MQA? What exactly does "authenticated" mean?

[miguelito]

MQA is the new Meridian compression algorithm - MQA stands for Master Quality Authenticated.

 

I've listened to it, and it sounds good, very good. Obviously a Meridian dealer will play good examples to you, no question about that.

 

Couple of things demoed:

 

1- MQA encoded master tapes vs their CD counterpart

The examples played where much better in MQA. However, levels where not matched (CD was played way too loud) plus this really highlights the blunders of the specific CD mastering more than MQA itself. But frankly in this case, if having an "MQA" logo means that the mastering will be done more carefully, then by all means.

 

2- MQA encoded off of a high res file vs the high res file itself

In this case MQA also sounded slightly better! For this particular file that is. Now obviously this means that MQA has altered the content (the difference was not subtle as say something with less jitter would be). So MQA is acting as a filter in some way. Not necessarily bad, but now we are talking about an euphonic filter rather than lossless smart compression.

 

My problem with MQA comes when answers are not provided:

1- "Authenticated"

What does that mean exactly? Do you have every single artist "sign off" the MQA transcription? I asked details about what this exactly means and what the process is. I got "secret sauce" as the answer. What bullsh*t. I am a scientist by training (have a PhD in Physics). If you're going to be advertising something that's "authenticated" you better have an "authoritative" answer to the question of what "authenticated" exactly means.

 

2- Is the compression lossless? Do you recover the same bits as went in?

I don't know the answer to this "authoritatively" but I can tell you that demo "2" above definitely indicates that the answer is no, you cannot recover the same data.

 

Now don't get me wrong... I am all for processing that improves the sound, I am not bit-identical-obsessed. However, I have a very hard time swallowing BS.

[sdolezalek]

Not sure I can answer your questions fully, but having attended Meridian's demo event here in California I have been interested in exactly what they are doing. My understanding is as follows (and there is published stuff on the web that goes into far more detail):

 

1. They are verifying all of the steps from the microphone to the master tape and using that information to determine what filtering they should apply in order to get the sound as close as possible to the microphone feed. So I think that "authenticated" simply means they have verified the equipment provenance they are "adjusting." Now, as to what they are adjusting, I too got "secret sauce" but by pushing a bit harder it seems that they are in essence doing the same kinds of things that you or I could do in HQPlayer (by first upsampling and then carefully choosing the filters we applied), except that they are specifically choosing those filters in response to the equipment used in the recording chain. You or I could do the much of same thing if (a) we knew the provenance (hence authentication) and (b) we constructed a separate HQPlayer filtering scheme for each piece of music.

 

2. They then do apply a combination of lossless (and I believe lossy) compression selectively to the upsampled output in order to reduce the overall file size back down to something the size of a 16/44 file. Here too there is detail on the web about what they are doing and why they believe we cannot hear the compression as they apply it.

 

Lastly, I heard MQA on a Meridian system that had been specifically modified to play it (so it was hard to tell how much was MQA and how much was modifications that wouldn't show up when played on your or my system. I'm not a fan of the big Meridian loudspeakers, but parts of what i heard did sound very very good, so I plan on following their progress closely.

[avta]

As I understand it Authenticated refers to the fact that what you are getting out of your system is exactly what was put in to the file at the time it was created. I believe the process can be applied to any file format and is lossless. When the original master is created it is encoded with their software and when played back through equipment that contains MQA software it ensures that the master and playback file are exactly the same i.e. you hear what was recorded.

[miguelito]

As I understand it Authenticated refers to the fact that what you are getting out of your system is exactly what was put in to the file at the time it was created. I believe the process can be applied to any file format and is lossless. When the original master is created it is encoded with their software and when played back through equipment that contains MQA software it ensures that the master and playback file are exactly the same i.e. you hear what was recorded.

And what exactly does that mean? How is a lossless file directly from the master not "authenticated"?

