New Meridian audio format creating quite a buzz

Jamie Lang above did a better job then I could, but yes, I agree with you on exact same sample rate, which tends to be 24/48-ish from what I've seen. Obviously there used to be a need to downsample to 44/16 to play in consumer's CD players, but no longer. We're getting to a point where non-audiophiles are going to be streaming everything and audiophiles have the right DACs, so there's nobody to lose out. The question might be harder to answer with analog tape but as Jamie says, that's basically a moot point now. Analog tape has presumably all been archived to 24/48+ at this point.

Arguments can persist on the extent to which people can hear differences between 44/16 and the masters, but they're basically irrelevant; there's no reason to do 44/16 anymore.

 

The only reasons I can think of for why things like this don't happen is perceived incompatibility issues among customers, which as I mentioned above is fading away as physical formats become out of vogue. Cynics will say the industry just doesn't care enough, but if nobody is actually recording in 44/16 and sales of CDs are declining, I have to believe one day they'll just skip the process of converting down to 44/16 before sending it off to the streaming companies.

 

I had been shrugging my shoulders about the early buzz over MQA. It did not seem all that interesting to me. Although I am hardly a Bob Harley fan, I thought his TAS article, coupled with the Bob Stuart interview, was exceptionally good in revealing many aspects of the technology and why it might bring about quite a revolution in audio as we know it. My non-technical, music first, audiophile friends unanimously agree with that assessment. All of us now see its likely impact well beyond just being a more compact, more efficient streaming and downloading format. Though, those are the areas of obvious initial application and appeal. The compatible, multi layered lossless compression scheme is really quite ingenious. It has superficial parallels in how it works to the old HDCD scheme, but goes way beyond that.

The real appeal might be greatly broadening interest in hi rez. And, the potential payoff is not just better, tighter lossless compression, but higher fidelity in the process. I believe if the claims are true, it will greatly hasten the demise of physical media in music distribution. I also see it as handwriting on the wall foreshadowing the death of DSD, though that format has a loyal small niche of diehard disciples. I have tons of SACDs myself, but I am not married to DSD if something just as good or better, if only in file size and convenience, comes along. And, interestingly, MQA is also seen as an archival format, much as DSD was over 15 years ago. Of course, as we know, Stuart and many other engineers, both on the equipment and recording sides, have hated DSD/SACD from the outset and never supported it.

I have never owned a Meridian product. However, I do believe the measurable time domain filter ringing issues are a blemish on the ultimate fidelity of PCM formats. I find Stuart's approach rather brilliant of saving in the noise of the media file itself the metadata about the original a-d conversion for later use in d-a, both now and in the future.

Kal Rubinson at Stereophile was not only impressed with the sound, he received assurance that Mch audio is very much part of the capability. It also has the potential to greatly speed up the acceptance of the latest Auro-3d or similar technology, if that can compatibly be included in the codec, which would seem doable.

As Stuart is keenly aware, there are major struggles ahead. The industry and audiophiles are very resistant to change, and cynicism abounds, sometimes with justification, sometimes not.

Agreed, the proof will be in the listening, so all bets are off until then. And, they can still screw up the potential it offers by, for example, being too greedy in licensing, among other things.

 

Yes, because all DACs produce the same analog output?

I don't have a view on MQA, but analog tapes are just fine as far as I know. I'd need Steve to break the news that most jazz/rock tapes from the 50s to 80s are ruined. Who's to say that new A/D technology can't come along and show us all how much there is to hear on the tapes?

As for consumer playback, well, the surface is really being scratched now by innovators at Lampizator and the like. None of us have heard how terrific today's DSD and redbook can sound at home aside from the few with many tens of thousands in digital front end.

 

Yes, in fact they do, within a relatively small margin. Relative to the speakers....relative to amps....relative to rooms.....absolutely--the DAC is the most consistently accurate part of that playback. If you ONLY count whatever level you think are "good" speakers, they vary SOOOO much more widely than the WORST to BEST DAC.

 

If you compare the output of a typical cellphone DAC to the Oppo 105 (stock) output and then all the way to EMM Labs, Berkeley Audio and the like, then the difference is as significant to an audiophile as would representative speakers or amplification. Any way you cut it, it is the difference between ho-hum and shock and awe. I'm not sure if my current speakers fall within the "good speaker" spectrum by your standard, but I can say without any hesitation that a DAC upgrade in my system can change the sound such that one may not even recognize that the same speakers are being used. That's a large margin if you ask me.

