MQA presentations. 

[barrows]

30 minutes ago, #Yoda# said:

The only largely reliable source to get MQA and HiRes from the same Master are the records published by the label 2L.

 

All MQA files from the major labels are not reliable from the same master as the HiRes, even if it is most likely that MQA Ltd. used the existing HiRes files as source for the conversion to MQA (is it still really MQA with the supposed benefits then?), maybe except the view albums that enjoyed a special mastering and are usually used for MQA presentations. 

Agreed!  And this is why we need to be very careful about comparisons, most I have seen are apples to oranges and not valid.

[beetlemania]

58 minutes ago, barrows said:

Agreed!  And this is why we need to be very careful about comparisons, most I have seen are apples to oranges and not valid.

 

Why did John Atkinson do this (see JA's reply in discussion)?

Quote

As I used 3 pairs of microphones and 2 types of A/D converter to produce the mixdown—see the photos—as well as the mixdown I sent Bob Stuart each of the 3 mike-pair recordings, including an impulse response recording for each, and full details of the mix.

 

He sent MQA the raw files including the frickin' impulse response! And we're supposed to believe he was comparing his PCM master to an MQA version of the same . . .

[mansr]

How does one obtain an impulse response of a microphone when no such thing as an acoustic impulse is possible?

[gmgraves]

3 hours ago, barrows said:

Here we go!

 

My concerns with MQA:

 

1.  If it gains wide acceptance it may reduce the availability of real, uncompressed hi res audio files, as record producers very

     well may prefer to release DRMed MQA versions instead.

 

2. MQA almost entirely requires consumers to purchase new hardware to take full advantage of it, when the complete "unfold"

    could be offered as software to be implemented in playback softwares.

 

3. MQA marketing is clearly disingenuous, the name itself screams BS loud and clear: who is "authenticating" what exactly?

 

4. DAC manufacturers who want to implement MQA are forced to accept MQA's world view on digital filtering as being "best",

    despite their own experience designing their own filter approaches, this reduces creativity and additional potential advances

    in audio playback performance.

 

Potential benefits of MQA:

 

1. If MQA is widely adopted and replaces MP3 for average non-audiophile people, then those people will get better SQ, and will

    very likely engage more deeply with music because of it.

 

2. Smaller files (although still lossy), I do not consider this that meaningful in our age where storage is not much a problem, but 

    I can see how this is an advantage for streaming from cloud based sources.

 

Anyone who is testing MQA needs to use the following approach, anything less will not be a valid test:

 

1.  The music files must be from the same master, testing MQA vs. a rebook source is absurd in the extreme. Example: use 

     a 24/192 native file vs. a MQA version of the exact same file for the comparison.

 

2.  The same DAC must be used, and this DAC's standard (non-MQA) filters need to at least approximate the MQA filter.  Or,

     alternative filters could be used which might be better than MQA, as long as this is disclosed by the tester; analysis of the

     filter responses needs to be included for full transparency.  There are still potential problems, as the DAC in question may be

     optimized to sound best with MQA (re analog reconstruction filter, etc). 

      

You know, I don't doubt for a moment that MQA can have audible side effects, I also don't doubt that it isn't all it's cracked-up to be. But, ultimately, is it better or worse than MP3? Where I see MQA's ultimate worth is as a replacement for for very lossy compression of streaming music - especially over Internet radio. Now, this last summer, the BBC broadcast the annual Proms  concerts live as 16/48 FLAC files to those who had installed the correct web browser to be able to decode them. I was able to compare the broadcasts with the 192kbps MP3 live broadcasts of the Boston Symphony over WCRB's Internet feed. There was no doubt that as listenable as streaming 192kbps MP3 is, that the BBC's FLAC files sounded much better, with no audible artifacts vs the Boston concerts with few artifacts. It was a big step forward. On the other hand, the Proms concerts were still only slightly better than CD quality in a time when streaming high-definition audio of at least 24-bit/88.2 KHz should be available as almost a commonplace occurrence. In my estimation, even if MQA is not good enough to replace current music storage standards, if it turns out to allow high-definition music to be broadcast in the same bandwidth budget that now constrains Internet radio to MP3 or CD-quality FLAC, and without the all too audible artifacts that accompany MP3, then it should be embraced by the Internet radio broadcasters, forthwith and probably over-the-air digital FM as well. 

[Miska]

11 minutes ago, gmgraves said:

Where I see MQA's ultimate worth is as a replacement for for very lossy compression of streaming music - especially over Internet radio.

 

It just does worse job on that than standard FLAC with equivalent resolution. So what's the point?

 

11 minutes ago, gmgraves said:

if it turns out to allow high-definition music to be broadcast in the same bandwidth budget that now constrains Internet radio to MP3 or CD-quality FLAC

 

MQA doesn't do that.

 

[Miska]

7 hours ago, GUTB said:

The little Pro-Ject box is a vastly superior DAC to the Node 2, and its MQA implemention may also be much better.