[miguelito]

Not sure I can answer your questions fully, but having attended Meridian's demo event here in California I have been interested in exactly what they are doing. My understanding is as follows (and there is published stuff on the web that goes into far more detail):

 

1. They are verifying all of the steps from the microphone to the master tape and using that information to determine what filtering they should apply in order to get the sound as close as possible to the microphone feed. So I think that "authenticated" simply means they have verified the equipment provenance they are "adjusting." Now, as to what they are adjusting, I too got "secret sauce" but by pushing a bit harder it seems that they are in essence doing the same kinds of things that you or I could do in HQPlayer (by first upsampling and then carefully choosing the filters we applied), except that they are specifically choosing those filters in response to the equipment used in the recording chain. You or I could do the much of same thing if (a) we knew the provenance (hence authentication) and (b) we constructed a separate HQPlayer filtering scheme for each piece of music.

 

2. They then do apply a combination of lossless (and I believe lossy) compression selectively to the upsampled output in order to reduce the overall file size back down to something the size of a 16/44 file. Here too there is detail on the web about what they are doing and why they believe we cannot hear the compression as they apply it.

 

Lastly, I heard MQA on a Meridian system that had been specifically modified to play it (so it was hard to tell how much was MQA and how much was modifications that wouldn't show up when played on your or my system. I'm not a fan of the big Meridian loudspeakers, but parts of what i heard did sound very very good, so I plan on following their progress closely.

I asked about the equipment chain argument and whether they used that to come up with a filter that's appropriate to that chain, and a related compression method also appropriate to the chain. I was told no that is not done - which makes the initial claim that those items are taken into account rather laughable and contradictory.

 

And frankly that would be a very onerous process and I don't believe for a second studios would submit to that. Moreover, many recordings use different microphones in the same tracks, different mixing consoles in different tracks, and I very much doubt there are reliable records of all of this - so again, I would regard the claim as disingenuous.

 

As I said, I'm all for filters that will improve sound - eg HQPlayer upsampling.

 

If the argument was: "We have surveyed all of these systems, determined what pieces of the audio spectrum matter most, even analyzed some "quiet" portions of recordings for noise signatures, etc... and we have come up with a compression algorithm that we find to be the most musical in the most cases, and all of these recording engineers (list available) have listened to it and agreed. And we have patented this algorithm and this is it, and it sounds great..." then I'm sold on this.

 

Instead there's the BS about "authenticated", "recording chain", etc... It's so lame... I just can't swallow that...

[esldude]

Actually simpler than you imagine. Firstly, no versus the actual hirez file it is not lossless bit correct upon replay. They tip toe around with "perceptually lossless".

 

The authenticated part has some validity though not the way you presume on first blush. And they are using your first thought against you in the grand honored tradition of marketing speak. Say something kind of true, but sure to be misinterpreted as better. Smile and don't say too much.

 

The key part in this authenticated business is I think the fact MQA among other tricks uses subtractive dither. Potentially much more effective and leaving less residue than additive dither.

 

Subtractive Dither Approach Streamlines Data Acquisition Designs - COTS Journal

 

Subtractive dither is addition of random info prior to digitizing. You must have a way to precisely subtract that dither upon conversion at the other end. It has several benefits if done. By its nature MQA must be 'authenticated' relative to its initial digital encoding for this to function. In that sense it can assure you that it was encoded with MQA equipment. That by itself does not authenticate other provenance or quality issues. Again, sort of true, sort of makes you likely to believe more is happening than is, so it works great for marketing though of course it is an issue they will always sort of sidestep around.

[Jabs1542]

Actually simpler than you imagine. Firstly, no versus the actual hirez file it is not lossless bit correct upon replay. They tip toe around with "perceptually lossless".

 

The authenticated part has some validity though not the way you presume on first blush. And they are using your first thought against you in the grand honored tradition of marketing speak. Say something kind of true, but sure to be misinterpreted as better. Smile and don't say too much.

 

The key part in this authenticated business is I think the fact MQA among other tricks uses subtractive dither. Potentially much more effective and leaving less residue than additive dither.

 

Subtractive Dither Approach Streamlines Data Acquisition Designs - COTS Journal

 

Subtractive dither is addition of random info prior to digitizing. You must have a way to precisely subtract that dither upon conversion at the other end. It has several benefits if done. By its nature MQA must be 'authenticated' relative to its initial digital encoding for this to function. In that sense it can assure you that it was encoded with MQA equipment. That by itself does not authenticate other provenance or quality issues. Again, sort of true, sort of makes you likely to believe more is happening than is, so it works great for marketing though of course it is an issue they will always sort of sidestep around.

