• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

MQA Deep Dive - I published music on tidal to test MQA

Status
Not open for further replies.

Raindog123

Major Contributor
Forum Donor
Joined
Oct 23, 2020
Messages
1,209
Likes
2,414
Location
Melbourne, FL, USA
My understanding, the problem with that song is not as much the dip as it is this. Is this what you guys call “fake ultrasonics”?

If indeed true, this is huge - as this is not some ‘blogger’ not knowing how to use the encoder, but the official release aliases the entire baseband into the ‘ultrasonics’, and leaves nothing else there!

The love or hate to Bruno Mars simply has to be tucked away here, and the case must be investigated by those equipped to do it - beyond the first 50 seconds of the track.
 
Last edited:

Docmoggy

Member
Joined
Mar 13, 2021
Messages
31
Likes
8
I can say that your analysis did indeed help me in realizing that I fell for the hype! I’ve cancelled my Tidal hifi subscription.

My question is what is the best value streaming service available in the market to achieve hi resolution lossless audio? I mean, it’s a great idea to hear master quality music, but it doesn’t appear that possibility is being provided by MQA codec.

is it Qobuz, Apple etc??
 

Raindog123

Major Contributor
Forum Donor
Joined
Oct 23, 2020
Messages
1,209
Likes
2,414
Location
Melbourne, FL, USA
Note that it doesn't show anything like he has. If I however I play the music and watch the spectrum, you do see a hump that correlates with music…


Amir, what are the green and blue curves in your plots?
 
Last edited:

DimitryZ

Addicted to Fun and Learning
Forum Donor
Joined
May 30, 2021
Messages
500
Likes
202
Location
Waltham, MA, USA
My understanding, the problem with that song is not as much the dip as it is this. Is this what you guys call “fake ultrasonics”?

If indeed true, this is huge - as this is not some ‘blogger’ not knowing how to use the encoder, but the official release aliases the entire baseband - and leaves nothing else - around the Nyquist frequency!

The love or hate to Bruno Mars simply has to be tucked away here, and the case must be investigated by those equipped to do it - beyond the first 50 seconds of the track.
@amirm is making a case it's shaped noise from a 44KHz master, upsampled to 88KHz by the label before asking MQA to encode it. Not unlike very high level ultrasonics in DXD, but much lower in level.

On the other hand, MQA bandsplitting process, due to their choice of low slope filters (see Jim Lesurf's writeup from years ago and werner's recent very good concise explainer on PFM) does allow aliases to creep into both bands, but according to MQA's claims, at entirely inaudible level. At -130dB to -190dB, this claim is met, if that's what we are observing (Amir's numbers). At -96dB (Archimago's numbers) it's close, but no cigar. Perhaps an inadvertent test occured - and they re-encoded to improve the numbers.

So maybe not huge. We shall see.
 
Last edited:

levimax

Major Contributor
Joined
Dec 28, 2018
Messages
1,085
Likes
1,523
Location
San Diego
I can say that your analysis did indeed help me in realizing that I fell for the hype! I’ve cancelled my Tidal hifi subscription.

My question is what is the best value streaming service available in the market to achieve hi resolution lossless audio? I mean, it’s a great idea to hear master quality music, but it doesn’t appear that possibility is being provided by MQA codec.

is it Qobuz, Apple etc??
I switched from Tidal to Qobuz to get away from MQA and it seems fine... interface a little cluckier but it works. It is also $5 per month cheaper than Tidal. Amazon Prime HD is OK but it has or at least had issues with WASAPI when I tried it. I am going to give Spotify a serious look when their new HD service comes out as it seems to have the best interface and suggestions.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
35,358
Likes
135,597
Location
Seattle Area
So how do you explain the difference from the other one. Do you think error by whoever made the graph or did they redo the MQA?
I don't know how he captured it. I used digital capture from Roon. Maybe he did an analog capture. But yes, it is very possible they re-encoded it as well.
 

bennetng

Major Contributor
Joined
Nov 15, 2017
Messages
1,380
Likes
1,311
Not related to MQA or filter-induced imaging, but faking a hi-res spectrum without "gap" and mirroring, and with correlated spectral content is entirely possible:
https://hydrogenaud.io/index.php?topic=108864.msg948350#msg948350

Therefore, if there is no original hi-res file to compare with, a nice looking spectrogram can still be fake, and I don't even think my faking methodology is good enough. There must be someone else or some better tools capable of making better looking fakes.

