MQA

[mansr]

The concern wrt material that has been though MQA and then subsequently altered, losing its official 'flagging' is that the effects of MQA will vary from case to case. So any detector that relies on detecting the 'usual signs' can't be 100% reliable. What I'm not yet sure about is how reliable it can be, and what the best tack may be. The obvious targets are:

1) The anharmonics adde by MQA passage.

2) The 'hole' effect around the folded Nyquist.

Thus far my feeling is that (1) could be detected by doing FFTs, and comparing them with their 'frequency flipped' results, correlating against an offset. That may make finding patterns of offset anharmonics easier as it can maximise the SNR of the search. May also be useful to differentiate the sample series first to 'pre whiten' a bit. But I'd need to play with this, so takes time. Hence if anyone else also wants to have a go, I'd be happy to be beaten to it!

One obvious target is the shaped dither that is always present in MQA files. The graph below was made by computing a series of short (2048 points) FFTs over an MQA file, computing a 100-point moving average of these, and finally taking the minimum of those.

This noise shape is a tell-tale sign of MQA and easily identified visually. Now to automate it...

 

[Rosewind]

Well, in the case of streamed or ripped music, wouldn't it be possible to include a test feature in any given music player program that at a click and, say, a 15 second scan or playback of the file, would generate this type of graph so that it could be compared to the profile of a non-MQA file?

 

[Alexh]

It does make me wonder how many people have heard what a plain old CD player can do. I have not gone into streaming yet and I am not sure if I will bother with all the CD`s that are still available.
It is because I find CD is able to produce a really good sound. A good transport and Dac will give good depth, width and no digital harshness at all.
There will still be the odd recording that sound like it was a late Friday evening job ... but I dare say that will apply to all formats.
But on the whole, I find CD very enjoyable.

 

[Jim Audiomisc]

One obvious target is the shaped dither that is always present in MQA files. The graph below was made by computing a series of short (2048 points) FFTs over an MQA file, computing a 100-point moving average of these, and finally taking the minimum of those.

This noise shape is a tell-tale sign of MQA and easily identified visually. Now to automate it...

Thanks for the and the previous posting. I'm wondering if the nominal noise shaping profile also may include the typical level of anharmonic fold-back. Not sure of the extent to which going for the minima removes that. I've in the past tended to go for 16k FFTs, but that is for very different purposes. Still, no harm in having more than one approach to try...

 

[Jim Audiomisc]

Well, in the case of streamed or ripped music, wouldn't it be possible to include a test feature in any given music player program that at a click and, say, a 15 second scan or playback of the file, would generate this type of graph so that it could be compared to the profile of a non-MQA file?

It will depend to some extent on how 'different' the MQA 'signature' is from the typical range of non-MQA examples, and on how varied each type may be from case to case. But if the shape is distinctive you can use a pattern pair like the type mansr has produced and take an overall difference to see the level of confidence. I'm glad he's a better programmer than me, though. Helps avoid the risk that people will die of old age before I got anywhere.

 

[mansr]

The MQA decoder relies on the dither noise being generated using a specific PRNG and filter. There can be no variation, or decoding would fail.

 

[Craig B]

Possibly. And she may also be an excellent presenter, etc. I have no idea as I've not seen her work. However given that the info I saw said her education was in 'media and english' I am unsure how deep her knowledge may be of topics like the application of Information Theory to digitial audio. Either in principle, or via analysis / development / testing. The BBC 'Click' programme is largely about 'tech toys' and on that 'wow' level. Its not really a science or engineering programme, but is presented as being about 'tech'. So I'm curious about the level/type of questioning she may bring to the event that WotHiFi arranged.

Dammit, Jim, I do need to get out more!

 

[misterdog]

Streaming gives me surprisingly variable results, some music sounds poor whereas other music is very good sounding.

Not all 1's and 0's are created equal.

By the way, for what it's worth (or was), CD's from different pressing plants can sound different, or could at any rate in the days when jitter wasn't well understood. Last I heard, which was decades ago, Sony was aware of and studying the problem, which they traced to variations in pit geometry.

From ASR.