 

Interesting approach, and this idea does fill in a huge gaping whole with MQA. If it was the "recording chain" itself I would be highly suspicious of them actually obtaining provenance on anything more than a very small percentage of what's out there. If Acoustic Sounds, HD Tracks, and everyone else selling high res can't get enough provenance to even say what the source is how is Meridian going to go beyond this and capture the actual equipment chain.

 

I think Dennis may be on to something. You could loosely term it as "provenance" when in fact it's just an algorithm used against the source file to determine the best way to sprinkle their magic fairy dust (a.k.a. voicing). I would like to add that their psycho acoustic stuff sounds a lot like sweetener / voicing.

[Fitzcaraldo215]

No doubt there is some marketing hyperbole in the word "authenticated". But, my read of the description of the technology is somewhat simpler. Fundamentally, the MQA chain is using apodizing filters to remove filter-induced time domain ringing. A DAC can do that on playback d-a ringing by knowing its own characteristics on output, but it does not know what happened during a-d in the original recording which also inevitably induces ringing. MQA solves that by including metadata about the original a-d, which is then used in the final d-a conversion simultaneously to minimize both sources of filter ringing.

 

Not sure if this completely squares with esldude's negative dither, but it parallels it in some ways. I do not see it as dither with MQA, in that it does not inject random noise from what I see.

 

I am also quite comfortable with descriptions of psychoacoustically perceptible compression as a meams of "smart" compression. I have long wondered myself why 24 bit S/N and ultra high sampling rates wasted all that space and bandwidth carrying extra bits consisting of nothing but inaudible random noise. MQA cleverly cuts through that.

 

I understand the skepticism, but I have a decent amount of respect for Bob Stuart and what he has done over the years. Personally, on paper at least, I think the concept is brilliant. But, time will tell.

[wisnon]

Why the confusion?

I seems clear to me.

 

Its a 2 step program…master quality and authentication…2 different processes.

 

MQ is about recreating the DA filter based on the AD filters used. They claim that there is a very limited universe of AD machines and filers back in the day and its either easy to find, or the set up the DA filters to math the say 5-10 AD filter universe and apply them one after the other and test to see which nullifies ringing, etc.

 

The 2nd step is authentication. This is to assure provenance and to get the artists to approve the MQA release and assure buyers that they are not getting counterfeits, but rather what the artists and Master engineers approve. This is a digital signature embedded in the file that virtually eliminates fraud. You are guaranteed the original re-release and the MQA device lights up to show you its authenticated.

[Tony Lauck]

Actually simpler than you imagine. Firstly, no versus the actual hirez file it is not lossless bit correct upon replay. They tip toe around with "perceptually lossless".

 

The authenticated part has some validity though not the way you presume on first blush. And they are using your first thought against you in the grand honored tradition of marketing speak. Say something kind of true, but sure to be misinterpreted as better. Smile and don't say too much.

 

The key part in this authenticated business is I think the fact MQA among other tricks uses subtractive dither. Potentially much more effective and leaving less residue than additive dither.

 

Subtractive Dither Approach Streamlines Data Acquisition Designs - COTS Journal

 

Subtractive dither is addition of random info prior to digitizing. You must have a way to precisely subtract that dither upon conversion at the other end. It has several benefits if done. By its nature MQA must be 'authenticated' relative to its initial digital encoding for this to function. In that sense it can assure you that it was encoded with MQA equipment. That by itself does not authenticate other provenance or quality issues. Again, sort of true, sort of makes you likely to believe more is happening than is, so it works great for marketing though of course it is an issue they will always sort of sidestep around.

 

Perceptually lossless. Like MP3?

 

Marketing spin.

 

Technology seems like a solution to a non-problem. That some listeners prefered the MQA over the original is strong evidence that the demonstration was rigged. Caveat emptor.

[esldude]

Perceptually lossless. Like MP3?

 

Marketing spin.

 

Technology seems like a solution to a non-problem. That some listeners prefered the MQA over the original is strong evidence that the demonstration was rigged. Caveat emptor.

 

Actually, MP3 never claimed to be perceptually lossless, at least those who made the format did not claim that.

