Tidal Masters & MQA Thread!

[pinnahertz]

  Why is it useless?  Your argument has been that when people say an MQA stream on Tidal sounds better than the Redbook version Tidal stream, they don't know what they are comparing.  Ignoring the fact that only a tiny number of re-masters have been done (Warner has acknowledged that 95%+ of their MQA conversions are automated encodings of their current master, not re-masters) so they were in almost all cases comparing from the same master, the 2L tracks take away any question.  You can directly compare Redbook and MQA encodings of the same source master.  Isn't that exactly what you were saying isn't possible?

Unless I'm not reading right, it seems the 2L files were native DXD, isn't that correct? That means they are transcoded to get to Redbook, and separately transcoded to get to MQA. Those are separate paths, and thus not directly comparable. The Warner files need to be verified, there have been far too many "masters" around and since we don't have any way to do that, those are not valid test files. I realize how pedantic this sounds, but unless we have total verifiability, we can't be sure of what is being auditioned.

That plot shows that the differences between software decoded Tidal MQA(Tidal only, no hardware decode), and a lossless, native 24/96 version of the same file, are minuscule and way below the ability of any human to differentiate(if you are an objectivist that is...), this comparison nulls at -90dB, objectively, period.

The "plots" are spectrograms, wrong tool for the job, and almost impossible to get any meaningful detail out of them. There are FAR better tools that could have been used. However, yes, I agree, a 90dB null is identical enough. I didn't get that in my tests, though I have a pretty good idea why.

The files become even more similar with hardware decoding, but since the difference was already inaudible, it's sort of moot (though I suspect many will say MQA decoded to 24/384 sounds better than software decoded to 24/96, but that, like so many other things is another argument...).  

The reality is, if two files null to a real -90dB, there can be no audible difference at all. So hardware decoding may be "better", but not audibly.  Given perfect timing match (not actually possible), the level match for a 50dB null must be within 0.017dB.  Assuming perfect level match, the phase alignment for a 50dB null must be under 6 degrees at all frequencies. So you see, a 50dB null represents no audible difference.  A 90dB null is practically perfect (and I believe impossible, BTW).  And, we're splitting inaudible hairs.

You said one cannot compare because there's no way to know what is being compared.  That is simply not true.

Beg to differ. I have yet to see an MQA file that was made, verifiably, from a specific 16/44 master.

There are lots of valid critiques of MQA, but saying it does not provide an acoustically transparent fascimile of HiRes is not one of them.

Agreed. I'm not saying that.

 People can subjectively compare themselves (with all the caveats that entails), something that anyone can actually do fairly easily, contrary to what you keep saying.  They can also objectively compare (as Archimago did), albeit not quite as easily.  

Until we have 100% verifiable and traceable production path from a specific 16/44 master to an MQA version, verified by someone other than a content provider (with their motives for difference and improvement) and MQA (with their motives for difference and improvement), we cannot do that comparison with any degree of validity. There are far too many players in the game with monetary stakes. All of those stakes are biases that must be under control.

 

[smartin]

I think I finally understand what you're arguing.
 
I doubt anyone in this thread is hanging their hat on the premise that MQA improves Redbook, and if so it would be a false premise, because MQA is not applied to native Redbook.  The labels that have signed up with MQA have only applied MQA to their HiRes masters.  While there are certainly some HiRes masters that are just upsampled crap (i.e. all of Madonna's stuff), a vast majority are native HiRes (depending on definition, there are 24/44 masters).  In most of those cases the Redbook version is transcoded from the same source (as is the case for all the 2L files).  There are no MQA files encoded from native Redbook, none.  It would make no sense to do so.  The patent doesn't even cover this scenario (it does include the possibility of encoding to a lower resolution than 24/48, but it's still from a HiRes master)
 
When people say they are enthusiastic for MQA, I suspect it's because they think it sounds good, better than the Redbook alternative that Tidal also has.  They are enjoying being able to stream a HiRes facsimile for the same price they pay to stream Redbook.
 
The results of the encoding process are the reproduction of a HiRes master.  If you enjoy that, then you are enjoying the results of the encoding.

 

[pinnahertz]

   I think I finally understand what you're arguing.
 
I doubt anyone in this thread is hanging their hat on the premise that MQA improves Redbook, and if so it would be a false premise, because MQA is not applied to native Redbook.  The labels that have signed up with MQA have only applied MQA to their HiRes masters.  While there are certainly some HiRes masters that are just upsampled crap (i.e. all of Madonna's stuff), a vast majority are native HiRes (depending on definition, there are 24/44 masters).  In most of those cases the Redbook version is transcoded from the same source (as is the case for all the 2L files).  