 

[Rosewind]

Craig. I found a video where she encounters a paint ball in slo-mo in a bathing suit. I dared not post it in case I was deemed to be a dirty old misogynist.

 

[leroyd]

So far as I've been able to tell the video of the WotHiFi 'interview' (?) is only available if you 'register' with WHF. If they object to YT's data hoovering/sharing, why not simply make it openly available, I wonder? If they think it is good I'd expect them to feel it would attract new subscribers for the mag.

Here it is

 

[Tony L]

Nice to see the word ‘sponsored’ used. It makes it clear there will be no scrutiny or analysis.

 

[Zombie]

I really wish Tidal well. They had no idea what MQA really was.

They have a very good catalogue with new music, which might not appeal to middle-aged anoraks.
Btw, do you have a genre for Danish music on Tidal and is it updated? (Not videos)

 

[Rosewind]

"for the benefit of the artists" (c. 6:24)

Zombie, I was not on Tidal for very long, certainly not after I realised that master quality was not hi res flac but lossy MQA files posing as lossless hi res. So I did not check for much Danish music while I was there.

I'll put on my anorak now to take a stroll in my Danish garden.

 

[DimitryZ]

The problem arises in what you take for granted. Consider the following:

For the sake of example I'll use buying files you download, but the same argument applies to streaming.

Someone sees that one of their favourite bands - Bob Zit and the Blacktops - has released a new song. So they get it from their supplier in two versions being offerred. File 'L' is a now-traditional 192k/24 LPCM file. File 'M' is an MQA file.

They have a setup that can play both types of file, and fully 'decode' MQA.

They may decide that one version sound 'better' than the other. Doesn't matter here, which one.

The problem arises if they ask themself: Was 'L' the data fed into the MQA encoder to produce 'M'? If it was, I can decide if I prefer to choose MQA, or I'd prefer to avoid it.

Snag is that AWUI MQA insist that people should NOT be allowed access to what went into the MQA encoder. So if the supplier adheres to this, 'L' *can't* be what was fed into the MQA encoder. That means L and M have differences that are not due to the actual MQA encoding-decoding cycle. So we can't fully assess any effects of MQA on LPCM. And any decision about 'liking' one version or the other may not be due to MQA itself but this insistence on preventing us from having what was fed into MQA meaning that L is different to that possible causing us form a preference.

In effect, we might have preferred what went into MQA encoding if we'd been given THAT instead of L.

We now have the reports that when people fetch 'L' type versions of material they actually get M. Which also tends to cloak what is happening and get in the way of a comparison deciding what changes the MQA encode-decode may make and how audible - for good or bad - they might be. Thus we lose the reference point for making choices.

I think you really overcomplicating this.

If you are uncertain where the system ends, just draw a larger box. If MQA cabal holds a special ritual over the master that subtly changes it, then this is part of MQA and the end result either speaks for itself or it doesn't.

In practice, 2L files are supposed to be those things - carefully handled hires files that are only different through a pass into an MQA encoder.

Of course, if you never trust anything or anyone, this may also be unsatisfactory, as 2L may also be lying.

 

[DimitryZ]

I would assume close to zero, possibly negative.

I wouldn't be so sure. You, for example, have the proper education but you often speak utter nonsense.

 

[Craig B]

Craig. I found a video where she encounters a paint ball in slo-mo in a bathing suit. I dared not post it in case I was deemed to be a dirty old misogynist.

Point taken, Peter.

 

[DimitryZ]

It will depend to some extent on how 'different' the MQA 'signature' is from the typical range of non-MQA examples, and on how varied each type may be from case to case. But if the shape is distinctive you can use a pattern pair like the type mansr has produced and take an overall difference to see the level of confidence. I'm glad he's a better programmer than me, though. Helps avoid the risk that people will die of old age before I got anywhere.

It's easy to hear, if you know what to listen for.

 

[Alexh]

I haven't heard MQA but I do follow the new technologies that that emerge. I came across this video that my be of interest:

 

[Tony L]

Goto 1

 

[DimitryZ]

If you want rid, look no further. As above, I reside in Portugal. RME DAC is on my radar when the time is right, would be interesting to hear a topping DAC.

Excellent, PM me your address.