Anyone heard about Meridian's new project called MQA

[Orb]

I forgot to also say.
And if the music industry are messing up hirez music at the moment (appreciate it is not all but seems it is a fair chunk), then what does this mean for the CD quality releases and lower (streaming type services) that are derived from the master files and would suffer the same problems but possibly exacerbated even more.....
Possibly part of the issue of why CD releases in general seem to be average in terms of sound quality even for labels known for good recording setup.
Appreciate such an argument is made more complicated due to mainstream pop genre release but quality trends go beyond that.

Cheers
Orb

 

[mep]

Sorry but where in the article does it mention it does not improve the process?
I get it cannot improve a 24-bit master file, but MQA is meant to be more than just a transmission solution.
Thanks
Orb

Orb is absolutely correct in saying that Meridian is more than just a transmission solution. Meridian is promising that their MQA versions of songs will sound better than any current hi-rez format. This was never solely about just being faster to stream.

 

[esldude]

Sorry but where in the article does it mention it does not improve the process?
I get it cannot improve a 24-bit master file, but MQA is meant to be more than just a transmission solution.
Thanks
Orb

One of the comments asked if MQA was supposed to improve sound quality or be a transparent conduit for it. The Admin (Mark Waldrep) replied he had asked that question directly of Stuart and been told it was a transparent conduit. I have already seen people give the impression the MQA process provides additional improvement upon files it encodes on various forums. The Robert Harley article and graphs sure give that impression. Stuart, who with others developed MQA say otherwise. So according to the fellow who developed the format it is not more than a transmission solution technically.

How it might be considered more is that you get 'approval' that the file being encoded to MQA is the quality you wish for. That way the end user seeing the MQA indicator lit knows he is getting what was intended. So then a question is Harley's graphs of temporal abilities of the various formats. That makes you think MQA does something extra. So did Harley invent those, did Meridian give those to him, is it a case of Meridian marketing saying one thing, and Stuart otherwise or what?

 

[amirm]

I spoke to Bob Stuart at the last conference I was at (CES?). He does indeed talk of two types of improvements:

1. Efficiency of transmitting high resolution audio. As we get above 20 Khz, the amount of information in the content drops like a rock. PCM treats the entire spectrum equally, assigning the same number of bits, e.g. 24, to the full spectrum. MQS segments the content and quantizes, i.e. assigns bits, dynamically to higher frequency content proportional to the level of information in them. This results is efficiency improvement and the higher the sampling, the more improvment.

2. Some kind of matching technology where they analyze things at encode time, and apply the correct counterpart at playback. I have to confess that I have not studied this, nor do I understand what it means in the context of a system that goes from digital to digital (i.e. not modifying the DAC). It is this feature that is said to improve fidelity.

I have not had any opportunity to listen and compare #2 with and without MQA. The demos I have heard have always had MQA on. There was no pre-treatment version of the song to compare. Is something here? I don't know .

 

[amirm]

One of the comments asked if MQA was supposed to improve sound quality or be a transparent conduit for it. The Admin (Mark Waldrep) replied he had asked that question directly of Stuart and been told it was a transparent conduit. I have already seen people give the impression the MQA process provides additional improvement upon files it encodes on various forums. The Robert Harley article and graphs sure give that impression. Stuart, who with others developed MQA say otherwise. So according to the fellow who developed the format it is not more than a transmission solution technically.

How it might be considered more is that you get 'approval' that the file being encoded to MQA is the quality you wish for. That way the end user seeing the MQA indicator lit knows he is getting what was intended. So then a question is Harley's graphs of temporal abilities of the various formats. That makes you think MQA does something extra. So did Harley invent those, did Meridian give those to him, is it a case of Meridian marketing saying one thing, and Stuart otherwise or what?

Our posts crossed . As I mentioned, I can tell you with 100% confidence that Meridian advertises two sets of features, not just one regarding transmission. Indeed that is how my conversations started with him at CES which him strongly emphasizing the improvement of the encoding/decoding, i.e. a matching chain.