Of course, people can just download my fake file and point out here and there to explain why it is fake, but it would be pointless because I already told everyone it is a fake, so don't do that.
 

Raindog123

Major Contributor
Forum Donor
Joined
Oct 23, 2020
Messages
1,209
Likes
2,414
Location
Melbourne, FL, USA

voodooless

Major Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
2,371
Likes
3,544
Location
Netherlands
Archimago's Musings: MUSINGS: On the RMAF 2018 MQA talk, pseudonyms, and the right to anonymity.

View attachment 133797



This appears to be real music with missing data surrounding 22 kHz.
I have seen many attacks on the Archimago but no direct comment on the data.

- Rich

How is this real music? It’s obviously a perfect aliasing image as I showed at least two times already in this thread. See https://www.audiosciencereview.com/...-music-on-tidal-to-test-mqa.22549/post-755493 for instance.

1622872725866.png
 
Last edited:

DimitryZ

Addicted to Fun and Learning
Forum Donor
Joined
May 30, 2021
Messages
500
Likes
202
Location
Waltham, MA, USA
How is this real music? It’s obviously a perfect aliasing image as I showed at least two times already in this thread. See https://www.audiosciencereview.com/...-music-on-tidal-to-test-mqa.22549/post-755493 for instance.

View attachment 133843
It very well maybe, as MQA bandsplitting produces it by a choice of low slope filters, trading low level aliasing for time domain integrity.

If the level is -130dB+ (@amirm) then MQA claims holds, if -96dB is where it is (@Archimago ), then close but not quite right.
 

noobie1

Active Member
Joined
Feb 15, 2017
Messages
226
Likes
144
Location
Bay Area
I think we need a DMCA lawyer to address whether or not MQA content qualifies as having DRM protection.

If someone were to develop an alternative decoder to completely unfold the MQA content, MQA might claim that the content was encrypted. And if the courts found that to be true, the use of the alternative decoder would be a DMCA violation.

If the alternative decoder doesn't infringe on MQA's patents, why would this matter? Isn't DMCA a matter for who owns the music?
 

DimitryZ

Addicted to Fun and Learning
Forum Donor
Joined
May 30, 2021
Messages
500
Likes
202
Location
Waltham, MA, USA
If the alternative decoder doesn't infringe on MQA's patents, why would this matter? Isn't DMCA a matter for who owns the music?
@mansr has produced such a public domain decoder several years ago, based on Bluesound's MQA firmware. By all accounts, it works just fine. He seems to be free :)
 
Last edited:

noobie1

Active Member
Joined
Feb 15, 2017
Messages
226
Likes
144
Location
Bay Area
I am not saying with certainty that MQA removed my files.
I have made very clear where facts end and my opinion/suspicions start. And addressed this point in the video and say quite clearly that this is my opinion and not something I can prove.

However I DO find the timing suspicious, and the fact that the publisher refused to answer my question about whether MQA requested the removal of the files (and how MQA's point #1 was worded), does not help.

But, these are of course opinions/suspicions and I am not presenting them as anything else.

MQA however has been quite clear about their encoding being lossless despite the fact that it is not. And dodges the question whenever they can.

View attachment 132331

I'm not dodging questions, I'm giving clear and transparent answers.

If MQA were advertising their product as something that they were of the OPINION that it is perceptually better, then that would be fine. We wouldn't have an issue. It's totally ok to have an opinion and to market a product on opinion as long as it is clear that that is what it is.
But marketing it as lossless OR perceptually better with no clear evidence for either is wrong.

I'm not really sure what else I can say here.

