• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

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

Status
Not open for further replies.

Rottmannash

Major Contributor
Forum Donor
Joined
Nov 11, 2020
Messages
2,969
Likes
2,606
Location
Nashville
Right, even if I will correct it a bit : he acquired a majority stake, Square is not owning 100% of Tidal. Jay-Z and the others are still there, Jay-Z got $300M, and the other $8M each but they are still there.
Do you know the percentage each own? Does this mean Jack controls the future of Tidal or just strongly influences it?
 

RichB

Major Contributor
Forum Donor
Joined
May 24, 2019
Messages
1,946
Likes
2,611
Location
Massachusetts
@RichB Rich, All good points. But I think they have been made here already…Respectfully…

Fair enough.
Phase 2 of this thread had some good data that you summarized well.
The 20 bit discussion for example. However, MQA is pushing encoding to 16-bit which is inferior at all levels.
That point seems to get lost.

The thread has gotten less technical, had more threats, despite protests to the contrary.
Technical questions, are address by with discussion of credentials and pedigree.
It's depressing.

Did anyone ever address the Bruno Mars analysis posted by Archimago?
If we are discussing the data, triangles and such, what about this chart?

MQA_Bruno_Mars_-_Imaging.png

Anyone, anyone, Bueller? :)

- Rich
 

DimitryZ

Addicted to Fun and Learning
Forum Donor
Joined
May 30, 2021
Messages
667
Likes
342
Location
Waltham, MA, USA
Fair enough.
Phase 2 of this thread had some good data that you summarized well.
The 20 bit discussion for example. However, MQA is pushing encoding to 16-bit which is inferior at all levels.
That point seems to get lost.

The thread has gotten less technical, had more threats, despite protests to the contrary.
Technical questions, are address by with discussion of credentials and pedigree.
It's depressing.

Did anyone ever address the Bruno Mars analysis posted by Archimago?
If we are discussing the data, triangles and such, what about this chart?

View attachment 133804

Anyone, anyone, Bueller? :)

- Rich
Well that graphs sure looks wrong.

@Archimago and myself share a disc player - Oppo UDP-205 - which surely confirms our superb taste in equipment.

When it came out, it didn't support MQA and in my conversations with Oppo it seemed like it wouldn't - which is why I ended up with the excellent M2TECH Young III. It's only after the player was discontinued, was the MQA functionality enabled in the player. That required both a player firmware update and a DAC firmware update.

This player is the only inexpensive option available for playing the notorious MQA-CDs, three of which I purchased from Japan.
 
Last edited:

RichB

Major Contributor
Forum Donor
Joined
May 24, 2019
Messages
1,946
Likes
2,611
Location
Massachusetts
Well that graphs sure looks wrong.

@Archimago and myself share a disc player - Oppo UDP-205 - which surely confirms our superb taste in equipment.

When it came out, it didn't support MQA and in my conversations with Oppo it seemed like it wouldn't. It's only after the player was discontinued, was the MQA functionality enabled in the player. That required both a player firmware update and a DAC firmware update.

This player is the only inexpensive option available for playing the notorious MQA-CDs, three of which I purchased from Japan.

I was an Oppo Beta tester and there are others here. :)

On point, there is no evidence that this is a player issue.
It could be an encoding parameter error, operator error, or a flaw in MQA triggered by the source material.
This was presented to MQA 4 years ago and, perhaps there has been a response, but I've not seen it.

- Rich
 

earlevel

Addicted to Fun and Learning
Joined
Nov 18, 2020
Messages
545
Likes
776
Here is the MQA-CD web page: MQA-CD


Technically, if the master is recorded at > 44.1 kHz and 24 bit, this statement is clearly false.
It safe the say the mastering engineer producing the desired sound without MQA.

If MQA did not alter it, the it would have the same dynamic range and frequency response as a RedBook CD.
This is clearly false since bits of fully audible potential dynamic range has been consumed.

Of course, MQA could sound the same, so it was pointless.
If MQA sounds different, then they have violated the mastering engineers intent.

Perhaps, in MQA classic form they have moved away from the "intent" into the sound in the studio, which of course is meaningless for music mastered with headphones.

Steve Guttenberg makes this point very well...
Forgive me for rehashing just a little bit in this paragraph, but it will make my point clearer. I've said be that, to me, MQA boils down to two things: 1) Compression, and 2) Potential (because I'm skeptical) audio improvement over what it's compressing (24/96k, for instance). And again, I feel the compression is too mild to produce enough value for the ensuing baggage of MQA. However, if the compression is not compelling, but the audio improvement is real and significant, an alternate implementation that would have far less intrusion on the audio chain and industry would be to build the process in players and stream 24/96k (if that's the bar) instead of MQA.

Now, I realize an argument could be that the encoding process (for deblurring or wherever the supposed improvement is) needs to be tuned by an expert and can't be done effectively by automation. If so, that just doubles down the issue with mastering engineers—someone else is deciding how it should sound. (Obviously, it's debatable whether anything was improved if someone other than the people in charge of the music production are tasked with "improving" it.)

