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

[RichB]

I don't want this thread closed. It sends a bad message.
Deal with infractions as you wish but the discussion should not be silenced.

If closed, I propose a MQA thread with measurements that show MQA anomalies.
Proponents of MQA should be welcome to present their measurements and provide their insights.

To some extent, I think that well regarded proponents feel set upon even when the discussion is technical.
IMO, this is insufficient reason to close this thread.

If the proposed thread is created, I encourage supporters that do not want to engage, stay away.
Fair warning: if you engage, you are engaged.

- Rich

 

[RichB]

Just read the quote I shared. The rest is obviously non-sense. I don't care who he is or about the DAC...



You're focusing on something completely irrelevant.

The quote I shared about him talking to MQA guys gives a little insight... ignore the rest of the webpage. It was easy for me to.

I am weird that way, I don't generally take anything away from articles that, as you say, contain non-sense.
The MQA stuff was boiler plate regurgitation of MQA PR.

- Rich

 

[HoweSound]

The infinite improbability drive is obsolete. Behold the infinite impossibility drive.

In which case MQA should be using 42 bits - the answer to "life, the universe and everything".

 

[KeenObserver]

MQA Thread, "I'll be back"

LOL
There are so many people asking so many questions that go unanswered that it is going to be like Wack-a-Mole.

 

[AdamG]

Notice: I am issuing a temporary reprieve until I can discuss options forward with the Mod/Admin Team. I’m really trying to accommodate what the ASR community wants. It’s a constant balancing act. More will be explained later today.

 

[RichB]

Notice: I am issuing a temporary reprieve until I can discuss options forward with the Mod/Admin Team. I’m really trying to accommodate what the ASR community wants. It’s a constant balancing act. More will be explained later today.

Thank you!

It will likely wind down on its own anyway.
Most supporter probably realize there is no joy to be had here.

- Rich

 

[AdamG]

@GoldenOne please check your PM’s.

 

[Pluto]

Is it not possible to create and store 0 samples immediately followed by 1's at some amplitude?

You can indeed store a mathematically perfect square wave in a PCM environment; you have a set of samples at +n, followed by a set of equal length at -n, and on and on…

The real devil in the detail is when the signal gets passed through the reconstruction filter.

 

[mieswall]

Your analogy establishes the value of MQA only if the letters are indeed useless - which they are not; and only if what the letters are being replaced with is "meaningful for our process" - which it is not.

So you've actually captured perfectly why and how MQA is simultaneously unnecessary and harmful: it does something harmful to the musical data in order to insert other data that is unnecessary.

I can't figure out how this incredibly obvious thing cannot be understood; I guess you just don't WANT to understand it.

Do you want me to use apples and pears instead? Letters in this quick example are the noise. The noise is irrelevant. Some people in the team of MQA has dedicated almost their whole life to research exactly this point, I guess few people in the world is more knowledgeable on the subject. If you bother to read any paper about MQA you would see up to which incredible extents they analyze and describe this issue (and, btw, you would understand also the absurdity of the 13 bits complaints, because you would have at least a clue of what noise shaping and dithering is about).

Trying to be "lossless" about noise is as stupid as it can be. More than that, if about testing is this discussion: the first axiom of MQA is that it will replace that noise for "pseudo noise" with true information hidden in it (you don't even have to read a thing, it is just obvious looking for 10 seconds one simple graph). Then if your test start to evaluate the match of the result with the original noise, your test is flawed from square one: those useless rotten pears are now very tasty apples, because the system was defined to do it that way. Everything else that follows is invalid.

But after 1800 comments people keep saying MQA is garbage... because that useless noise is not the same anymore... This would be understood even by a K-12 kid if not as biased as almost everybody seems to be in this thread.

 

[GoldenOne]

@GoldenOne, all: For my benefit, what exactly does a single point (in dBFS) on the Delta Spectrogram mean? How exactly does it calculate values for different frequencies (from the original time-domain samples)? Are the timelines guaranteed to be aligned between the two files?

(If there a description somewhere online, just point me at it. )

The scale can be seen on the righthand side. The resolution of each point is just dependent on the FFT bin size i'd set it to which i've just got at 65536.
The timelines are guaranteed to be aligned yes. Deltawave checks for and can compensate for clock drift and sub-sample alignment.
Though that is more for when comparing two analog recordings. For digital versions of the same file it is typically safe to assume that there won't be any clock drift. But still good to check.

Yes, the transition between 0 and 1 when decoded must be not a square wave.
Therefore, you did not embed square waves in your file. Correct?

- Rich

It is not possible to embed a true square wave in PCM no. But you wouldn't expect a true square wave output. A square wave test is effectively just another way of looking at impulse response.
This is what the square wave looks like in terms of the PCM samples:

Typically for a short linear phase filter you'd expect to see this as the outcome:

For minimum phase something like this:

And then for a longer filter, this one being a 1 million tap linear phase filter for example it'd look like this (zoomed out a bit for better view of ringing):

But, with MQA, we see this:

So no, it's not that there was a genuine square wave encoded in the PCM file (because that's impossible), but we weren't expecting to get a square wave output. The point is that it shows that there is weird nonlinear ringing characteristics, an odd 'notch' as such to the transition, and all sorts of odd quirks.

For the file in which the square wave PCM info is contained though, higher sample rates will result in less filter ringing and faster transition time.
(Image taken from AtomicBob's Dangerous Convert 2 measurements).

