2L MQA Comparisons: Supplementary Listeners’ Notes For The 2L Test Bench

The following document was provided by MQA and contains information and listening notes relating to files from record label 2L which have been MQA encoded.

Bob Stuart (MQA) & Morten Lindberg (2L), Jan. 2016

For some time 2L has maintained a webpage giving access to high definition music files. Recently both MQA and CD release versions have been added for some tracks. (www.2l.no/hires)

Where provided, MQA is available in only one form, having been encoded directly from the original master in its native resolution. The test bench includes several encoded from DXD as well as:

  1. 2L-120-01 Carl Nielsen: Chaconne op 32 (originally recorded at 44.1 kHz)
  2. 2L-048 Ola Gjeilo: North Country II (originally recorded at 96 kHz)
Playback
To get the best result it is important to ensure that your computer is not modifying the output and that you are using a bit-accurate player without post-processing or level control.1

If you have an MQA decoder then the light or display will confirm MQA playback. When the MQA indicator shows then the signal path is bit accurate.

About the MQA Files
MQA uses a process called ‘music origami’ to ‘fold’ a high-sample-rate signal down to a smaller, lower-data-rate file which can be played back without a decoder. 2L-048 is folded once to 48 kHz 24bit; 2L-120 has no folds; the others are folded three times from 352.8 to 44.1 kHz 24 bit.

An MQA decoder will restore the original recording and ‘unfold’ it to optimally match its D/A converter. So, for example, on a mobile device, 2L-111 (which is from DXD) can be unwrapped to 44.1, 88.2 or 176.4 kHz whereas a higher-performance DAC can unfold all the way to 352.8 kHz. The ability of one file to be used in several contexts adds considerably to convenience; you don’t need to make a down-sample of big download files to play them in your car, ‘phone or ‘legacy’ equipment. If you have an MQA file it can be listened to without any decoder, it’s a simple FLAC file.

The ‘origami’ process embeds processes that make each version and the undecoded rendering highly optimal. For that reason, MQA claims that the sound quality of the MQA file, when played without a decoder, is at least equal to CD. To facilitate this comparison, 2L have added their CD release of the songs in question.

If you have an MQA decoder then the sound quality will be extremely high and in most cases will avoid shortcomings in the original PCM and uncompensated DAC.

Morten’s Notes
Listening with the Prime I would range the listening quality in the following order:

Ranking With Decoder (Meridian Prime) No Decoder (Oppo HA-1)
1 MQA DXD
2 DXD MQA
3 CD 44.1 /16 CD 44.1 /16

In this ranking therefore the un-decoded MQA is preferred to the CD and with a decoder it is preferred to the original.

Note: the 24-bit Master and MQA (decoded) peak and noise curves overlay and are not separately visible.

These graphs confirm that 2L’s Original, CD and MQA versions of the files are consistent in level and response. Of course spectral plots using FFT have no time-domain information, but we can use them to compare the peak spectrum of the Original, CD, and MQA with and without a decoder.

Also shown is a comparison of the background noise throughout each version and the reference level for 16-bit tpdf dither in a channel sampled at 44.1 kHz. 2 3

In both graphs the peak and noise-floor curves overlay for both MQA decoded and Original master.

In 2L-111-15 we can see that the noise floor in the recording approaches the 18 bit level by 20 kHz and is reproduced by MQA.

We can also see that the shaped HF noise introduced by the MQA encoder and ‘heard’ without a decoder is removed by the decoder and is also below that of the CD release, even without decoding.


1 There are notes on the webpage to help MAC users. PC users can safely use Foobar, also in Exclusive mode.
2 The analysis uses 21.53Hz bins (=44100/2048 and 351800/16384) giving an offset of +13.33 dB wrt 1Hz.
3 2L sensibly use shaped quantisation for their CD releases.

COMMENTS
Casimir's picture

"To get the best result it is important to ensure that your computer is not modifying the output..."
Does it mean that the USB Regen and Jitterbug shouldn't be part of the audio rig ?

Michael Lavorgna's picture
...are concerned with physical connectivity issues, if you will, which is separate and apart from MQA.
Reed's picture

When I look at them, it appears that CD actually outperforms from a noise floor perspective below 15khz, which I thought was the heaviest band the human ear is capable of discerning. I'm sure I'm missing something, but the 120 graph is so close to CD that I can tell it's a wash below 15khz. The 111 graph, however, shows a drastic shortcoming in terms of noise floor below 15khz.

Again, I may be interpreting it wrong, but it would help to lend some understanding as these graphs are becoming more prevalent.

Michael Lavorgna's picture
...is alert MQA to your questions since this is their baby.
Hi-Reality's picture

Dear Michael,

Is there currently any off-the-shelf MQA Encoder / Analogue Digital Converter in the marketplace that you heard of? (I have been sifting through the web for hours to find an answer)

Thanks so much in advance!

Regards, Babak
Hi-Reality Project

Ps. I sent the same question to Jason under his 'Meridian Updates its Products to MQA' post at Stereophile.com

Michael Lavorgna's picture
However, I do know that Mytek is working on one, so you may want to reach out to them re. availability.
Miska's picture

For fun, I made conversion of the 2L-111 to 120 kHz sampling rate 18-bit FLAC. The result was significantly smaller than the MQA file (less bandwidth consumption), while preserving more of the source content while not requiring special decoding capabilities. 120/18 FLAC preserves both all the harmonic content of the source and all dynamic range of the content, while being standard FLAC...

96 kHz 18-bit FLAC would be even more standard and even smaller...

Why such variants are not included in the comparison?

MQA encoded content in FLAC container has significant size overhead, because the encoded MQA content looks like white noise from FLAC encoder point of view and thus is not compressible. Resulting in relatively huge 48/24 FLAC files (larger than 96/20 or 120/18).

