MQA: A Review of controversies, concerns, and cautions

[Soniclife]

MQA is a lossy codec

It's a mix of lossless and lossy. The file starts as a lossless file, but some of its bit depth is sacrificed to hide the lossy data for the first unfold. After the first unfold the data below 44.1 is lossless, and it's lossy above. The bit depth improves with unfolding, but is still way below what the file advertises itself as.

 

[amirm]

This is quite a prediction! You really think that MQA is completely dependent on Tidal?

It is the only content source. Why would it not be the killing blow if Tida goes away? Who is buying DVD-A discs players today?

 

[amirm]

Bob, Bob, Bob......so what. His commercial enterprises and format attempts so far are all failures commercially.

Nonsense. Dolby TrueHD is from him (MLP) and has shipped in millions of Blu-ray players and used by major studios to release content. It is the basis of Dolby Atmos format as well. Do some research before making statements like this.

You know how hard it is to get a format adopted by Blu-ray?

Last I checked, Meridian is still in business, decades after it started unlike many high-end audio companies that have come and gone. My car infotainment system has Meridian logo on it.

Fastest way to lose credibility in your argument is to manufacture dirt like this.

 

[scott wurcer]

After the first unfold the data below 44.1 is lossless, and it's lossy above.

MQA is lossy, period. The original presentations where folks came away thinking otherwise were spin and IMO intellectually dishonest.

 

[amirm]

It’s astounding that on a site that claims to be about the “science” of audio, the MQA snake oil is accepted and promoted.

This is not a police state forum like the ones you may be used to. People can have different points of view. Mine is from having been in the industry, read and learned from Bob Stuarts research, and am a user of MQA. You are none of this. Nor is the owner of the forum you hang out in normally.

MQA is not promoted on this site. There are no ads for it, no sponsorship, and no articles from me on it. My personal opinion is just that and it is not up to you to force me to change for the sake of it.

Now, if one wants to know about real snake oil, all you have to do is read your signature on CA Forum:

So please don't lecture me on MQA and audio science. You don't have use for audio science.

 

[Soniclife]

MQA is lossy, period. The original presentations where folks came away thinking otherwise were spin and IMO intellectually dishonest.

Do you have a link that details this?

 

[amirm]

MQA is a lossy codec, more similar to the old ADPCM (adaptive bit depth reduction) than to the "new" perceptual encoders (MP3, AAC, etc...).

There is no reason to dedicate 24 bits of dynamic range to ultrasonic content in music. Only a few bits will suffice. This is how MQA is able to encode high-bandwidth audio in a backwards compatible way with just a few bits. The technique is very close to how HDCD worked in encoding 20 bit content in CD's 16 bit envelop. It need not work like perceptual codecs although the encoder likely has some logic as such.

Here is a random example file in my library encoded at 88.1 kHz:

You see the ultrasonic range? The "highest" level is at -100 dB. Assuming we need to encode down to -144 dB, that is only 44 dB of dynamic range requiring about 7 bits, not 24 bits.

We could also perform spectral/statistical analysis and realize what part of that is just noise and just discard it instead of attempting to encode it.

Such logic doesn't exist in perceptual codecs as they primarily deal with the audible band.

 

[GrimSurfer]

MQA is not promoted on this site. There are no ads for it, no sponsorship, and no articles from me on it. My personal opinion is just that and it is not up to you to force me to change for the sake of it.

FWIW to anyone tuning in, these are true statements. (Coming from a very cynical and sometimes sarcastic guy who enjoys calling out internet sock puppets at any time.)

 

[AudioSceptic]

It is the only content source. Why would it not be the killing blow if Tida goes away? Who is buying DVD-A discs players today?

So no one else will pick it up? Seems really odd that there's just one "provider", plus MQA CD (which I don't understand at all).

 

[scott wurcer]

Do you have a link that details this?

It was years ago, the point being that it was not presented in an open fashion from the start. As amrim says a 24/96 file contains lots of useless information, this is true, the claim that their algorithm correctly identifies exactly what is useless was not substantiated. With respect to 44.1/16 there is no more information so I don't see the point, claiming an improvement would violate Shannon. With respect to 96k/24 it does not return the same exact bits so it's lossy.