 

I have the little Pro-Ject box and the original hires certainly sounds better through it than the MQA'd version.

 

And MQA sounds better through it with bare "core decoding" in software, when using one of the better ESS DAC chip filters instead of the poor MQA ones. And even better with the software decoded MQA upsampled to DSD512. But that is still not as good as the original non-MQA hires upsampled to DSD512 through it.

 

[gmgraves]

5 minutes ago, Miska said:

 

It just does worse job on that than standard FLAC with equivalent resolution. So what's the point?

 

 

MQA doesn't do that.

 

 

The point is that a say, 24/88.2 Flac file would take up far more bandwidth than a similar MQA file. Add to that the fact that someone without the ability to decode FLAC on the fly can't listen to the file at normal CD quality (think of it as an audio analogy to early color television (at least here in the USA). People without color sets could still watch the programming in Black and white). 

[Miska]

14 minutes ago, gmgraves said:

The point is that a say, 24/88.2 Flac file would take up far more bandwidth than a similar MQA file.

 

No it certainly doesn't... MQA doesn't hold 24-bit worth of resolution, when you encode FLAC at equivalent resolution that the MQA is, the result consumes less bandwidth than the MQA file.

 

Reason is very simple, MQA adds random-looking noise to the data which is then put through FLAC compression and as a result it compresses poorly due to this random noise.

 

MQA reduces word length based on the amount needed for the upper octave content folding. More there is content to fold, less there is word length left.

 

14 minutes ago, gmgraves said:

Add to that the fact that someone without the ability to decode FLAC on the fly can't listen to the file at normal CD quality

 

1) MQA is so far always delivered as a FLAC, so if you cannot decode FLAC on the fly you cannot listen MQA either

2) Any proper player software can convert hires to CD-resolution if needed of some ancient DAC. And if not, if nothing else, the OS can do it for you, macOS, Windows and Linux have that capability (as well as iOS and Android).

 

[semente]

On 10/12/2017 at 9:01 PM, #Yoda# said:

Why should I do? Most members here are doing not as well and it doesn't matter concerning the MQA topic and my posts. But I can assure you, most of my equipment has been reviewed by Stereophile, Audiostream or InnerFidelity very positve and some "Product of the Year". 

Does anyone know the annual cost of a "Product of the Year" award?

[#Yoda#]

32 minutes ago, gmgraves said:

You know, I don't doubt for a moment that MQA can have audible side effects, I also don't doubt that it isn't all it's cracked-up to be. But, ultimately, is it better or worse than MP3? Where I see MQA's ultimate worth is as a replacement for for very lossy compression of streaming music - especially over Internet radio. Now, this last summer, the BBC broadcast the annual Proms  concerts live as 16/48 FLAC files to those who had installed the correct web browser to be able to decode them. I was able to compare the broadcasts with the 192kbps MP3 live broadcasts of the Boston Symphony over WCRB's Internet feed. There was no doubt that as listenable as streaming 192kbps MP3 is, that the BBC's FLAC files sounded much better, with no audible artifacts vs the Boston concerts with few artifacts. It was a big step forward. On the other hand, the Proms concerts were still only slightly better than CD quality in a time when streaming high-definition audio of at least 24-bit/88.2 KHz should be available as almost a commonplace occurrence. In my estimation, even if MQA is not good enough to replace current music storage standards, if it turns out to allow high-definition music to be broadcast in the same bandwidth budget that now constrains Internet radio to MP3 or CD-quality FLAC, and without the all too audible artifacts that accompany MP3, then it should be embraced by the Internet radio broadcasters, forthwith and probably over-the-air digital FM as well. 

Probably you missed the conversation in this thread:

JB Radio2, one of my favourite internet radio stations, is streaming their program in FLAC format for more than a year now. The current URL is http://199.189.87.9:10999/flac.m3u. Actually they are sending an upsampled 320 kbps mp3 signal as far as I know, I think because of digital right issues. This example shows that in principle it is possible for internet radio stations to stream in at least cd quality without significant drop-outs, even for a small non profit radio station.

[semente]

4 hours ago, mansr said:

But MQA files are not smaller than standard FLAC at equivalent resolution. In fact, they are quite a bit larger.

That must be the authenticated digital signature of the Authenticator.

[#Yoda#]

32 minutes ago, semente said:

Does anyone know the annual cost of a "Product of the Year" award?

I'm buying nothing just because of a review in a magazine or achieved rewards. The only instance I really trust for evaluating HiFi equipment are my ears and the signal processing unit in between. Fortunately I'm living in an area with many well-assorted, customer friendly HiFi dealers and the currently most important HiFi show to make my own judgements.

[semente]

5 minutes ago, #Yoda# said:

I'm buying nothing just because of a review in a magazine or achieved rewards. The only instance I really trust for evaluating HiFi equipment are my ears and the signal processing unit in between. Fortunately I'm living in an area with many well-assorted, customer friendly HiFi dealers and the currently most significant HiFi show to make my own judgements.