 

[esldude]

Orb is absolutely correct in saying that Meridian is more than just a transmission solution. Meridian is promising that their MQA versions of songs will sound better than any current hi-rez format. This was never solely about just being faster to stream.

Apparently Robert Stuart disagrees with you. Which does beg the question of who provided Robert Harley with the opposite idea?

If they release the process it would be easy enough to test. If they continue to keep it very close to the vest, and keep themselves right in the middle of it while being vague about what it does I think they are effectively killing any chances it has.

 

[esldude]

I spoke to Bob Stuart at the last conference I was at (CES?). He does indeed talk of two types of improvements:

1. Efficiency of transmitting high resolution audio. As we get above 20 Khz, the amount of information in the content drops like a rock. PCM treats the entire spectrum equally, assigning the same number of bits, e.g. 24, to the full spectrum. MQS segments the content and quantizes, i.e. assigns bits, dynamically to higher frequency content proportional to the level of information in them. This results is efficiency improvement and the higher the sampling, the more improvment.

2. Some kind of matching technology where they analyze things at encode time, and apply the correct counterpart at playback. I have to confess that I have not studied this, nor do I understand what it means in the context of a system that goes from digital to digital (i.e. not modifying the DAC). It is this feature that is said to improve fidelity.

I have not had any opportunity to listen and compare #2 with and without MQA. The demos I have heard have always had MQA on. There was no pre-treatment version of the song to compare. Is something here? I don't know .

Looking at their patent applications, #2 would appear to be a version of subtractive dither. It may also use psychoacoustic info to let the dither do more than just what dither normally does. More sophisticated than just say noise shaped dither for instance.

It does make me wonder if subtractive dither is used, what effect might that have if I use playback software or hardware that does digital volume control. Will that be an issue, will volume need to be analog for it to really be MQA?

 

[Peter Breuninger]

We have a new video with some challenges and replies on it coming out soon. You can be the judge.

 

[esldude]

From the Dr. AIX blog comments section.

1st comment (which as full disclosure was my comment)

Having studied the patent some I was pretty sure MQA didn’t improve anything, but was only supposed to be an audibly transparent conveyance in a low bandwidth package.

The reply form Mark Waldrep:

I asked the inventor that very question and he agreed with my assessment. It doesn’t get any clearer than that. Although, there were several attendees that experienced the MQA demo in room 306 and told me that the presented did say that the technology improved the fidelity of the audio. I don’t think those presenters have a complete grasp on the issues at hand. And Robert Harley’s over the top assessment didn’t help things.

And now apparently Amirm has gotten a different answer in talking with Bob Stuart.

 

[still-one]

Apparently Robert Stuart disagrees with you. Which does beg the question of who provided Robert Harley with the opposite idea?

If they release the process it would be easy enough to test. If they continue to keep it very close to the vest, and keep themselves right in the middle of it while being vague about what it does I think they are effectively killing any chances it has.

During the MQA demos the Meridian Regional Sales Manager did say that MQA could improve older material. (I am not going to restate what he said because I will get it wrong) This has been stated pretty consistently across the board by Meridian. I think whoever is quoting Bob Stuart didn't hear or understand his response properly. Now whether or not we will ever hear this improvement will have to wait until we can do our own comparisons.

 

[mep]

Apparently Robert Stuart disagrees with you. Which does beg the question of who provided Robert Harley with the opposite idea?

If they release the process it would be easy enough to test. If they continue to keep it very close to the vest, and keep themselves right in the middle of it while being vague about what it does I think they are effectively killing any chances it has.

Look-Go read Meridian's info on MQA on their website before you tell me that they aren't advertising better sound than what we hear now through current digital. Here is one example from Meridian's website and not something that somebody claimed Meridian said:

MQA allows listeners to experience every intricate detail the microphone heard, offering music fans the purest ever sound. And it’s based firmly in science. For the first time in history, music fans will be able to hear at home what the artist created and approved in the recording studio, and MQA confirms its exact delivery.