IF MQA or Bob Stuart reached out to someone at Tidal or your publisher to take down the files, I doubt it would be to the guy who responds to emails/calls.
 

noobie1

Active Member
Joined
Feb 15, 2017
Messages
226
Likes
144
Location
Bay Area
DMCA makes illegal circumvention of copy or other IP protection, IIRC. It would be clearly unconstitutional in the U.S. if it did not implement two International treaties, IIRC, IMHO, & etc. It is a gross over-extension of intellectual property law, which is authorized in the U.S Constitution for the purpose of “progress in the sciences and useful arts,” IMHO. The U.S. had a constitutional safe-harbor kind of common sense personal or educational use doctrine called “fair use” but the DMCA greatly jeopardizes that doctrine it seems to me.

Long story short: If you are messing with stuff that involves circumvention of IP protection, be careful out there. The rules far exceed any “rule of reason” or socially useful doctrine of limitation or any normal type of effort to narrowly tailor the law to achieve the intended constitutionally authorized objective. IMHO. The law has a kind of special permission to go too far because it is enacted pursuant to two international treaties. IMHO, YMMV. :)

Actually my point was that MQA wouldn't be a party to DMCA dispute as they don't own the music rights. They can only be a party to any patent disputes.
 

DimitryZ

Addicted to Fun and Learning
Forum Donor
Joined
May 30, 2021
Messages
500
Likes
202
Location
Waltham, MA, USA
Actually my point was that MQA wouldn't be a party to DMCA dispute as they don't own the music rights. They can only be a party to any patent disputes.
That's probably not correct re. public domain decoder. This is strictly MQA patent issue.

And in case of the current public domain decoder developed by @mansr , they chose not to legally pursue this matter.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
35,358
Likes
135,597
Location
Seattle Area
I think we need a DMCA lawyer to address whether or not MQA content qualifies as having DRM protection.

If someone were to develop an alternative decoder to completely unfold the MQA content, MQA might claim that the content was encrypted. And if the courts found that to be true, the use of the alternative decoder would be a DMCA violation.
MQA content is distributed with no copy protection. I see no reason for the labels to bring any kind of DMCA claim against you or even want to be involved. The business end of this will be patent and copyright infringement of the code, not the content.
 

DimitryZ

Addicted to Fun and Learning
Forum Donor
Joined
May 30, 2021
Messages
500
Likes
202
Location
Waltham, MA, USA
MQA content is distributed with no copy protection. I see no reason for the labels to bring any kind of DMCA claim against you or even want to be involved. The business end of this will be patent and copyright infringement of the code, not the content.
Can we all please stop "MQA = DRM meme? It's honestly ridiculous.
 

voodooless

Major Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
2,371
Likes
3,544
Location
Netherlands
It very well maybe, as MQA bandsplitting produces it by a choice of low slope filters, trading low level aliasing for time domain integrity.

If the level is -130dB+ (@amirm) then MQA claims holds, if -96dB is where it is (@Archimago ), then close but not quite right.

Once again: the band splitting is supposed to be lossless.. And slow filters don’t have very well time domain integrity.

This looks much more like the described upsampling the renderer does. You would also get something similar if you’d use linear interpolation for upsampling. Some other form of interpolation other than a decent low pass would give similar results, no matter how fancy the name is.

Whatever it is, it should not be there!
 

Hayabusa

Senior Member
Joined
Oct 12, 2019
Messages
308
Likes
243
Location
Netherlands
I could not let this one slide :p Have you actually looked at the graph:

index.php


Notice anything strange except for the hole in the middle? No? Then let's do some "folding" and show you (I've done this one before, but one more won't hurt anyone):

View attachment 125593

See whats going on here? The HF is a perfect mirror image of the LF, just with a lower level and a sloping.

You know what they call that in circles outside of MQA: aliasing... of the worst kind! So please stop telling us that MQA somehow solves aliasing, because all evidence points towards the expact opposite.

@amirm: why do you condemn this kind of behaviour when reviewing DAC's, but not when MQA does it? If you were given an upscaler to review that showed the exact same properties, I bet you it would get a scolding review.

You get this by upsampling by just duplicating samples. Same effect as a NOS DAC with a sample and hold on 44.1 KHz
 
Status
Not open for further replies.
Top Bottom