If you're saying that all the masters in the Warner catalog are hi-res, I'll have to doubt that strongly. The final mixdown would be to redbook in every single case because it must then be used to create the CD master, and the AAC/MP3 release. To think there are 24/96 masters for all of that is ludicrous. The multitracks, sure, but the final master, not a chance.  But masters are the stereo mixdown, equalized..."mastered", and ready for CD.  They're never going to release a non-"mastered" stereo mixdown.  And you don't release a CD, or standard file from 24/96. 

There are no MQA files encoded from native Redbook, none.

Proof? I have none either way.

It would make no sense to do so.  The patent doesn't even cover this scenario (it does include the possibility of encoding to a lower resolution than 24/48, but it's still from a HiRes master)  

MQA claims to be able, with a single filter, correct the impulse response anomalies of every single ADC and DAC used in production all the way back to analog masters. That's a whole lotta redbook. The improvement of Redbook is one of their biggest points.

When people say they are enthusiastic for MQA, I suspect it's because they think it sounds good, better than the Redbook alternative that Tidal also has.  They are enjoying being able to stream a HiRes facsimile for the same price they pay to stream Redbook.

And I say, "Nonsense". It's highly unlikely they are actually hearing any difference at all. But we don't know, so my argument isn't any better than yours.

The results of the encoding process are the reproduction of a HiRes master.  If you enjoy that, then you are enjoying the results of the encoding.

MQA has two points.  One is delivery of high resolution audio in smaller files.  The second is an improvement over the original by "correcting timing errors".  
 
From the MQA site, "With MQA, we go all the way back to the original master recording and capture the missing timing detail. We then use advanced digital processing to deliver it in a form that’s small enough to download or stream. The result is astonishing. Every nuance and subtlety of the artist’s performance – every tiny drop of emotion – is authentically reproduced. When you listen, you’ll be transported right into the very moment of creation. You’re there. MQA keeps all this timing information, and folds it into a file that's only a little bigger than a CD. Played back on any existing system, it will sound better than CD. But on an MQA-capable system, the music will be unfolded, reproducing every element that's in the original recording."
 
The implication is clear: capture the missing detail, better than the CD.  How you capture something that's missing is the problem. And that's any master, all the way back to the analog days, and of course redbook.  Yes, they are processing redbook masters.  For the vast bulk of the global library, that's all there is to work with.  

 

[old tech]

   I think I finally understand what you're arguing.
 
I doubt anyone in this thread is hanging their hat on the premise that MQA improves Redbook, and if so it would be a false premise, because MQA is not applied to native Redbook.  The labels that have signed up with MQA have only applied MQA to their HiRes masters.  While there are certainly some HiRes masters that are just upsampled crap (i.e. all of Madonna's stuff), a vast majority are native HiRes (depending on definition, there are 24/44 masters).  In most of those cases the Redbook version is transcoded from the same source (as is the case for all the 2L files).  There are no MQA files encoded from native Redbook, none.  It would make no sense to do so.  The patent doesn't even cover this scenario (it does include the possibility of encoding to a lower resolution than 24/48, but it's still from a HiRes master)
 
When people say they are enthusiastic for MQA, I suspect it's because they think it sounds good, better than the Redbook alternative that Tidal also has.  They are enjoying being able to stream a HiRes facsimile for the same price they pay to stream Redbook.
 
The results of the encoding process are the reproduction of a HiRes master.  If you enjoy that, then you are enjoying the results of the encoding.

But where did those 24/44 masters come from?  I'm pretty sure that Fleetwood Mac for example would have originally been a copy from the analog tape.  Analog tapes are not hi res, they are lower res than CDs.  So it is a false premise unless MQA can prove the music source was at least a 24/44 recording with no analog processing in the production chain.

 

[pinnahertz]

  But where did those 24/44 masters come from?

Up-sampling?

 I'm pretty sure that Fleetwood Mac for example would have originally been a copy from the analog tape.

If memory serves, and it may not, either Rumors or Tusk was done on an early digital system, though there wouldn't have been digital multitracks at that point yet. But yeah, analog, digitized...happened a lot. AAD, ADD, etc.

 Analog tapes are not hi res, they are lower res than CDs.

Correct.

So it is a false premise unless MQA can prove the music source was at least a 24/44 recording with no analog processing in the production chain.  