 

[esldude]

During the MQA demos the Meridian Regional Sales Manager did say that MQA could improve older material. (I am not going to restate what he said because I will get it wrong) This has been stated pretty consistently across the board by Meridian. I think whoever is quoting Bob Stuart didn't hear or understand his response properly. Now whether or not we will ever hear this improvement will have to wait until we can do our own comparisons.

Well I also wonder how it could do that. Mark Waldrep has said Bob Stuart is going to MQA encode some of Waldrep's files so he can hear the result. How can they improve upon a file that was originally 96/24 or 192/24. I have seen it implied they will be able to tell what converter was used, and improve the result. I have elsewhere seen it implied they will be able to determine which type of filters were used in the AD stage and make improvements based upon that. I am no expert with DSP. Such would potentially be possible I suppose. The graphs in the Harley article imply they will be able to reduce "temporal blue" of digital filters. Is this only on the reconstruction end, is it on the complete encode/decode process? Is that even really much of a problem? How can they undo the blur in the original encoding ADC?

 

[esldude]

Look-Go read Meridian's info on MQA on their website before you tell me that they aren't advertising better sound than what we hear now through current digital. Here is one example from Meridian's website and not something that somebody claimed Meridian said:

MQA allows listeners to experience every intricate detail the microphone heard, offering music fans the purest ever sound. And it’s based firmly in science. For the first time in history, music fans will be able to hear at home what the artist created and approved in the recording studio, and MQA confirms its exact delivery.

Their statements are like the one you quote. That says almost as much by what is left out as by what is stated. For the first time you can be sure of hearing exactly what the artist created and approved with MQA confirming it. The confirming it alone part is new, and in many cases would be a step forward vs how HD Audio is peddled currently. That statement however actually doesn't say it improves upon what was recorded at the studio by encoding into MQA, only that you will hear it and know it was unadulterated between studio and you by anything in MQA.

Nevertheless, not a big sticking point to me. Until they let more of the cat out of the bag, make some music available or let some software do the decoding, we won't know. If it helps via better transparency to higher quality masters or whether it improves on those thru superior encoding would both be obvious enough if true. The rollout seems very slow so far.

 

[Fitzcaraldo215]

From the Dr. AIX blog comments section.

1st comment (which as full disclosure was my comment)

Having studied the patent some I was pretty sure MQA didn’t improve anything, but was only supposed to be an audibly transparent conveyance in a low bandwidth package.



The reply form Mark Waldrep:

I asked the inventor that very question and he agreed with my assessment. It doesn’t get any clearer than that. Although, there were several attendees that experienced the MQA demo in room 306 and told me that the presented did say that the technology improved the fidelity of the audio. I don’t think those presenters have a complete grasp on the issues at hand. And Robert Harley’s over the top assessment didn’t help things.

And now apparently Amirm has gotten a different answer in talking with Bob Stuart.

I think there may be unjustified controversy stirred up here. And, it is all still brand new. We are not seeing actual quoted replies to actual quoted questions of which there is some record. It is beginning to sound like a heresay "he said, she said" of impressions where one or another person has filled in the gaps to his or her liking.

Having read a fair bit about the technology myself in a number of places, including Harley's interview with Stuart of allegedly verbatim responses by him, my impression is that Amir has the more correct version of things from the horses mouth. Again, it is early in the game. But, I see nothing anywhere else, except for Waldrep's comments and other idle speculation, that this is merely a "lossy compression scheme". But, Waldrep himself is not all that high on the credibility charts, in my opinion.

The disappointment really seems to be that in the early demonstrations of MQA, the opportunity to present clear and convincing evidence of improvements beyond just data compression has been missed. That does not necessarily mean that MQA has no SQ improvements beyond compression. But, what does it mean? Perhaps there are more facts and more effective demos to come. I suggest that anyone who thinks he has this all figured out already is overreaching.