Don't worry, I understood your reference to the award in response to Gutburg, since he will take nothing less than high-end or highly touted equipment...

[Archimago]

1 hour ago, gmgraves said:

You know, I don't doubt for a moment that MQA can have audible side effects, I also don't doubt that it isn't all it's cracked-up to be. But, ultimately, is it better or worse than MP3? Where I see MQA's ultimate worth is as a replacement for for very lossy compression of streaming music - especially over Internet radio. Now, this last summer, the BBC broadcast the annual Proms  concerts live as 16/48 FLAC files to those who had installed the correct web browser to be able to decode them. I was able to compare the broadcasts with the 192kbps MP3 live broadcasts of the Boston Symphony over WCRB's Internet feed. There was no doubt that as listenable as streaming 192kbps MP3 is, that the BBC's FLAC files sounded much better, with no audible artifacts vs the Boston concerts with few artifacts. It was a big step forward. On the other hand, the Proms concerts were still only slightly better than CD quality in a time when streaming high-definition audio of at least 24-bit/88.2 KHz should be available as almost a commonplace occurrence. In my estimation, even if MQA is not good enough to replace current music storage standards, if it turns out to allow high-definition music to be broadcast in the same bandwidth budget that now constrains Internet radio to MP3 or CD-quality FLAC, and without the all too audible artifacts that accompany MP3, then it should be embraced by the Internet radio broadcasters, forthwith and probably over-the-air digital FM as well. 

 

Remember @gmgraves, MQA is not similar to MP3 in terms of bitrates.

 

At the highest bitrate, MP3 is 320kbps. At the lowest bitrate, MQA with "hi-res" capabilities is around that of a FLAC encoded 24/44.1 stream or approximately 1,000kbps; I'm actually being very generous here with the amount of compression since the MQA encoding typically results in lower ability for lossless compression. Don't forget that many MQA streams are 24/48 or more like around 1200-1500kbps.

 

So ultimately, yes, streaming MQA better sound "better" than even the highest quality MP3 given higher bitrate (at least 3x). But why bother if we can already just stream "flat" FLAC lossless compressed 24/48 which would sound awesome because it's really capable of 24-bits rather than MQA's ~16-bit resolution, typically compresses better especially if you zero out the lowest few bits to save space, is free of any potential DRM nonsense, doesn't require special hardware, and nobody has to pay a licensing fee?

 

Obviously that "licensing fee" part is beneficial for MQA Ltd.; nobody else really. And that DRM part might be desired by some content providers.

 

Clearly the Stereophile articles have not added anything we don't know, and Jim Austin's article just shows us a few impulse responses we've known about for months if not years. Just more hot air trying to justify the unjustifiable in the eyes of consumers IMO.
 

I've said my piece previously: "Why I Don't Like MQA"; which echoes the sentiments here and elsewhere.

[arcman]

On 12/10/2017 at 4:11 PM, GUTB said:

 

I think we all know why you don’t.

 

But you’re right, it’s irrelevent.

 

What does matter is the quality of MQA. I just listened to the MQA and non-MQA version of Hotel California 40th annervisary expended edition with the Pro-Ject Pre S2; if you can’t tell the difference I don’t know what to tell you. The MQA version sounds like hi-res, and the non-MQA sounds like Redbook.

The non MQA hi res version is 24/96 while the mqa version is 24/192. I have same Dac and noticed the MQA version sounds slightly more “spacious”. Is that due to mqa or the extra resolution (96 vs 192)? bTW the 24/96 version is on Qobuz

[mansr]

27 minutes ago, arcman said:

The non MQA hi res version is 24/96 while the mqa version is 24/192. I have same Dac and noticed the MQA version sounds slightly more “spacious”. Is that due to mqa or the extra resolution (96 vs 192)? bTW the 24/96 version is on Qobuz

MQA is never more than 96 kHz. There's a 5-bit field in the MQA file that tells the decoder what rate to display. That is all.

[semente]

5 hours ago, arcman said:

The non MQA hi res version is 24/96 while the mqa version is 24/192. I have same Dac and noticed the MQA version sounds slightly more “spacious”. Is that due to mqa or the extra resolution (96 vs 192)? bTW the 24/96 version is on Qobuz

"Spacious" can very well be the result of artifacts that are correlated with the music programme. Harmonic distortion has that effect too.

[MikeJazz]

12 hours ago, Miska said:

 

I have the little Pro-Ject box and the original hires certainly sounds better through it than the MQA'd version.

 

And MQA sounds better through it with bare "core decoding" in software, when using one of the better ESS DAC chip filters instead of the poor MQA ones. And even better with the software decoded MQA upsampled to DSD512. But that is still not as good as the original non-MQA hires upsampled to DSD512 through it.

 

BTW Miska, and how much do you like the Pro-ject doing upsampling to DSD512 compared to the iFi machines...?

Sorry for the Ot.

[mevdinc]

Here we go!
https://www.stereophile.com/content/mqa-some-claims-examined