The MQA comprises of three important elements.

1) MQA files incorporate a digital ‘signature’ buried within the audio data which is examined by the replay system’s MQA-compatible D-A converter. This provides authentication that the reconstructed audio file delivered is exactly the same as that encoded by the file’s author, i.e. to file being recognised as a genuine MQA file.

2) MQA is principally a method of digitally capturing and storing original master recordings in by compressing them in a new way, with greater compression ratio than FLAC of ALAC formats. The MQA comes typically at half their size (but in some hi-res cases it might be 2/3).

MQA encoding is formally lossy, but after the decoding by using proprietary dithering techniques the result would be a lossless file.

Using a form of file fingerprinting, when paired with an MQA decoder, the MQA file truly reveals the original master recording (in 17-bit), but it can play back on any device to deliver higher than CD-quality (playback limited on non-MQA devices effectively to 13-bit).

It works on the principle that limited amount of energy in higher frequency bands (above 22.05 or 24kHz) is “sent” into the lowest bits of a 24-bit/44.1kHz or 24/48 file. It uses time-domain ADPCM and bit rate reduction instead of perceptual encoding based on psychoacoustic models (mp3, wma).

Its bit rate is about 1.5Mbps, when coupled with additional FLAC compression.

The MQA quality is higher than "normal" 48/16, partly because of the novel sampling and convolution processes (using a triangle function, too technical to be explained here).

One of its aims is to pack better-than-CD-quality audio files down to a size suitable for Internet streaming.

3) The third advantage of MQA is its effort to avoid standard PCM audio ‘smears’ of timing information and render a very precise time-domain performance. (There are opinions that PCM is inferior to analogue mostly due to quantization /solved mostly by very high sample rates, but not quite, and at the expense of huge memory consumption/ and bad temporal precision).

The wide-spread use of linear-phase ‘brick-wall’ filters in digital equipment, gives a near-perfect frequency-domain characteristic. But the filters’ time-domain behaviour is very bad, with significant pre- and post-ringing. They typically have ringing tails extending for several hundred microseconds before and after the main impulse. Especially the pre-ringing, where sound energy builds up in advance of the sound’s actual starting transient, and that's not what happens in nature. This is traditionally being partly solved by using a ‘minimum-phase’ filtering - but on its bad side it creates louder and longer post-ring tails - or by using higher sample rates (where the audio data rate grows enormously with only moderate reductions in the ringing duration).

So the apodizing filter was designed to render a relatively gentle cut-off slope bove the pass-band, but still achieve desired attenuation well before the Nyquist limit. This is only practicable, though, where the sample rate is significantly higher than the required audio bandwidth (i.e. for 96kHz sample rates and above).

Since the overall impulse response of the chain is the convolution of their individual responses, the convolution process allows to a cascade series of filters to cause a shorter impulse response than that of any individual filter in the chain. So, an apodizing filter can almost completely remove the pre- and post-ringing tails associated with a chain of conventional linear-phase filters.

But the best thing is if the system’s native impulse response was already ‘perfect’. MQA target is delivering a perceptual time smear of about 10µs for existing digital recordings (it’s around 100µs on a conventional 24/192 system), and 3µs for newly recorded material. (Recent international research found out that the human auditory system can resolve transient timing information down to at least 8µs, (and possibly even 4µs.)

This level of time-domain performance is only possible if MQA is used as a complete end-to-end system, encompassing both the original A-D sampling and encoding within the mastering process, as well as decoding stages. So, with implicit and precise knowledge of all digital filtering processes involved in the chain, the system’s overall impulse response can be made almost perfect. In theory, it could even correct for the time-domain behaviour of studio mics and mixing consoles by keeping track of their impulse response characteristics and counter acting them by adequate filtering.

For the hi-res reproduction, with MQA-compatible converter, the original (so called 'origami') process is completely reversed to fully reconstruct the 24/192 output signal.

There is also a very long technical story behind all this.


MQA music files are available to download from Highresaudio, Onkyo Music (powered by 7digital), and 2L, as well as e-onkyo music, Kripton HQM's store and groovers (Japan).

Hardware decoders are manufactured by Pioneer, iFi Audio, Onkyo, Mytek, Meridian, Cocktailaudio and Bluesound. Selected Meridian products will support MQA thanks to a firmware update. On software side, the desktop application by Tidal supports MQA.