 

But yes marketing spin. Implying the authenticated nature of hires provenance while trying to sidestep the issue of this being a lossy format above 30 khz.

[esldude]

Why the confusion?

I seems clear to me.

 

Its a 2 step program…master quality and authentication…2 different processes.

 

MQ is about recreating the DA filter based on the AD filters used. They claim that there is a very limited universe of AD machines and filers back in the day and its either easy to find, or the set up the DA filters to math the say 5-10 AD filter universe and apply them one after the other and test to see which nullifies ringing, etc.

 

The 2nd step is authentication. This is to assure provenance and to get the artists to approve the MQA release and assure buyers that they are not getting counterfeits, but rather what the artists and Master engineers approve. This is a digital signature embedded in the file that virtually eliminates fraud. You are guaranteed the original re-release and the MQA device lights up to show you its authenticated.

 

I am not sure I see this. Do you have other information on it?

 

The recreation on the DA end is tied up with the AD end due to the needs to properly implement subtractive dither. That means encoding into MQA will have to follow certain parameters. It doesn't necessarily follow that the original hires file being made into MQA had the original AD filtering in the full original recording recreated. At least not unless you have some info they have released saying that. Of course their playing it close to the vest is annoying to say the least. They act like someone with something to hide while proclaiming a breakthrough.

[Tony Lauck]

I am not sure I see this. Do you have other information on it?

 

The recreation on the DA end is tied up with the AD end due to the needs to properly implement subtractive dither. That means encoding into MQA will have to follow certain parameters. It doesn't necessarily follow that the original hires file being made into MQA had the original AD filtering in the full original recording recreated. At least not unless you have some info they have released saying that. Of course their playing it close to the vest is annoying to say the least. They act like someone with something to hide while proclaiming a breakthrough.

 

Subtractive dither is a win, about 6 dB reduction in noise floor compared with TPDF dither. However, it's not something that can be implemented in the analog domain, so this means that it is a technique for getting better performance while throwing away bits, e.g. 24 bits reduced to 15 bits with subtractive dither will have about the same noise floor as 16 bits with TPDF dither. In addition, although there are fewer bits and about the same noise level the sound quality may be better because the noise will statistically independent of the music. TPDF dither is uncorrelated only in the average (1st order) and RMS(2nd order). Higher order moments are correlated. Unfortunately, none of this magic does nothing to improve the sound, as the original 24 bits are going to sound better than any noisy version. The benefit is in "saving" bits, but bits are cheap these days.

 

I place no value on "approval" by the artists and the mastering engineers. These are the same people who "approved" horrible sounding releases that had all the musical dynamics stripped out and distortion added in the interest of "loudness". But the real problem is with the customers who buy these horrible recordings. No "seal of approval" is going to fix this problem.

[goldsdad]

I place no value on "approval" by the artists and the mastering engineers. These are the same people who "approved" horrible sounding releases that had all the musical dynamics stripped out and distortion added in the interest of "loudness".

 

+1

[Fitzcaraldo215]

Actually, MP3 never claimed to be perceptually lossless, at least those who made the format did not claim that.

 

But yes marketing spin. Implying the authenticated nature of hires provenance while trying to sidestep the issue of this being a lossy format above 30 khz.

 

 

I have not read the AES paper, so I do not know for sure that they have attempted to sidestep anything. The TAS article was pretty clear that they do reduce bit depth with increasing frequency based on their studies of perception and actual signal levels in recordings. Rather than the rectangular distribution of constant bit depth vs. frequency used in PCM, actual recorded signal levels and perception are triangular, with decreasing bit depth vs. frequency. Ergo, if they are right, at least half the bits are wasted.

 

30kHz? Who uses recording mikes that go that high? Speakers that do are also few in number. I am of the impression that actual signal in that frequency range and above in hi rez PCM is little but random noise and inaudible currently. What is audible is the pushing up further in frequency of the reconstruction filter and its artifacts, making them less audible.

 

My big question concerns whether apodizing filters at 48 or 96k for 96 or 192k sampling are really significant in improving sound quality - is filter ringing that high in frequency still audible? They are saying it is. I will have to hear it myself under good auditioning conditions to be convinced how important it is.

