Tidal chucking MQA?

Status
Not open for further replies.
Jump to Latest
#162
Bruno Putzeys posted this on Facebook in October 2017...
This isn't a prelude to suddenly becoming active on FB but I felt I had to share this.
Yesterday there was an AES session on mastering for high resolution (whatever that is) whose highlight was a talk about the state of the loudness war, why we're still fighting it and what the final arrival of on-by-default loudness normalisation on streaming services means for mastering. It also contained a two-pronged campaign piece for MQA. During it, every classical misconception and canard about digital audio was trotted out in an amazingly short time. Interaural timing resolution, check. Pictures showing staircase waveforms, check. That old chestnut about the ear beating the Fourier uncertainty (the acoustical equivalent of saying that human observers are able to beat Heisenberg's uncertainty principle), right there.
At the end of the talk I got up to ask a scathing question and spectacularly fumbled my attack*. So for those who were wondering what I was on about, here goes. A filtering operation is a convolution of two waveforms. One is the impulse response of the filter (aka the "kernel"), the other is the signal.
A word that high res proponents of any stripe love is "blurring". The convolution point of view shows that as the "kernel" blurs the signal, so the signal blurs the kernel. As Stuart's spectral plots showed, an audio signal is a much smoother waveform than the kernel so in reality guess who's really blurring whom. And if there's no spectral energy left above the noise floor at the frequency where the filter has ring tails, the ring tails are below the noise floor too.
A second question, which I didn't even get to ask, was about the impulse response of MQA's decimation and upsampling chain as it is shown in the slide presentation. MQA's take on those filters famously allows for aliasing, so how does one even define "the" impulse response of that signal chain when its actual shape depends on when exactly it happens relative to the sampling clock (it's not time invariant). I mentioned this to my friend Bob Katz who countered "but what if there isn't any aliasing" (meaning what if no signal is present in the region that folds down). Well yes, that's the saving grace. The signal filters the kernel rather than vice versa and the shape of the transition band doesn't matter if it is in a region where there is no signal.
These folk are trying to have their cake and eat it. Either aliasing doesn't matter because there is no signal in the transition band and then the precise shape of the transition band doesn't matter either (ie the ring tails have no conceivable manifestation) or the absence of ring tails is critical because there is signal in that region and then the aliasing will result in audible components that fly in the face of MQA's transparency claims.
Doesn't that just sound like the arguments DSD folks used to make? The requirement for 100kHz bandwidth was made based on the assumption that content above 20k had an audible impact whereas the supersonic noise was excused on the grounds that it wasn't audible. What gives?
Meanwhile I'm happy to do speakers. You wouldn't believe how much impact speakers have on replay fidelity.
________
* Oh hang on, actually I started by asking if besides speculations about neuroscience and physics they had actual controlled listening trials to back their story up. Bob Stuart replied that all listening tests so far were working experiences with engineers in their studios but that no scientific listening tests have been done so far. That doesn't surprise any of us cynics but it is an astonishing admission from the man himself. Mhm, I can just see the headlines. "No Scientific Tests Were Done, Says MQA Founder".
Click to expand...
 
  • Like
Reactions: lrisbo and Turbowatch2
#164
What I perhaps meant was that there is no need for any newer things as the one we have is in any sense "perfect". I'm sure it can be done "differently", perhaps more efficient, but not better. I did not follow the links i couldn't find a reason to use electric energy to do so.

//
 
#168
Shanon and Nyquist proved mathematically that any bandwidth limited signal can be perfectly reconstructed from a set of discrete-time samples as long as the sampling rate is at least twice the bandwidth of the signal.

Let's let that sink in for a moment.

Proved mathematically. That's a pretty darn high bar. Probably the highest. I certainly give that a lot more credibility than experimental data. And by "experimental data" I mean properly executed scientific experiments. Which I give a lot more credibility to than someone's individual experience or the sighted, uncontrolled trials commonly performed by audiophiles (and many others).

The newer sampling theorems apply to a subset of signals. For example, compressed sensing requires that the signal satisfies the sparcity and incoherence criteria required for compressed sensing to work. According to Wikipedia, compressed sensing works well for MRI images. A "sparse" signal is a signal with lots of zero samples. I somehow doubt audio is an example of such a signal, but hey... Maybe you like to listen to silence. I'm not here to judge. 🙂

Tom
 
  • Like
Reactions: kevinkr
#169
The signal itself doesn't have to be sparse, rather it has to be representable sparsely in some domain. The trick is finding the right representational domain. For example, x-ray images show tissue density but they may be represented sparsely in the gradient domain. That's just a way of saying that x-ray images don't have sharp edges in them. X-rays are kind of blurry due to scatter in tissue. That means some solutions to the reconstruction problem are not likely or not possible. Thus it is possible to discard planar ray images of a CT scanner that are badly affected by scatter from dental fillings in the head of a patient. With the problem rays having been discarded, the remaining CT planar ray images are now undersampled. However, if they are transformed into the gradient domain representation they can be reconstructed into a 3D image using sparse sensing techniques. The resulting reconstructed CT images look almost perfect without any streaking artifacts from metal fillings. This was from some work done by Stanford University Radiation Oncology Dept.

Also, sparse signals don't necessarily have to have a lot of zero samples. As a practical matter it is often sufficient to have equation term coefficients that are near zero. They may then be set to zero so that the system of equations can be solved.