 

I could be wrong, but I'm not sure "accuracy" is necessarily the same thing as "sounds good-ness" so to speak. Being "error-free" isn't the same thing as having more expensive capacitors providing a more pleasing sound, etc etc.

 

How can a lossy MQA format (at least for anything above 24/48) be every bit as good as a high(er) res master?..

 

I'm not sure how many masters are above 24/48. Or to put it slightly different, I'm not sure how many masters have usable data above 24/48.

 

Answer: because MQA is not lossy in any audible sense. The noise in the last few LSB's of 24 bit hi rez audio is completely inaudible. So, if those bits are repurposed to carry higher sampling rate signal and metadata instead of inaudible random noise, no one will hear the difference. They are still well below the threshold of audibility by humans. But, the system can access those bits and use them for encoding higher sampling rates losslessly and to improve the fidelity of the d-a process by apodizing corrections. I think this explanation is quite clear and logical in the TAS article.

 

Some observations based on extensive reading and having attended a MQA demo last week.

MQA is NOT a 24-bit system. Their patent Doubly Compatible Lossless Audio Bandwidth Extension covers the spectral folding. Pay attention to figures 7A and 7B. The input to the bandsplitter is a noise-shaped 17-bit signal, and that is also the output of the bandjoiner in the playback device.

Meridian did NOT demo the 24-bit original versus the noise-shaped 17-bit version, so impossible to know what audible effect, if any, this step is having.

Meridian claim that the legacy 16/48 version available to people without MQA decoders will sound better, yet did NOT demo it versus a direct down-conversion of the 24-bit original.

Meridian make a big issue out of the audibility of pre-/post-ringing of anti-alias/reconstruction filters but did NOT demo playback with/without ringing to prove their point. In contrast, when Stereophile looked into the audibility of filter ringing, they found it extremely difficult to identify differences:

http://www.stereophile.com/reference/106ringing/index.html

In much of the discussion of MQA the focus seems to be on 24/192, but Meridian's own materials indicate that sampling at 384 and 768 kHz is even better, and 384 kHz files are part of their demo material. They also talk about inter-aural time differences and advances in neuroscience, etc. as the justification for the need of higher sampling rates. No indication is given of when this became an issue within Meridian. Bob Stuart wrote in the late 90s that properly noise-shaped 14-bit/58 kHz PCM (or conventional 20-bit/58 kHz PCM) attains audible transparency. That at a time when both higher PCM sampling rates were appearing and DSD was being introduced. Moreover, inter-aural time differences on the order of 10 microseconds were already well known, and were used as the motivation for DSD's 100 kHz bandwidth claim. Companies like dCS published papers at the time on listening tests of 48/96/192 PCM and DSD, and suggested even then that the perceived improvement at higher rates was due to better time-domain behavior (especially when slow roll-off filters were being used, which would not extend the frequency response).
Meridian remain very tight-lipped on what is involved in 'characterizing' a mastering engineer's digital chain, and they are making people sign NDAs before they do discuss it. It will not be open to third parties to do the end-to-end characterization and have the necessary data embedded in MQA's subcode. This must be done by Meridian. It is unclear what this means if you have a converter with numerous filtering options, or if you upgrade/change gear, etc.

In order to ensure a 'perfect' end-to-end digital chain, the DAC needs to be adaptive to the data contained in the subcode. While Meridian talk about software solutions, it is clear that they are not thinking in terms of a plug-in for something like Foobar/JRiver. They talk a lot about the mobile phone space, despite doing their demo at a high-end dealer on a $100+K system!! Since a software plug-in cannot know the capabilities of your pricey downstream DAC, it means the authentication cannot be completed without some change in the owner's equipment!

They claim to have 72 companies on board at this time, but will not say who most of them are. Seventy-two Sonys, Matsushitas, Samsungs, etc., like in Blu-Ray, and the thing will have traction. However, the big boys are notorious for 'not invented here' syndrome, and did not adopt HDCD, for example (neither did Meridian, BTW!!). The big boys are also wary of creating another Dolby situation, which is what Meridian seems to want to do here (encoders in all studios, decoders in all consumer products).