For a NOS dac, in theory they should be able to play the true square wave from PCM info. Limited only by the DAC's own bandwidth/slew rate.

They are panicking and trying to alter evidence..

I don't know, it's getting a little too conspiracy-esque to say that
But it is odd...

 

[RichB]

You can indeed store a mathematically perfect square wave in a PCM environment; you have a set of samples at +n, followed by a set of equal length at -n, and on and on…

The real devil in the detail is when the signal gets passed through the reconstruction filter.

Thanks, I think that is what @GoldenOne has done.
Clearly, this is not a representative of music but DACs, for example, must be tested with such files to make sure they are handled without incident.
Is this reasonable?

- Rich

 

[Zensō]

So give me an example of any music you purchased yourself that came in MQA instead of CD. Not "here is a CD in Japan." But any music you personally purchased. Is there anyone in this thread with such an experience?

It has been years since these dire predictions been made. Where is the exitance proof?

Millions of WMG MQA titles on Tidal?

LOL
There are so many people asking so many questions that go unanswered that it is going to be like Wack-a-Mole.

Yup. Until the legitimate questions are answered, or MQA goes away, this will continue to be discussed here, or elsewhere.

 

[GoldenOne]

Thanks, I think that is what @GoldenOne has done.
Clearly, this is not a representative of music but DACs, for example, must be tested with such files to make sure they are handled without incident.
Is this reasonable?

- Rich

It's not intended to be representative of music no. It can be used to test a variety of things including how resampling/oversampling/reconstruction filters respond to these types of signals (same as impulse response tests and white noise). As well as to test things like DAC slew rate/bandwidth (this would need a high sample rate though as 44.1khz wouldn't be sufficient)

 

[voodooless]

Is it not possible to create and store 0 samples immediately followed by 1's at some amplitude?
I am not an expert but I thought this as possible.

I thought there were complaints about @GoldenOne doing this in his sample. How did he do that if it were not possible.
Am I missing something?

- Rich

Yes, you can do that, but it will not encode a square wave, only an approximation on one within the bandwidth that the sampling frequency delivers. These signals are often used to look at the filter behavior, because they show pre- and post ringing quite well.

 

[FireLion]

From someone that watched a few videos on MQA it appears its main focus was retaining the quality but using lower bandwidth. Well, internet has got faster right?

Can someone point me to an MQA test where they show the advantages over FLAC or are we just to take it on the faith of golden ear engineers who have history in the industry? If it's just another file format and can do origami then I understand the extra cost.

Here is my science! Yes that is an MQA origami pink panther, look how he is after that folding process, lots of noise.

Here is the FLAC version, no folding.

 

[RichB]

It's not intended to be representative of music no. It can be used to test a variety of things including how resampling/oversampling/reconstruction filters respond to these types of signals (same as impulse response tests and white noise). As well as to test things like DAC slew rate/bandwidth (this would need a high sample rate though as 44.1khz wouldn't be sufficient)

Excellent thanks to voodooless as well.
I picked the very first DAC I found in Stereophile:
exaSound s88 multichannel D/A processor Measurements | Stereophile.com

Containing an impulse a single response:

This is used to test the reconstruction filter.
Since MQA forces a reconstruction filter, then it seems to me there is nothing invalid about testing it.

What @GoldenOne has done then, it is found a way to test MQA and is not materially deferent from Stereophile, except that it is unsanctioned.
Is this correct?

If so, is complaints about the inclusion of digital representations (approximations) of square waves in the MQA sample valid?

- Rich

 

[RichB]

From someone that watched a few videos on MQA it appears its main focus was retaining the quality but using lower bandwidth. Well, internet has got faster right?

Can someone point me to an MQA test where they show the advantages over FLAC or are we just to take it on the faith of golden ear engineers who have history in the industry? If it's just another file format and can do origami then I understand the extra cost.

Here is my science! Yes that is an MQA origami pink panther, look how he is after that folding process, lots of noise.
View attachment 127035


Here is the FLAC version, no folding.

View attachment 127036

Come on folks, it's funny right?
If you don't think so, then you might need a mirth booster shot

There have been some great laughs on this thread.

- Rich

 

[voodooless]

If so, is complaints about the inclusion of digital representations (approximations) of square waves in the MQA sample valid?

That depends on what you would like to test.

 

[Cebolla]

So I really do not know what Tidal is doing at the moment.

After their recent announcement that they'd be moving MQA to a new tier, and offering an actual price tier for just hifi/lossless, I decided to check if the files being served had changed at all. Or if it was still just the same file with MQA flagging removed.

Firstly I checked Sam Smith's "Too Good at Goodbyes", and was pleasantly surprised to see that now there is no longer a bit-perfect match between the master and hifi tier. A 100dB null now.

Not sure why you expected that particular MQA track when streamed via the TIDAL HiFi quality (16bit/44.1kHz only resolution) connection to be bit-perfect with the original MQA track (as streamed via the Masters quality connection). The original MQA track (undecoded) is hi-res 24bit/44.1kHz (with an original MQA sample rate of 88.2kHz)!

The TIDAL HiFi quality version could still be sourced from the original (hi-res in this case) MQA track, though not bit-perfect but truncated to16-bits and very likely still recognised as MQA when played to a full decoding MQA DAC.

 

[RichB]

That depends on what you would like to test.

I suppose, but sometimes, you test and find things you did not expect.
I write software, so that is pretty much a daily occurrence.

- Rich