But absent that, if the deblurring/etc. were built into DACs, then mastering engineers and studios would own such DACs and none of this would be an issue. A happy side effect of that is that if the improvement proves to be—or not to be—real or significant, it's going to be obvious out of the gate. MQA succeeds or fails right away and we move on, without first "re-imagining" an industry. :p
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,374
Likes
234,454
Location
Seattle Area
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
"Appears" is the key word here. A bunch of guessing on Archimigo's part with incorrect assumptions. I just analyzed the spectrum of this 88 kHz. Here is the scan of the first minute or so of this track:

1622854475330.png


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:


1622854364694.png


Note how low the levels are: we are talking about -130 dB to -190 dB. This can very well be noise shaping introduced by MQA and in the low order bits of 24 bit samples. It is taking advantage of the extra spectrum it has due to 88 kHz sampling to push the dither noise into that region.

I went to look for the original master and nothing beyond 44.1 kHz/24-bit is available. This support the notion then that the roll off is correct around 22 kHz and what is after is noise shaping.

You see what the problem is with reading the tea leaves as Archimago is doing? You make a bunch of assumptions that are too superficial and gets you in trouble. You have to have the original source file and encoder to do proper analysis. Otherwise with tendency to NOT give benefit of doubt to MQA, you arrive at conclusions that are simply wrong.

So if I were Archimago, I would not put this clip and example forward as proof that anyone could find this information. Anyone can but clearly most are not qualified to know what they are looking at and potential causes behind that graph.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,374
Likes
234,454
Location
Seattle Area
Indeed.
Can you post the spectrum of the complete track, please?
I post what I captured (50 seconds). I can't stand Bruno Mars so don't make me deal with the whole track!!!
 

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
I post what I captured (50 seconds). I can't stand Bruno Mars so don't make me deal with the whole track!!!
I'm not sure I understand your problem with the criticism of MQA-ing this track. Whether it has faked ultrasonics or no ultrasonics, is there any reason to convert and put this up as a so-called hi-res track?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,374
Likes
234,454
Location
Seattle Area
I'm not sure I understand your problem with the criticism of MQA-ing this track.
??? My "problem" was Archimago automatically thinking the problem was MQA encoding without access to the original to know what its spectrum looked like.

Whether it has faked ultrasonics or no ultrasonics, is there any reason to convert and put this up as a so-called hi-res track?
Yes there is a reason. A label gave them 88 kHz 24-bit content so they encoded it. Even outside of sample rate, I value 24 bit bit depth and don't want someone else to chop that down to 16 bits.
 

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
??? My "problem" was Archimago automatically thinking the problem was MQA encoding without access to the original to know what its spectrum looked like.


Yes there is a reason. A label gave them 88 kHz 24-bit content so they encoded it. Even outside of sample rate, I value 24 bit bit depth and don't want someone else to chop that down to 16 bits.
Because that track (and don't deny that you secretly love it) is so dynamic you need >16 bits?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,374
Likes
234,454
Location
Seattle Area
Because that track (and don't deny that you secretly love it) is so dynamic you need >16 bits?
It doesn't matter if you "need it." Once you encode it into 24 bits, if you don't convert it properly to 16 bits you can distort it. If you want 16 bits, 16 bit capture works better. But then that can cause problems in post production. So 24 bit is used and if there was a law that every mastering person converted it using proper dither life would be good. But there is no such law.

I just don't understand why people insist on getting 16 bits when 24 bit version exists.

And oh, I absolutely hate Bruno Mars music. He stole all the Motown music, change them a bit and called them his own.
 

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
It doesn't matter if you "need it." Once you encode it into 24 bits, if you don't convert it properly to 16 bits you can distort it. If you want 16 bits, 16 bit capture works better. But then that can cause problems in post production. So 24 bit is used and if there was a law that every mastering person converted it using proper dither life would be good. But there is no such law.

I just don't understand why people insist on getting 16 bits when 24 bit version exists.

And oh, I absolutely hate Bruno Mars music. He stole all the Motown music, change them a bit and called them his own.
OK, that's fine. I'll take 16 bits but if you prefer higher that's fine. Back to the track, let's look at it purely out of curiosity instead of teying to defend a position. Can you give the spectrum of the whole track so we can even have an idea if it matches the "gapped" example that seems to show faked ultrasonics? Did they redo this, possibly?
 

DimitryZ

Addicted to Fun and Learning
Forum Donor
Joined
May 30, 2021
Messages
667
Likes
342
Location
Waltham, MA, USA
It doesn't matter if you "need it." Once you encode it into 24 bits, if you don't convert it properly to 16 bits you can distort it. If you want 16 bits, 16 bit capture works better. But then that can cause problems in post production. So 24 bit is used and if there was a law that every mastering person converted it using proper dither life would be good. But there is no such law.