A long time ago my management invested in a small start up whose business plan hinged on the computer industry sticking with 640/480 CRT monitors and developed a small DSP to smooth the display artifacts. This was clearly stupid since 100's of millions was already being poured into flat panel hi-rez monitors. The point being any argument for MQA that has anything to do with file size, download speed. etc. is specious, that aspect of the industry will simply march on.

 

[AudioSceptic]

There is no reason to dedicate 24 bits of dynamic range to ultrasonic content in music. Only a few bits will suffice. This is how MQA is able to encode high-bandwidth audio in a backwards compatible way with just a few bits. The technique is very close to how HDCD worked in encoding 20 bit content in CD's 16 bit envelop. It need not work like perceptual codecs although the encoder likely has some logic as such.

Here is a random example file in my library encoded at 88.1 kHz:

View attachment 34923

You see the ultrasonic range? The "highest" level is at -100 dB. Assuming we need to encode down to -144 dB, that is only 44 dB of dynamic range requiring about 5 bits, not 24 bits.

We could also perform spectral/statistical analysis and realize what part of that is just noise and just discard it instead of attempting to encode it.

Such logic doesn't exist in perceptual codecs as they primarily deal with the audible band.

Shouldn't that be about 7 bits?

How many recordings have genuine musical content above 22 kHz?

 

[GrimSurfer]

About as many as my old ears can hear.

 

[amirm]

Shouldn't that be about 7 bits?

Yes, corrected.

How many recordings have genuine musical content above 22 kHz?

Fair number of them, especially those from vertically integrated labels that record and distribute themselves.

I did some analysis a while back and published them in youtube (watch full screen to see all the detail):

I examine multiple tracks here including one that is MQA encoded (around 10:00 minute):

 

[beefkabob]

Oh! I have a Meridian car stereo too.

 

[Wombat]

Have you forgotten that it made a big difference with studio tape recording before that, and continued to do so until the advent of digital?

Not at all. Just commenting re consumer use.

 

[Steve H]

Nonsense. Dolby TrueHD is from him (MLP) and has shipped in millions of Blu-ray players and used by major studios to release content. It is the basis of Dolby Atmos format as well. Do some research before making statements like this.

You know how hard it is to get a format adopted by Blu-ray?

Last I checked, Meridian is still in business, decades after it started unlike many high-end audio companies that have come and gone. My car infotainment system has Meridian logo on it.

Fastest way to lose credibility in your argument is to manufacture dirt like this.

Time to add some facts. Bob Stuart and Meridian had a pretty good run from 1997 to 2002. From then until he left the board Meridians operating results have been poor enough that I wonder why people listen to him. How big were the operating losses? They averaged over 2 million Pounds per year.

 

[Blumlein 88]

This is not a police state forum like the ones you may be used to. People can have different points of view. Mine is from having been in the industry, read and learned from Bob Stuarts research, and am a user of MQA. You are none of this. Nor is the owner of the forum you hang out in normally.

MQA is not promoted on this site. There are no ads for it, no sponsorship, and no articles from me on it. My personal opinion is just that and it is not up to you to force me to change for the sake of it.

Now, if one wants to know about real snake oil, all you have to do is read your signature on CA Forum:

View attachment 34921

So please don't lecture me on MQA and audio science. You don't have use for audio science.

This seems like a low blow to me. Unnecessary to any argument you have.

 

[restorer-john]

The main problem now is keeping a laptop going with a PCMCIA slot.

Pick up an old Stinkpad R61. Indestructible beast.

 

[Blumlein 88]

Do you have a link that details this?

Bob Stuart referred to it as "perceptually lossless" which obviously could just be called lossless unless it isn't. And it isn't.

It isn't lossless even at 44.1 because they use the 15th bit to identify it as MQA and to hold information about the filters to use. The other bits below the 13th bit in CD versions are where they fold the low bit rate lossy info for ultrasonic frequencies. So that part can't be lossless for 44.1/16 bit either.

If you search for posts by mansr he has done more than most to figure out what is going on with MQA. Both here and more so over at computeraudiophile.

 

[restorer-john]

Last I checked, Meridian is still in business, decades after it started unlike many high-end audio companies that have come and gone. My car infotainment system has Meridian logo on it.

The company was sold out in 2007 onwards and the car infotainment systems are made in China, not by, or in Meridian factories. The "Meridian" label is just a sticker, and an expensive one for the company who really only has MQA and some Chinese OEM car stereos as its claim to fame and income stream.

The brand will disappear soon enough.