Well, not according to MQA. Their claims have not been definitively proven either way. Unlikely, but the jury hasn't even heard the evidence.
 
BTW, not sure how much was done in 24/44, more like 24/48 or 24/96 and of course 16/44.1. Yeah, the .1 to make 16 bit stereo format into an NTSC video field/frame. A lot of early stuff was 16/44.1 because sample rate conversion back then was kind of not too nice, adding 3dB at least. Even when we could record at 48 we didn't if it was headed to CD. And most of it was.

 

[old tech]

  BTW, not sure how much was done in 24/44, more like 24/48 or 24/96 and of course 16/44.1. Yeah, the .1 to make 16 bit stereo format into an NTSC video field/frame. A lot of early stuff was 16/44.1 because sample rate conversion back then was kind of not too nice, adding 3dB at least. Even when we could record at 48 we didn't if it was headed to CD. And most of it was.

Sorry I did mean 24/48 and you are right most recordings were done in 24/48.
 
A lot of early CDs that were sold as DDD were actually DAD as it wasn't later that digital consoles were used.
 
I don't know if this is correct, but I understand that Dire Straits Brothers in Arms was recorded in 16bit digital (and processed through an analog workstation, though classified as DDD).  It sure sounded nice though with a lot more dynamic breadth than most modern pop recordings.

 

[pinnahertz]

  Sorry I did mean 24/48 and you are right most recordings were done in 24/48.
 
A lot of early CDs that were sold as DDD were actually DAD as it wasn't later that digital consoles were used.
 
I don't know if this is correct, but I understand that Dire Straits Brothers in Arms was recorded in 16bit digital (and processed through an analog workstation, though classified as DDD).  It sure sounded nice though with a lot more dynamic breadth than most modern pop recordings.

Most most recordings were done in 16/44.1 for the first 20 years of digital audio.  
 
There were plenty of actual DDDs, mostly classical, as a good many of those were  mixed to stereo directly and recorded on 16 bit stereo digital systems.  I worked in that world.  My two CD projects,mid 1980s, were live mix to 2 track stereo, Sony PCM 1630, edited on a DAE-1100.  DDD all the way.  The pop stuff that needed multitrack would have had DASH machines after 1982, up to 48 tracks on 1/2" reel to reel (digital) tape.  Fully digital consoles were rare birds for the first decade, but got cheap fast after that.  Still, we worked in 16/44.1 to avoid a format bump for quite a while.  Film guys did 48kHz because they weren't going to CD. 
 
You know those early Soundstream tapes?  16/50!  

 

[old tech]

  Most most recordings were done in 16/44.1 for the first 20 years of digital audio.  
 
There were plenty of actual DDDs, mostly classical, as a good many of those were  mixed to stereo directly and recorded on 16 bit stereo digital systems.  I worked in that world.  My two CD projects,mid 1980s, were live mix to 2 track stereo, Sony PCM 1630, edited on a DAE-1100.  DDD all the way.  The pop stuff that needed multitrack would have had DASH machines after 1982, up to 48 tracks on 1/2" reel to reel (digital) tape.  Fully digital consoles were rare birds for the first decade, but got cheap fast after that.  Still, we worked in 16/44.1 to avoid a format bump for quite a while.  Film guys did 48kHz because they weren't going to CD. 
 
You know those early Soundstream tapes?  16/50!  

Thanks for that info.  Yeah I didn't click on the live mixes.  Now I remember that a 1987 Telarc CD in my collection (Andre Previn Holts the Planets) was a pure digital recording and phenomenal sounding too.
 
Those early days of digital must have been quite interesting with the technology moving so fast and the need to experiment and learn how best to use the new media.  The use of digital in music production goes back to the mid/late 70s when digital delay lines were used on LP cutting lathes, but it wasn't until CDs were released that the technology accelerated.
 
This is getting a bit off topic though... but the point remains, do all the recordings from this era and the analog era prior to it suddenly become hi res because the masters are MQA authenticated? It is more than a stretch to believe so. 

 

[gregorio]

  The relevant argument is whether MQA improves on Redbook, not 'Redbook or above'.  MQA is designed to be a HiRes streaming protocol.  MQA is ~60% the size of native 24/96.  If, as Archimago showed, MQA can provide an acoustically transparent version of 24/96 (or higher), then it's met its goal.  As I said above, you could provide a mathematically better (but not necesarrily acoustically better) format (i.e. 18/96) at similar bit rates, but at the moment noone is (and from a marketing standpoint, it's easy to understand why), so that's a strawman argument.  If you want to argue thet 24/96 is no better than Redbook, that's fine, but that is a completely different argument.