With the widespread use of 24/96, 24/192, and especially DSD, the time smear issue is not as severe as it was, especially when minimum phase filters are used. Sigma-delta schemes like DSD have the narrow impulse response and lack of ringing by default, so it would be interesting to compare MQA to DSD at different rates.

This may seem like I am down on Meridian and MQA but that is not the case. There are some clever bits (especially the spectrum folding), but they did not demonstrate any of the benefits (there were no audio comparisons at all), and I remain skeptical that this will gain widespread acceptance by the major players (even if it can be shown to improve audio performance).

As I understand it, the intention is not to return to the analog tapes, but to recode the digital masters using the appropriate characterization data for the digital chain used to create the digital file. That assumes that that information is cataloged with the digital master file, and I highly doubt that that is the case!

Meridian's tests of 1,000 music samples showed that 2.4% would NOT fold due to the spectrum shape. They will not say (outside of a NDA) what can be done in those cases. They also use a string quartet in their Shannon diagram rather than a more extreme spectral example!

 

@Black Elk, could you give your impressions of having heard MQA in a way that doesn't read like a white paper? I tried reading all that but gave up.

 

I couldn't tell you anything about how it sounded. They had a Sooloos feeding some $15k Meridian pre-amp. feeding $80k 8000SE digital active speakers in a large-ish demo room furnished more like a home than a studio. They only set up three low listening chairs, and I never got a chance to sit in one, but did kneel down behind the middle chair. There were loads of people standing behind the three seats. They just played various pieces that I presume they thought would impress, but without knowing the system/room, it was impossible to draw any conclusions about MQA since there were no comparisons to anything.

I attended with another Forum member, who also has a lot of experience with Hi-Res, and he too was very underwhelmed, both by the overall quality and the lack of any comparisons.

 

Cool, thanks for that. So they didn't bother with before and after, this is the original and this is the MQA? They just played great sounding tracks on a six figure stereo that would blow people away regardless? Hmmm.

Oh lawd. If that's the case, this is beyond worthless.

 

Time will tell.

 

Correct.

I wish I could say that was the case, but I didn't hear anything that made me go, "Wow!" This is a report from the same store in which the MQA demo was held:

http://www.stereophile.com/content/meridians-spectacular-speaker-unveiling

The system was set up as shown in the photos, except all the large KEF speakers were arranged down the left side of the room. As I wrote earlier, I did not get a chance to sit in one of the three low seats they had set up. The sound did improve when I was on my knees, but I did not hear anything like what is described in the Stereophile piece (and in similar MQA reports from Stereophile, TAS, etc.). I cannot find measurements on the 8000SE speakers, and it could be that you need to be at or slightly under tweeter height for everything to snap into focus. In that case they would have been better off raising the speakers up, since the vast majority of people attending had to stand to listen.

They also had a headphone system set up comprising a Macbook feeding a $2k Meridian Prime DAC/headphone amp. and a variety of headphones. I started with the Audeze LCD-X, but no matter how much I played with the headband settings I could not get a comfortable fit. They just seemed like a ton weight on my head. The Audeze EL-8s were more comfortable, but we managed to find a pair of Sennhesier HD800s, which was much more to my liking. I was familiar with a lot of the music they had on the Macbook, and have even heard the master tapes with the original recording engineer for a few of the tunes. Again, the sound did not impress. I sensed a hard edginess to the sound. I would not put this down to the HD800s. It would be interesting to know from some of the big headphone users here whether this is a feature of the sound of the Prime. I use a tube amp on my HD600s at home.

Apart from a live Beethoven piano sonata recording, all the demo pieces on the big system were either jazz or pop. Pieces I can recall were:

Dave Brubeck - Take 5
Elton John - Goodbye Yellow Brick Road (?)
Dire Straits - Money For Nothing
will.i.am/Justin Bieber - That Power

For the will.i.am/Bieber they said that it needed to be played loud. While I have certainly heard louder in studios, I felt the system ran out of gas when things really got going.

On the headphone system there was music from The Doors, Steely Dan, Wilco, Van Morrison, etc.

I can tell you that I have been more impressed by demos of unknown gear at Hi-Fi shows, especially allowing for the crappy hotel rooms and limited set up time.