 

I posted this on another thread, then saw your's here. So i'm placing an edited reply here too...

As @pratyahara correctly identifies, the nub of the problem is that MQA would need to be employed across the entire recording-production-mastering-distribution-reception chain to show any substantial benefit whatsoever - and even then it would be noticeable on top-tier DAC kit only, whereas the average modern phone/car/laptop user wouldn't notice any significant improvement whatsoever. So, that's four commercial licences that no other commercial interests (other than the owners of MQA) feel any particular inclination to pay for - and all to satisfy the obsessive 'needs' of a fractional percentage of hardcore nerd audiophiles who remotely give a shit anyway about this level of 'perfection' anyway...

IMHO this technology will remain substantially stillborn until/unless it becomes open source, which frankly isn't likely to happen. And its premise is essentially irrelevent for the average music consumer anyway, 'cos whatever benefits MQA can supply, it certainly can't transform crappy content into incredibly profound masterpieces... which indeed would be software tech worth paying for.

 

Well yes, those theoretical informations are known, some questions that come to my mind on a practical level are:

• What will be the standard delivery format in the future? DSD, DSF, MQA? For now it's all very confusing
• How do I play them in the car? These are audio files with very high bitrate and I don't think it is possible to use them from smartphones: do I need a device with an analog out?
• If this encoding is so efficient, should we expect a daw with native MQA-recording inside?

Then, there is also the problem of the proprietary format: experience has shown that this is very limiting for its diffusion

 

Last edited by a moderator: Oct 14, 2020

I agree, it will never become open source simply because it's a useless software to hardware related business created by Meridian, and we don't need another lossless format rather than flac, even for streaming, i'm glad that Qobuz didn't choose that and stick with flac for lossless streaming instead of Tidal with his "masters" format, which is mqa, i don't like it at all to be honest, i even have a bat file created to scan your audio files and check if 48/24 ones are "genuine" flac or mqa just to avoid that, if someone is interested just ask

Some articles:

https://www.linn.co.uk/blog/mqa-is-bad-for-music
https://www.realhd-audio.com/?p=5948

Only for nerds:

https://archimago.blogspot.com/2016/10/musings-keeping-it-simple-mqa-is-codec.html
http://archimago.blogspot.com/2017/01/comparison-tidal-mqa-music-high.html
http://archimago.blogspot.com/2017/02/musings-discussion-on-mqa-filter-and.html
http://archimago.blogspot.com/2017/04/comparison-mqa-authentication-sound.html

 

Last edited: Oct 15, 2020

We all deserve to listen to the music with the best format possible, listening to mp3 in 2020 is a crime against humanity.
Said that, a MQA can’t go any further of a 96k24bit flac recorded from analog master (see comparative test in the @Epidemico links for nerds)
So we could say "Flac is already the best format available, let's use that". The topic is less simple than it seems: often the right idea is not the one that is adopted in reality, there are strong commercial interests involved, in short: MQA is more profitable for business.

We will probably have to accept this new delivery format without arguing

 

It is a real breakthrough in sense of advancement of audio processing. Time-domain problems are practically solved, as well as irregularities and disbalance in processing chains - and that was the biggest unsolved drawback of digital audio compared to analogue.
Its sounding benefits can be heard to the most on very high quality Hi-Fi equipment it's true, and convincingly more natural sound is apparent even on mid-priced Hi-Fi. Hi-Fi impact and share on the market is really minute, so MQA can not make any commercial success in spite of its essential improvements in quality. But its streaming advantages and possible spread among streaming services seem to offer a much greater potential for it to survive as a standard. That is questionable though because of greater expenses and the fact that most such services rely heavily on non-authentic, re-encoded stuff.
Anyway, for a real music lover, it would be a shame not to have a chance to have a taste of it, even in some demonstration before it eventually goes dead. And a theorist might get satisfaction of exploring that great idea and studying its nuances.
Personally, I do not expect I'll have a chance to use it as a professional, and I don't have a Hi-Fi system, but I have learned some lessons I'll try to implement in my future mastering work. Especially in using apodizing techniques, or some plugin arrangements that can simulate them. I'm on my way already.