I just don't understand why people insist on getting 16 bits when 24 bit version exists.

And oh, I absolutely hate Bruno Mars music. He stole all the Motown music, change them a bit and called them his own.
I would gently suggest you are wrong about Bruno.

His talent (and its' a considerable one) is to re-stylize older RB and Soul music for modern audiences. The arrangements of his music are quite different than would have been done back in the day.

My dad was a jazz musician who introduced Soviet audience to Brazilian Bossa Nova in the late 1960s and early 1970s. When I listen to his old Melodya records I hear Antonio Carlos Jobim, but also him. This practice is very common among musicians and it's more of a homage than "stealing." A form of musical dialogue - and the recipient doesn't have to be alive. Jazz musicians have been talking to Parker, Coltrane, Mingus and Ella long after their death.
 
Last edited:

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
snip

And oh, I absolutely hate Bruno Mars music. He stole all the Motown music, change them a bit and called them his own.
He changed it for the worse. His talent is taking songs I often like and recording them so I can't stand them.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,374
Likes
234,454
Location
Seattle Area
OK, that's fine. I'll take 16 bits but if you prefer higher that's fine. Back to the track, let's look at it purely out of curiosity instead of teying to defend a position. Can you give the spectrum of the whole track so we can even have an idea if it matches the "gapped" example that seems to show faked ultrasonics? Did they redo this, possibly?
You didn't read my response??? I captured only 50 seconds and gave you the spectrum for the whole 50 seconds in the first graphics. What do you expect to find in the rest of it that is not in the first 50 seconds?
 

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
You didn't read my response??? I captured only 50 seconds and gave you the spectrum for the whole 50 seconds in the first graphics. What do you expect to find in the rest of it that is not in the first 50 seconds?
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?
 

RichB

Major Contributor
Forum Donor
Joined
May 24, 2019
Messages
1,946
Likes
2,611
Location
Massachusetts
Forgive me for rehashing just a little bit in this paragraph, but it will make my point clearer. I've said be that, to me, MQA boils down to two things: 1) Compression, and 2) Potential (because I'm skeptical) audio improvement over what it's compressing (24/96k, for instance). And again, I feel the compression is too mild to produce enough value for the ensuing baggage of MQA. However, if the compression is not compelling, but the audio improvement is real and significant, an alternate implementation that would have far less intrusion on the audio chain and industry would be to build the process in players and stream 24/96k (if that's the bar) instead of MQA.

Now, I realize an argument could be that the encoding process (for deblurring or wherever the supposed improvement is) needs to be tuned by an expert and can't be done effectively by automation. If so, that just doubles down the issue with mastering engineers—someone else is deciding how it should sound. (Obviously, it's debatable whether anything was improved if someone other than the people in charge of the music production are tasked with "improving" it.)

But absent that, if the deblurring/etc. were built into DACs, then mastering engineers and studios would own such DACs and none of this would be an issue. A happy side effect of that is that if the improvement proves to be—or not to be—real or significant, it's going to be obvious out of the gate. MQA succeeds or fails right away and we move on, without first "re-imagining" an industry. :p

Any "tuning" of MQA during the process to improve the source would be a subtle attempt at remastering that comes with a restriction of using the MQA leaky filter.
I pro-filter choice ;)

- Rich
 

RichB

Major Contributor
Forum Donor
Joined
May 24, 2019
Messages
1,946
Likes
2,611
Location
Massachusetts
"Appears" is the key word here. A bunch of guessing on Archimigo's part with incorrect assumptions. I just analyzed the spectrum of this 88 kHz. Here is the scan of the first minute or so of this track:

View attachment 133810

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:


View attachment 133809

Note how low the levels are: we are talking about -130 dB to -190 dB. This can very well be noise shaping introduced by MQA and in the low order bits of 24 bit samples. It is taking advantage of the extra spectrum it has due to 88 kHz sampling to push the dither noise into that region.

I went to look for the original master and nothing beyond 44.1 kHz/24-bit is available. This support the notion then that the roll off is correct around 22 kHz and what is after is noise shaping.

You see what the problem is with reading the tea leaves as Archimago is doing? You make a bunch of assumptions that are too superficial and gets you in trouble. You have to have the original source file and encoder to do proper analysis. Otherwise with tendency to NOT give benefit of doubt to MQA, you arrive at conclusions that are simply wrong.

So if I were Archimago, I would not put this clip and example forward as proof that anyone could find this information. Anyone can but clearly most are not qualified to know what they are looking at and potential causes behind that graph.

The levels your talking about are -130 to -190 dB in the 88 kHz track and not the MQA encoded version.
It could be noise shaping with noise levels introduced my MQA.
If your analysis is not the MQA track, then this is speculation.

Experts speculate, novices assume. :)

Me thinks Archimago is in trouble because he asks the wrong questions.

- Rich
 
Status
Not open for further replies.
Top Bottom