[wisnon]

I am not sure I see this. Do you have other information on it?

 

The recreation on the DA end is tied up with the AD end due to the needs to properly implement subtractive dither. That means encoding into MQA will have to follow certain parameters. It doesn't necessarily follow that the original hires file being made into MQA had the original AD filtering in the full original recording recreated. At least not unless you have some info they have released saying that. Of course their playing it close to the vest is annoying to say the least. They act like someone with something to hide while proclaiming a breakthrough.

 

Cant recall where now, but could be from Chris here or from Mike at AudioStream or somewhere else (PSA forum?) about 2 months ago when there were reams of articles about it. If I find something, I will link you up.

 

EDIT : Interesting reading here:

Meridian Demos MQA in NYC - Page 2 - AVS | Home Theater Discussions And Reviews

 

and here:

http://www.techhive.com/article/2866274/meridian-audio-s-mqa-technology-is-the-best-product-at-ces-so-far.html

[miguelito]

No doubt there is some marketing hyperbole in the word "authenticated". But, my read of the description of the technology is somewhat simpler. Fundamentally, the MQA chain is using apodizing filters to remove filter-induced time domain ringing. A DAC can do that on playback d-a ringing by knowing its own characteristics on output, but it does not know what happened during a-d in the original recording which also inevitably induces ringing. MQA solves that by including metadata about the original a-d, which is then used in the final d-a conversion simultaneously to minimize both sources of filter ringing.

There seems to be more than this as I heard what was purportedly a high-res recording (in digital form already) and it's MQA-processed form. The MQA version sounded different - and better to my ears - so there's something in the vein of filtering done.

 

Not this this is a problem in my book - I love what HQPlayer does for redbook CD.

[miguelito]

Actually simpler than you imagine. Firstly, no versus the actual hirez file it is not lossless bit correct upon replay. They tip toe around with "perceptually lossless".

 

The authenticated part has some validity though not the way you presume on first blush. And they are using your first thought against you in the grand honored tradition of marketing speak. Say something kind of true, but sure to be misinterpreted as better. Smile and don't say too much.

 

The key part in this authenticated business is I think the fact MQA among other tricks uses subtractive dither. Potentially much more effective and leaving less residue than additive dither.

 

Subtractive Dither Approach Streamlines Data Acquisition Designs - COTS Journal

 

Subtractive dither is addition of random info prior to digitizing. You must have a way to precisely subtract that dither upon conversion at the other end. It has several benefits if done. By its nature MQA must be 'authenticated' relative to its initial digital encoding for this to function. In that sense it can assure you that it was encoded with MQA equipment. That by itself does not authenticate other provenance or quality issues. Again, sort of true, sort of makes you likely to believe more is happening than is, so it works great for marketing though of course it is an issue they will always sort of sidestep around.

All for better sound. All for filters that make the music sound more real (whatever that means).

 

Not for BS that smells rotten right out of the gate. Seriously. I expect better from Meridian.

[miguelito]

Perceptually lossless. Like MP3? Marketing spin.

Trudat.

 

Technology seems like a solution to a non-problem. That some listeners prefered the MQA over the original is strong evidence that the demonstration was rigged. Caveat emptor.

Disagree. Most people that have listened to HQPlayer upsample redbook will agree it is an improvement. MQA seems to be doing some form of this. No problem - just be honest about what it does.

[wisnon]

Meridian does it in PCM by uncovering what the AD process did. they claim that only 5-6 AD encoders were used back in the day, so they can experiment and implement those reverse filters on the master copy till they hear which one sounds right, even if they dont have provenace data for every album.

 

I read somewhre that these 5 to 6 A-D filter setups will cover some 90% of older albums out there. if this is accurate, then conceptually the big picture is easy to get.

[Tony Lauck]

 

30kHz? Who uses recording mikes that go that high? Speakers that do are also few in number. I am of the impression that actual signal in that frequency range and above in hi rez PCM is little but random noise and inaudible currently. What is audible is the pushing up further in frequency of the reconstruction filter and its artifacts, making them less audible.