 

[esldude]

I think there may be unjustified controversy stirred up here. And, it is all still brand new. We are not seeing actual quoted replies to actual quoted questions of which there is some record. It is beginning to sound like a heresay "he said, she said" of impressions where one or another person has filled in the gaps to his or her liking.

Having read a fair bit about the technology myself in a number of places, including Harley's interview with Stuart of allegedly verbatim responses by him, my impression is that Amir has the more correct version of things from the horses mouth. Again, it is early in the game. But, I see nothing anywhere else, except for Waldrep's comments and other idle speculation, that this is merely a "lossy compression scheme". But, Waldrep himself is not all that high on the credibility charts, in my opinion.

The disappointment really seems to be that in the early demonstrations of MQA, the opportunity to present clear and convincing evidence of improvements beyond just data compression has been missed. That does not necessarily mean that MQA has no SQ improvements beyond compression. But, what does it mean? Perhaps there are more facts and more effective demos to come. I suggest that anyone who thinks he has this all figured out already is overreaching.

I definitely don't think I have it all figured out. The less than clear presentation of what MQA offers is the problem. A problem in the sense of not knowing what it does if you care like I or other audiophiles do. A bigger problem if you are rolling out a new format, and it confuses potential customers about what it accomplishes. I guess I am just being impatient. Rather than being announced and clearly touted what is toutable about it, this seems to be some measured rollout to build up interest, to create a buzz before releasing it. Which is okay, I just have a distaste for manufactured buzz.

 

[mep]

Their statements are like the one you quote. That says almost as much by what is left out as by what is stated. For the first time you can be sure of hearing exactly what the artist created and approved with MQA confirming it. The confirming it alone part is new, and in many cases would be a step forward vs how HD Audio is peddled currently. That statement however actually doesn't say it improves upon what was recorded at the studio by encoding into MQA, only that you will hear it and know it was unadulterated between studio and you by anything in MQA.

Nevertheless, not a big sticking point to me. Until they let more of the cat out of the bag, make some music available or let some software do the decoding, we won't know. If it helps via better transparency to higher quality masters or whether it improves on those thru superior encoding would both be obvious enough if true. The rollout seems very slow so far.

I don't think you are reading the same articles about MQA including the Meridian website that everyone who has been following MQA has read. Let there be no doubt that Meridian is claiming better sound than what we are now hearing with digital in addition to the smaller file size for faster streaming. MQA is not being touted as a one trick pony that only has smaller file size and the file has been "authenticated" by the original artist(s) assuming they are still alive.

 

[Orb]

When talking about the patent,
the focus is on WO2014108677) DIGITAL ENCAPSULATION OF AUDIO SIGNALS when talking about MQA as an end-to-end solution in the studio-to-listener?
https://patentscope.wipo.int/search...=&sortOption=&queryString=&tab=PCTDescription

The other patent is GB2503110 DOUBLY COMPATIBLE LOSSLESS AUDIO BANDWIDTH EXTENSION.

Unfortunately there are another 5+ applications associated with the technology to be published, so I can not get how anyone is making such conclusions at the moment with regards to its limitations or scope *shrug*.
Anyway I am one who does feel filter implementations can be critical (more so I appreciate relating to CD quality), and also how the studio copes with multiple releases from the master file and its relation to the native recording rate, and that distributed for consumers.
No idea where they are going with DSD recordings though and also their transcoding release for PCM hirez and CD quality , although that might be part of one of the applications to be published *shrug*

I do agree with what many say with regards that it is unlikely to gain a large footprint any time soon, and an unknown how well it is received and adopted longer term.