Michael Lavorgna's picture
I'm not sure if Bob Stuart is going to respond here, he seems to be rather busy ;-). From my perspective, the most relevant question at present is -- how does MQA sound when decoded using an MQA-enabled DAC (one just arrived). Question 2 is how does an MQA-encoded file sound when played through a non-MQA DAC?

I understand and appreciate the work you and Juergen (and others) are doing as well as the valid questions/concerns that have come up re. DRM, etc. The unfolding story (pardon the pun) begins with MQA's performance as it relates to how the process sounds, imo.

Casimir's picture

Hi Michael,
which an MQA-enabled DAC will you use for the sound tests?

Michael Lavorgna's picture
.
Casimir's picture

If you've got the Prime Power Supply too, try the Audioquest Diamond USB cable between the amplifier and the power supply and... enjoy :). Try the Jitterbugs from the computer side and the USB Regen from the PSU side. They make a difference and it's my favorite combination.
My headphones are Hifiman HE-560. Audeze sounds with Prime very good, too.

Casimir's picture

And for the USB Regen you can use the Prime Power Supply, too.
You need an additional 12V power cable. In the combination with the linear power supply the USB Regen sounds better, too.

Casimir's picture

Have a look hier:
http://www.2l.no/hires/index.html
the MQA files are 7-8 times smaller as the originals.

Miska's picture

MQA version of 2L-111 is 16.7 MB in size, standard 120/18 FLAC is 13 MB in size, 176.4/18 FLAC is 17.3 MB in size and 96/18 FLAC is 13 MB in size.

MQA itself has 17-bits of resolution. So standard 176.4/17 FLAC would be same size or smaller.

Miska's picture

Sorry 96/18 standard FLAC is 12 MB in size.

Casimir's picture
JR_Audio's picture

Hi Michael

Thank you, and good to see, that the graphs from 2L do fill some gaps on the MQA origami.

The non MQA decoded graphs, do partly match, what is "guessed" by Archimago, Miska and me.

And it looks like, that you have also read my comments on Archimago and on Computer Audiophile. :-).

Looking forward to your report after the MQA firmwares are released and the answers from Bob to CA.

Juergen

JR_Audio's picture

Really, I wished, there would be more information about the MQA origami from the technical point of view. At this moment, we all do mainly guess, what is going on.

I think you have read my listening impressions on MQA files on non MQA converters. The sound is pleasing and non fatiguing at all, but lesser resolution, than CD format and that respect, that even when the main part of the music has a very good sense of timbre and paste (what makes listing so “pleasing”), it looses information about the room reverberations and details, when looking at the “great picture” (the main instruments and the room reflexions at the same time). I know, it is a bit difficult to explain, so I hope, it does not lead to more confusion.

But fact is, that from the technical point if view, it is a lossy codec, but doesn't mean, that the psychoacoustic model is not good enough, that at the end, from ADC to DAC, when fed this codec with high res informations (bandwidth and timing) the outcome for the ear is superior.

But shouldn't this not can't be done with “plain” PCM files in 24 Bit 96 k (or a bit higher), with slow roll off digital filters for ADC and DAC? Then we would have a file format, that sounds good, and that is compatible with digital volume control and digital room DSPs?

Hopefully in one months or so, we all do knew more and we can leave the field of guessing and enter the field or knowledge, about the MQA origami.

Until then. Enjoy the music. This is what is all about.

Juergen

Casimir's picture

Bob Stuart explains:
https://youtu.be/BrgjycGhoSM

Casimir's picture

Well,
the 2L MQAs sounds for me much better than my 2L 24/192 records...
I think I have to buy some records again :)

Tapetech's picture

So I go to convert my new Explorer2 to MQA and find out my MAC OS is not current enough. One needs OS 10.9 or higher. My old MAC will not accept 10.9 so I will have to find out how to upgrade some other way.

The Explorer2 info states the are are two MQA modes:

-a "MQA file" which is indicated by a green LED. I believe it means the music was recorded with an MQA A>D converter. And possibly this is the highest quality(?) And does this also mean the recording sample rate is always 352kHz?

-a "MQA Studio File" which is indicated by a blue LED. I believe this means the music was recorded by a non-MQA A>D, but that the characteristics of that particular A>D have been measured by Maridian and that time-domain "corrections" have been applied to the file, this making it an MQA file. And the recording sample rate can be any sample rate.

Anyone correct me if the above is not accurate. So would the blue light "MQA Studio File" almost always be lesser quality than the green light "MQA File"

Casimir's picture

@Tapetech
If you have a green or blue light the audio is identical. The difference is only about it is the mastering studio or if it is the artist personally that have signed off.

Tapetech's picture

Yes, I just saw a comment yesterday from Bob Stuart saying the same thing.

However, on their website they say things like:" The listener’s decoder reconstructs the entire original signal, bringing up an indicator confirming that what they are hearing is exactly the same as the original master.... Its most advanced form, “MQA Studio”, uses the technology in the studio to generate an even higher quality master recording as the source."

It sort of suggests that there are technical differences between MQA and MQA Studio, but I guess not.

Casimir's picture

@Tapetech
The information send me Mr.Lindberg from 2L
Where have you seen Mr.Stuart answer?

Tapetech's picture

Turns out the info was from MQA Ltd and not directly from Mr Stuart.

"Edit 2: This just in from MQA ltd., "There is no sonic difference between files marked as green or blue, it is only about Provenance or Approval." In addition, "Today Alan Silverman asked us to move the Judy Collins [album] up to Studio."

http://www.computeraudiophile.com/content/681-my-first-24-hours-mqa/comm...

X