 

 

I have a number of recordings that have energy above 30 kHz which is correlated with the music and so can not be random noise. I see this with orchestral music, particularly brass and cymbals. I have downloads of the late Beethoven String Quartets by the Cypress Quartet in 96/24 that have energy above 40 kHz. These recordings were made with microphones that go up to about 80 kHz. My speakers have beryllium tweeters and go up to 40 kHz.

 

If the energy in the high frequencies is not loud then it does not take a lot of bits to represent it, thus many bits do not need to be wasted. If the high frequency content is low then it is easy to predict a quite a few samples from previous samples. Lossless compression, such as FLAC, then encodes the differences between the actual and predicted using the usual data compression algorithms (e.g. arithmetic coding or LZV). If these differences are small then the encoded differences will use a small number of bits. If there were really no high frequency energy in the original, typical lossless encoders will use few bits to encode the high frequencies. Most of the actual bits that are wasted will be high frequency microphone preamp noise or possibly sigma-delta modulator noise from the original ADC.

 

My big question concerns whether apodizing filters at 48 or 96k for 96 or 192k sampling are really significant in improving sound quality - is filter ringing that high in frequency still audible? They are saying it is. I will have to hear it myself under good auditioning conditions to be convinced how important it is.

They have an AES paper to back up that these kinds of filters are audible.

 

Whether you will hear the effects of apodizing a recording will depend on the recording and your playback chain. If the original recording has no energy in the transition band of the apodizing filter that you use (e.g. because it was already apodized to a lower frequency) then your apodizing filter will not ring. It will in fact do nothing except possibly add in some dither noise at the bit depth used downstream from the filter. This can easily be demonstrated with an audio editor such as RX4 with actual music files that have been apodized. You can get a null where the maximum error is at absurdly low levels, even measured by the most conservative method, the worst case error found by examining the samples of a null file. (You can take a low sample rate file and upsample it with RX4 and this will give you an apodized file to work with at the higher sample rate. If you want to do null tests of further filtering on the apodized file you will want to use linear phase filters. In Rx4, these filters have zero delay and you can get very deep nulls when comparing the results created by these filters on an apodized file. This will not work with minimum phase filters which have delay. Also, note that earlier versions of the iZotope SRC have a constant sub-sample delay and you may not get good nulls when comparing these filters.)

 

There is no need to apodize an already apodized file. If an original recording was not apodized the chances are good that it already has distortion from aliasing. Once this distortion has been encoded into the file then it will be impossible to remove without also removing (some) high frequencies in the recording. Whether apodizing will be recognized as an improvement (e.g. less digital glare due to inharmonic components) or as deterioration (loss of high frequencies and hence "air") will be a matter of taste. It seems dubious that Meridian has some "secret sauce" that will make these tradeoffs automatically.

[Jud]

I have a number of recordings that have energy above 30 kHz which is correlated with the music and so can not be random noise. I see this with orchestral music, particularly brass and cymbals. I have downloads of the late Beethoven String Quartets by the Cypress Quartet in 96/24 that have energy above 40 kHz. These recordings were made with microphones that go up to about 80 kHz. My speakers have beryllium tweeters and go up to 40 kHz.

 

If the energy in the high frequencies is not loud then it does not take a lot of bits to represent it, thus many bits do not need to be wasted. If the high frequency content is low then it is easy to predict a quite a few samples from previous samples. Lossless compression, such as FLAC, then encodes the differences between the actual and predicted using the usual data compression algorithms (e.g. arithmetic coding or LZV). If these differences are small then the encoded differences will use a small number of bits. If there were really no high frequency energy in the original, typical lossless encoders will use few bits to encode the high frequencies. Most of the actual bits that are wasted will be high frequency microphone preamp noise or possibly sigma-delta modulator noise from the original ADC.

 

They have an AES paper to back up that these kinds of filters are audible.

 

Pardon the OT: I have lots of versions of the late quartets, and the Cypress is my absolute favorite for both performance and sound quality. (I don't have the Takacs.)

[Tony Lauck]

Pardon the OT: I have lots of versions of the late quartets, and the Cypress is my absolute favorite for both performance and sound quality. (I don't have the Takacs.)

 

More OT...