RE: remastering/recoding

This is still a confusing part of the story because they are keeping so tight-lipped. They can, and will, re-master from the analog tapes, but they realize that that is a slow and expensive process if you want to get a ton of material ready for launch. So, knowing the characteristics of the digital system that created a digital master file, they can recode it with the apodizing/correcting filter response(s) and create a larger library. That's the way I understand it at this time.

 

Whether or not there is merit to MQA, they better hire the top mastering engineers if a bunch of original source tapes are going to be used. The last thing we need is an archiving effort mastered for loudness or goofy eq, etc.

 

This sounds like PONO?

Is Meridian seriously going to be doing their own remastering of old analog recordings?

The market really does not need further fragmentation and perpetual revisiting of old war horses.

 

No, that's not my understanding. I get the impression that they want to get 'encoders' into the mastering studios for projects going forward, but also want to re-encode existing digital masters so that they have some 'apodizing goodness'.

 

I'm positive if they use enough digital processing, they can eliminate the buzz.

 

I'm still on the fence about MQA remastering but having heard the DSP8000's many times I can say confidently that they add an uncanny sense of clarity and realism to the performance.
To bad at $80,000 ( $108,000 Canadian ) there out of reach for most music lovers.

 

Thanks for all the contributions. I finally got around to reading the MQA article in The Absolute Sound (on the plane from San Jose to Houston) and it was such an impressive article, it had me totally wishing this was all true. It seems from what Black Elk heard, that is not the case. I hope this does go forward a bit further though - the technical solution is quite fascinating.

 

A question to the thread:

Does anyone believe that a lack of DRM is what's holding back the major record companies from releasing catalog titles in resolutions > Redbook?

To me, the elephant in the room regarding MQA is the condition of catalog analog master tapes and the willingness of record companies (or others who own the tapes) to let them be transcribed yet again.

 

I have no clue about anything related to DRM.

Original analog master tapes are currently used by Mofi, Audio Fidelity, Analogue Productions and Universal Japan (SHM series)- to name a few. Original tapes, for the most part as far as I know, are not shipped long distances. That's why Analogue Productions, for example, uses Kevin Gray's studio in the west for some titles and another studio in the east for others. It's also why the Japan SHM series titles are sometimes mastered in the US and other times in the UK. Point being that numerous studios will be used if a major archival project is undertaken which would then presumably retire the tapes. Should anything like this transpire, then the new digital source better be 99.99% as good as the tapes. I do assume that someday the back catalog of the most important recordings of all-time will be archived. It would nice for Steve to chime in though.

 

I am way out over my skis on all of this stuff, but I had a question if they even needed to go back to the original tape assuming a hi-res transfer of the tape had been made for some previous mastering project.

In the comments of Atkinson’s article here the question comes up “How does one compensate for the temporal effect of the A/D converter? and Atkinson answers “I understand that mastering with MQA uses Meridian's "apodizing filter" but fine-tuned to the actual A/D converter originally used. According to Bob Stuart, this is possible because 1) there is only a small population of professional A/D converters and 2) record companies actually keep good records on what converter was used for the original sessions and/or mastering. Almost all CDs from the early 1980s, for example, were mastered with one of the Sony PCM1600 family.”

This article here there is: “In the meantime, Robert walked me through the thinking that went into the MQA development. He laid out the production path from recording or mastering studio to home listening as an “analog to analog” signal path with a challenging digital chunk in between. Their goal is to preserve the exact sound of the that final analog experience (even if it is being sourced from a digital audio workstation like Pro Tools) including the individual characteristics of the Analog to Digital and Digital to Analog converters. If the engineer, producer, and artist are listening in a studio equipped with dCS or Apogee converters, then the metadata about those devices will be “encapsulated” into the MQA sound file so that the playback mirrors exactly the experience in the studio…and at approximately one tenth the bandwidth.”

So as the MQA conversion process relates to the original tape, do they even need to use it? Can they make these filtering choices, along with the other codec advantages that are over my head to cite, with any hi-res flat transfer already laying around and that presumably a record company would be more willing to release for this process? If I have not asked an extremely dumb question already, would you then try to match what was done for some a/d transfer in the ‘80s or the more recent hi-res flat transfer. I guess in the end the client will choose. I assume you try a couple of filter combinations out based on what was done for what you want to replicate at the mic as was said, the client or the label probably still says yea or nay.