 

A very rich topic.
I had never even heard of this before a few days ago when @pratyahara mentioned it in another thread.
I then found and mentioned this article in that thread and found it to be very informative.
https://www.soundonsound.com/techniques/mqa-time-domain-accuracy-digital-audio-quality

I can see how lots of interesting discussions could (might) emerge about
- very complex technical details (lots of which are well over my head - but some basics still very interesting)
- discussions about commercial viability, etc (also kind of interesting)

But the bit in the SOS article that got my attention the most was this quote from the SOS reviewer...
However, what I can say with confidence is that my initial impression of what I heard was of a generally more believable and realistic sound stage, with instruments and ensembles appearing more vibrant and genuinely three-dimensional in a way which exceeded anything I’ve heard with conventional hi-res audio on high-quality monitoring systems. Everything was located within very natural-sounding and acoustically defined spaces, and transient-rich instruments — cymbals and drums in particular — became noticeably more tangible.

I note that our audio-engineering culture does have a way of chasing after any and all tricks that can improve audio quality,
and does seem to put a huge shambolic collective effort into achieving even just small incremental advances,
and we all just want those advances - even if they can be ridiculed by cost effectiveness arguments.

I thought the stuff in that quote above about achieving more precise 3-D placement of our sounds is a far more interesting incremental advance than whether the technology achieves more efficient use of bandwidth whilst retaining high quality audio.
But that latter less interesting (to me) topic - (caricatured as) clever efficient semi-lossless format, etc.
seems to be where most of the arguments are focussed in other articles that I found (and got bored with)

 

MQA is clever, no doubt about it. The problem (as i think i understand it) is that it is the software mechanism itself that is subject to Meridian's patent. Accordingly, should anyone else attempt anything remotely similar they will immediately fall foul of Meridian's lawyers.

Open formats are the way forward. If the MQA format fails, Meridian will have nobody to blame but themselves.

Way back in the day i used to be keen on hi-fi components. I bought the magazines, then the kit. After a while i became aware what a perpetual BS merry-go-round circle-jerk consumer journalism was - it's entirely the same today - staff writers will always tell you that this new bit of kit will make the old kit obsolete. Indeed, it was the quote above that triggered my suspicion that the entire SOS article was sponsored by Meridian to plug its nascent format to raise profile. I no more believe the objective accuracy of the above SOS quote than i will when i see the next VST plugin video telling me how this new compression plugin is the most faithful emulation of a xxxx etc...

 

Lol, absolutely not. You can't even differentiate between 384/64 file and a 320 kbps mp3 so I would suggest a check on your ambitions

 

it's late, go to bed: you have school tomorrow

 

Question: most of these articles are 3 to 4 years old. Why does the topic pop up on the radar now? Did a big player incorporate it yet so that it becomes a "game changer" or what is the cause for it bubbling again?

 

A very healthy skepticism. I didn't actually say I believed it either, but it was the only part of the article that really had the potential to get me saying "I want it now please". I won't go back editing my post with "if it works"
How about a fun prediction of a new thread that says DAWs with MQA sound better than DAWs without.
We could all argue about it now for 25 pages before we've even got one

 

The graphs of digital ringing look exactly the same as diffraction rings in optics. In astronomy, and photography in general, apodizing lenses are widely used (although expensive), and nobody denies that they contribute greatly to the image sharpness and detail.
So the graphs shown in the article mentioned above are fully credible. As seen in the graphs, MQA improves the physical characteristics of audio waves (removing the majority of distortions in time domain), so it must be heard also, because the differences are great.
It is corroborated by the fact that there is ample room for improvement in existing audio standards - simply by inferring from the fact that no combination of existing mainstream software/hardware can make you not to discern the difference between a recorded sound and the natural one, even for a moment.
I believe MQA partially fills this gap, and it's substantial for human achievement in recreating natural sounds.

 

It's been forever since I last used ADPCM. Stop right there @pratyahara , I know it's only a small part of the process .
As a techie it's interesting, but these patents and the need of involving all the chain for me it's a deal-breaker.