 
That's not a different argument, if you're going to play the "audibly transparent" card then you have to apply that card to everything! If MQA were audibly transparent compared to a 24/96 master and if a noise-shaped dithered 16/44 version of that same master is also audibly transparent to the original, then the answer to your "relevant argument" must be no, MQA does not improve on Redbook. Furthermore, if it can be shown that say AAC, a high bitrate MP3 or a FLAC of that same 24/96 master are also audibly transparent, then MQA is not an improvement over any of them either. If we're talking about accuracy rather than audibly transparent, then with the exception of FLAC, all of these formats/versions are effectively lossy compared to the 24/96 original, including MQA.
 
I get your point about "current" masters but how do you know that the non-MQA version being streamed by tidal (and used by many posters for comparison) is also that current master and not some other (non-current) master? Even if it is the same master, that's still only the first and most obvious variable, there's the slightly less obvious but just as important variable of biased hearing which also has to be removed BEFORE the differences/improvements that some are reporting can be attributed to anything the MQA codec is doing!
 
G

 

[AxelCloris]

I've gone through and cleaned up this thread a little. Please keep the posts from getting personal.

 

[headfry]

OK folks:
 
 
I've found a Tidal Masters track that sounds so musical and artistic
I want all of you interested to check it out:
 
 
 
John Coltrane   -     Giant Steps (Mono 2014 reMaster)
 
 
Title Track (1st one!)
 
 
 
Sounds so direct and expressive. Sounds to me that Coltrane
was out to make his trumpet portray some of the fullness of an orchestra.
 
 
...Success!
 
 
...really good!

 

[headfry]

One more amazing track from Tidal Masters:
 
 
 
Joni Mitchell    Hejira (2014 Tidal Master'ing)
 
 
 
track:   Coyote
 
 
 
Seriously. Just listen!
 
 
 
I think this is one of Joni's innovative masterpieces.....as good as anything
on Court and Spark. (to my tastes her highest artistic achievement.....great album!)

 

[old tech]

 
One more amazing track from Tidal Masters:
 
 
 
Joni Mitchell    Hejira (2014 Tidal Master'ing)
 
 
 
track:   Coyote
 
 
 
Seriously. Just listen!
 
 
 
I think this is one of Joni's innovative masterpieces.....as good as anything
on Court and Spark. (to my tastes her highest artistic achievement.....great album!)

Ok, I have listened to Coyote on Tidal through my Bluesound streamer.  Before I give my thoughts we'll get one thing straight. This is not high res music.  Hejira was recorded in 1976 using the analog tape and the analog equipment of the day, which in resolution is less than 16bits of digital. The only way it could be hi res was if the remaster involved recalling Joni back to the studio to re-record the album at least 24/96 - it did not happen and will never happen.  Having said that, nearly all of Joni Mitchell's recordings (as with John Coltrane, which also is not hi res for the same reason) are of very high quality.
 
Anyway, I agree that this is both great music and a great sounding stream.  I would be happy with that but if you are also seeking comparisons, unfortunately I only have this album on LP which is stored away at the moment.  However, given I was listening through the streamer, I thought I'd compare it with a random track (Down to You) from my DCC Court and Spark CD, the one in the link below.
 
Now this is not a fair comparison as it is different tracks on different albums and most importantly, different mastering engineers.  Comparing them side by side, Down to You has overall better sound quality.  What I mean by better is smoother, more dynamic and less pumped up and compressed like MQA's Coyote.  To the uninitiated the Coyote track can sound better initially as it is louder with more tizz but after a few minutes Down to You proves to be a more relaxed listen.
 
http://avaxhome.unblocker.xyz/music/joni_mitchell_court_spark_dcc.html
 
How about finding some true hi res music for us to comment?

 

[canthearyou]

Kinda reminds me of noise reduction that is used in some recordings.  Some people like it, saying "wow it has eliminated all the tape hiss" but any serious audiophile avoids it like a plague as it affects the depth and "good" parts of the music.

So, tape hiss is the good part of music?

 

[winders]

So, tape hiss is the good part of music?

No. But there is sound you want to hear at the same frequencies as the tape hiss. So using some kind of EQ to reduce tape hiss also reduces the sound you want to hear at those frequencies.
 
There were Dolby B encoded tapes to reduce tape hiss. How? The frequencies where tape hiss were prevalent were boosted on the recording. When played back, those frequencies were reduced to a normal level and that lowered the tape hiss as well.
 
There was also dbx......