 

Perfromance wise, the Cypress Quartet late Beethoven quartets are definitely up there. It might be the 1952 Budapest mono recordings on Columbia LP beat it musically, but I've not listened to these for a few decades. These were the version I grew up on. When it comes to sound quality, however, the Cypress Quartet recordings are definitely the best. As I understand it they were made by the spouse of one of the musicians. It shows what can be done with simple but good equipment, a good venue, and a suitable combination of talent and luck on the part of the engineer.

 

Back in my college days, my requirement for a girlfriend was that she had to like classical music, more specifically to appreciate the late Beethoven quartets. In the end I married a pianist who loved and performed the late Beethoven piano sonatas... Unfortunately, the only suitable recording I have of her late Beethoven is the Op. 101 from a recording in 1977 made in Jordan Hall, Boston.

[bmoura]

MQA is the new Meridian compression algorithm - MQA stands for Master Quality Authenticated.

 

I've listened to it, and it sounds good, very good. Obviously a Meridian dealer will play good examples to you, no question about that.

 

I'd describe MQA as an encode/decode system vs. a compression algorithm. The mastering engineer uses Meridian-licensed equipment to encode the music which includes "pre-distortion" and storing dynamic range information in the noise floor of the CD Quality spectrum. The original music is recovered on the other end by an MQA equipped DAC owned by the consumer. If you don't own such a DAC, then you end up playing the music without the complimentary decoding. Much as happened back in the day when Warner Music released a number of HDCD encoded albums and many buyers didn't own audio equipment with HDCD circuits.

 

As for the sound, it's going to be a case of auditioning the system with and without MQA and deciding if it merits buying a new DAC to get the MQA feature.

 

I heard MQA recently at the West Coast debut event on a $100,000 Meridian system and a headphone demo with MQA decoders in a separate room. I found MQA decoded music of several songs I am very familiar with to be inferior to the 24/96 and DSD versions of the same music on my much less expensive audio system.

 

Interestingly, most of the Meridian presentation centered on how processing music to CD Quality with MQA encoding was a boon to streaming services like Tidal. That may be where MQA makes the most sense.

 

If you are looking for the highest quality sound, MQA doesn't provide it to my ears.

[miguelito]

I'd describe MQA as an encode/decode system vs. a compression algorithm. The mastering engineer uses Meridian-licensed equipment to encode the music which includes "pre-distortion" and storing dynamic range information in the noise floor of the CD Quality spectrum. The original music is recovered on the other end by an MQA equipped DAC owned by the consumer. If you don't own such a DAC, then you end up playing the music without the complimentary decoding. Much as happened back in the day when Warner Music released a number of HDCD encoded albums and many buyers didn't own audio equipment with HDCD circuits.

I get this and it makes sense, but given the demo I heard of an MQA's 24/192 file vs that file was markedly different, I would argue there's more to it that HDCD-style information "folding" (a term used by Meridian in fact - which I like).

 

As for the sound, it's going to be a case of auditioning the system with and without MQA and deciding if it merits buying a new DAC to get the MQA feature.

There's really no reason why there wouldn't be a software decoder - as long as Meridian gets royalties. I wonder if TIDAL would add decoding to their player or would tell people that in order to hear the improved sound they will need Meridian hardware - a losing proposition in my opinion.

 

I heard MQA recently at the West Coast debut event on a $100,000 Meridian system and a headphone demo with MQA decoders in a separate room. I found MQA decoded music of several songs I am very familiar with to be inferior to the 24/96 and DSD versions of the same music on my much less expensive audio system.

What I heard was very good, but certainly their picks not mine.

 

Interestingly, most of the Meridian presentation centered on how processing music to CD Quality with MQA encoding was a boon to streaming services like Tidal. That may be where MQA makes the most sense.

If this means that only people with MQA hardware decoding equipment - ie Meridian equipment - will be able to play MQA-encoded streams, then this is clearly a complete bust - IMHO of course.

 

If you are looking for the highest quality sound, MQA doesn't provide it to my ears.

I wouldn't regard MQA to the"euphonic filter must-have". I would regard it as a good way to stream a higher than CD quality stream using CD quality bandwidth - which in and of itself is great. Why this is better than a new high-bit-rate MP3 encoding for high res files is not clear to me.

 

It certainly made the 24/192 file post-processed with MQA sound better - but then again I can think of many reasons why that would be since after all information was REMOVED.