Cheers
Orb

 

[amirm]

Sometimes I hate you guys because I was in no mood to dig into DSP and psychoacoustics heavy paper by Stuart et. al. but you guys made me do it. I did not read it carefully still as there is a ton in there but dig more information than I had ingested before. Here is are the key parts in the paper:

A Hierarchical Approach to Archiving and Distribution
J. Robert Stuart , Peter G. Craven

And the conclusion:


4.5. Distribution System
Using the coding concepts described above, it is possible
to re-code a PCM signal so as to preserve both spectral
and temporal features of the content in a smaller coding
space. The encoding kernel should be chosen to best
match each song (track) and should be kept constant for
that segment, it may also take into account knowledge of
the A/D converter or prior processing. To maximize
potential sound quality and efficiency, both ends of the
chain must be involved.

The receiver (decoder) should implement an appropriate
up-sampling reconstruction, a flattening filter matching
the chosen encoding kernel, and a platform-specific D/A
manager. Ideally we should improve efficiency and
ensure consistency by using end-to-end subtractive
dither.

Conceptually, we are trying to connect the A/D and D/A
modulators together with a signal that encapsulates the
entire sound of the original but without artefacts that
imply lack of resolution, and to package it for efficient
distribution. The authors have used lossless buried-data
signaling within the channel to carry instructions,
metadata and authentication. We are illustrating a
distribution method which, since the encapsulated
version is monitored in the studio, is not only lossless in
delivery, but also more dependable than if arbitrary D/A
converters are used at playback.

This method is also efficient. For example, the Ravel
segment illustrated in Figure 10 can be encapsulated into
a distribution file containing all the relevant spectral and
temporal information of the 192-kHz 24-bit original (9.2
Mbps) using an average data of 922 kbps.

----

The core of the system is one where they create a new sampling rate for the stream that is designed to contain the actual amount of information that exists in the content. For example if there is nothing above 50 Khz, there is no sense in preserving that. In this manner, the content is resampled in a dynamic manner to match the specific track being encoded. This by itself is said to be lossless (although this is a perceptual statement, not mathematical like FLAC would be for example).

The improvement bit comes from tailoring the signal processing in this conversion to take into account the response of the typical signal processing blocks embedded in analog to digital converter and digital to analog converters as shown in the figure. So if I know the A/D converter used, I can model its time and frequency response and take that into account as I create a new sampling for it. Those adjustments therefore allow one to back out response variations that are introduced in the A/D for example. Similar capability then exists in playback time with regards to how one's DAC works, assuming an accurate model of that exists.

In other words, the argument is that PCM audio is a source and destination independent system. It digitizes samples to a specific value and frequency (e.g. 24-bit at 96 Khz). MQA also creates sampling but is able to modify its algorithms to reverse out what changes from ideal may have been introduced in the A/D and D/A conversion. An example of subtractive dither is given where the noise that was added in A/D conversion can be taken out at playback if we know the exact nature of it. Which we would in MQA with profiling of the encoder, and do not in the case of PCM today that is devoid of hardware characteristics.

Notice the endpoint "archive" in the top A/D converter. The goal is to recreate those bits, rather than what came out as PCM. Likewise, if the DAC is known, to back out any side-effects of its upsampling logic (most DACs resample audio to higher sampling rate but lower bit depth prior to conversion to analog).

So this explains why different people touching the elephant are expressing different views. This system does not improve on what digital audio is. So in that sense, it is a transmission system. It is just that it is able to back out of the effects of processing in A/D and D/A conversion while it is at it. This can be and is described as delivering a transparent channel. No enhancement is made to the sound itself. If by backing out the effects of processing blocks in A/D and D/A conversion can be shown to be audible, then we are getting fidelity improvements and getting closer to what got into the A/D converter at the start.

 

[amirm]

Adding on, the paper has no less than 92 references it draws upon! It can make for a year long research, digging into every reference to confirm what has been said, and true context.

 

[esldude]

Adding on, the paper has no less than 92 references it draws upon! It can make for a year long research, digging into every reference to confirm what has been said, and true context.

Thanks Amir.