 

Read all the FabienTDR's (Tokyo Dawn Labs) posts about MQA stuff and forget this sh"t. (Do not remember, if it was a Mastering section, maybe yes, because I only read NewProductAlerts and Mastering sections). This was a popular thread year or 2-3 years ago. Deep analysis, documentation analysis, listening experience, and more. As I remember this was some kind of high-end ultraquality lossy-codec with some psychoacoustic enhancers, ditherings or so, for streaming mostly etc. I don't think you need this. Not a go-to must-have. Make good music, use proper SRC, dithering, flac/wav, use non-lossy HQ samples, sample-rates, use good gear and plugins, monitors, room acoustics, expand skill/knowledge/experience (in the field of audio, DSP, production, sound-design, mastering, mixing, recording, restoration editing, other processes, be better at production and performance techniques). MQA and other zillions of high-end codecs will not make a big hype in audio. Not now.

 

I do like this discussion.

In the blue corner:
@pratyahara is providing very compelling arguments why the MQA claims can at least be supported by some very credible underlying theory. Certainly not easy to write it off as snake oil.

and in the red corner: we have @Paul Pi providing a very wise I'll believe it only if and when I hear it brand of skepticism.

and I'll paint @Xupito as the bored referee saying
"who cares? the commerce and technical upheaval would screw it up anyway"

So, a good time to sit on the fence. I think I'll snooze until someone says "it's available for your DAW - try it"

But this quote from @pratyahara is the holy grail that keeps me interested.

So true, I want to stand blindfolded 20 feet in front of a small orchestra
and the worlds best speakers playing a recording of that orchestra.
I want to switch back and forth between the two and not notice the difference.
Allegedly possible on board the Star Ship Enterprise.

 

Last edited: Oct 15, 2020

I don't believe this topic is out of date at all.
Locally, most of the people here (including rather experienced) first heard about it after this thread was started. They are interested, and they have the right to be. What does it matter if some other people in some other times, on some other sites discussed it?
After all (4 years time is almost no time at all in science) we are now in a better position to discuss it because of accumulated experience, knowledge and analysis made in the meantime (including the affined field of optics).
Globally, it is an approach that is really novel, constructive, contributing and opens new horizons in audio processing and thinking.

 

Last edited by a moderator: Oct 15, 2020

1. it is only formally (transitory) a lossy codec, but all losses get fully recovered after applying a specific kind of dithering - so the end product is lossless,
2. Here are no 'psychoacoustic enhancers' involved,
3. It is not designed for streaming only, but also for perfect recording and reproduction concerning time-domain distortions,
4. 'Make good music', be a clever guy and do not waste your time, stick to what you already have, and so on, is off topic.
And how clever must one be to understand the graph attached...

 

Last edited by a moderator: Oct 15, 2020

Musicians, composers, producers etc can only test this technology once it's been adopted by DAW/plugin/hardware manufacturers/streaming services etc.

Because it processes digital audio streams differently, MQA would compel DAW software developers to NOT ONLY re-write their entire multitrack audio-read/write and export mechanisms to integrate MQA, they'll also have to enter into an prohibitive up-front licencing/financial obligation with an avericious UK hi-fi manufacturer for a theoretically better-sounding proprietry technology that nobody actually needs or many have even heard about... and if i'm not mistaken, every USB/thunderbolt audio interface in the world would also have to be replaced with models that integrated the proprietry MQA tech too...

The hi-def audio streaming advantage of MQA would have perhaps compelled adoption 15-20 years ago. Given universal bandwidth standards nowadays (along with the extremely bleak economic realities for industry, creatives & consumers in 2020) i just don't see MQA taking hold, principally 'cos the marginal sonic improvements on offer are all together too costly to implement across the board.

@pratyahara - are you Peter Craven's mum?!?

 

As a matter of fact @Olymoon asked me to start this thread. I don't know if he has any connection with Meridian, but I doubt it.
People might be interested in technology breakthroughs in order to educate themselves, and not stay ignorant. They also might get some new insights and ideas based on that knowledge.
Theoretically better-sounding is better sounding (or math does not apply to reality). Since this invention is aimed at the most authentic sound recording/reproduction quality known to this day, of course that it requires for some other improvements in different fields.
Maybe it will become available in specialised labs only, or in museums, but it is the best recording/reproduction system humanity has achieved so far.
Talking about replacing 'everything'... just think about the future of quantum computers. We know they are better, because math says so. Should we stop the research or debate about them because we 'can only test this technology once it's been adopted by xyz'?
The world wasn't made in one day.