Regarding the incoherence criteria, it is conceptually similar to the requirement for solving a system of simultaneous equations. The equations can't be equivalent to each other; there have to be enough non-equivalent equations as there are unknown variables to be solved for.
 
Last edited:
#174
Markw4 said:
There is no "proof' in science, as I think you know. There is such a thing as anecdotal evidence which is a lot what we have when dealing with humans. A lot of drug side effects in the USP monographs are based on patient reports. Sometimes such information is the best we have. We just have to deal with it. Not every different capacitor design is going to get formal, publication quality listening tests. Nobody has the resources to do such things on a massive scale. Doesn't mean its all imaginary.
Click to expand...
Drug approvals very large sample sizes and the usual placebo tests. Golden ears give reasons why they can't provide large samples
Markw4 said:
BS. I know people with a remarkable ability to debias themselves in sighted listening.
Click to expand...
Have they been tested to prove this or do you give out medals to each other when you meet?
Markw4 said:
Moreover, ThorstenL is back and active in the forum in Marcel's RTZ FIRDAC thread. He has described something that I completely agree with, which is to the effect that if multiple people independently report hearing the same thing, there is a good chance there is something to it.
Click to expand...
Lots of people hear voices in their heads. They tend to get sectioned if they act on it. Are the voices real if enough people hear them?
Markw4 said:
Also, a lot of what people have claimed to have heard has turned out to be real, its the engineers who were in denial. Such things include Bybees, purple felt pens on CD edges, Schumann frequency, etc.
Click to expand...
none of which have been proven beyond some blokes having lols and claiming stuff.
Markw4 said:
Sometimes we know and other times we don't. There is a very good vinyl setup here to compare with CD.
Click to expand...
vinyl is for enjoyment of the experience, not for fidelity. Anyway I have 2 very good vinyl setups provided you are not being snobbish and saying that the cost of the setup is what matters not the performance...
Markw4 said:
If you take the recording to my house and all of a sudden it sounds far more real, you can get a pretty good idea of where the problem most likely is.
Click to expand...
2 channel audio can never sound real as it cannot recreate a full soundfield. It can however provide good entertainment, invoke emotional responses, get the wife in the right mood etc. Just not real.
 
  • Like
Reactions: bwaslo
#175
billshurv said:
Golden ears give reasons why they can't provide large samples
Click to expand...
If you say so; I don't know any golden ears.

Also, I don't have a drug company budget nor do I know anyone who does. Would like to though.

billshurv said:
Lots of people hear voices in their heads.
Click to expand...

I can't tell you how offensive I find that statement. Schizophrenia is a dead serious life changing illness. Nothing to make light of.
billshurv said:
...vinyl is for enjoyment of the experience, not for fidelity.
Click to expand...
Yean, but if it is from master tapes, and if its optical cartridge playback with its exceptional transparency, then you might be surprised. You can get a pretty good idea of how the master tape sounds. You can also plainly hear if the master was recorded to digital, such as with old Telarc recordings.
 
#178
tomchr said:
That's entirely possible.

On the flip side: Have you seen any hard evidence from MQA that their system works and does not degrade the original 16-bit data? I haven't. MQA is the one making the claims out audiophile superiority. They should be the ones backing up their claims with measurements.

Tom
Click to expand...
That is a kind of whataboutism .

It wasn't a flipside to me pointing out that GoldenSound appears to be one single person's opinion and yet is quoted as the "proof" something is a lie ..

If you listen to their videos and reasoning, they haven't performed the most obvious scientific requirement - to eliminate the effect of external factors on the thing they are testing, in this case Tidal.

All they have done is examine how Tidal works.

And no, it was not up to MQA to prove to the consumer that something works :

A) it's a product for entertainment : there is no moral duty here, it is not important. You don't have to use it , it doesn't kill people if it doesn't work .

B) their clients were major labels, streaming services, equipment manufacturers who pay licencing fees. It was they who had to feel convinced it made a difference and there are many engineers working with labels etc who felt that it did exactly what it intended on their systems, to their ears .

C) it was the incentive of streaming services, equipment manufacturers and labels to convince the consumer that it made a positive difference in order to make sales.

Personally I think it was a mistake for the Meridian guy to get so personally involved in discussions about it - I think he made some major marketing mistakes in thia modern social media environment by doing so.

This led to entrenched positions whereby even the claims made were distorted and the purpose of MQA was being distorted.

The reason for MQA was, in a nutshell, that timing, impulse accuracy is more important for human perception than frequency response accuracy and therefore the former should be the focus if audiophiles want better realism... and I personally guess that would first require people to drop the idea that frequency response and noise measurements are the be all and end all of everything.

So when you ask for measurements to prove MQA works , what exactly are you focusing on and what measurements?

How do you actually show that something is genuinely perceived to sound better when the whole purpose of the tech is psychoacoustics ?

Certainly, if your replay system isn't focused on temperal coherence .. then subtle attempts to improve that aspect in the source are not going to recreated by speakers that can't .. therefore one could argue it's unfortunately not going to be relevant to most consumers whether it works or not!
 
Last edited:
#179
acko said:
Isn’t CD quality considered lossy to starting with, no?
Click to expand...
"Lossy" has a technical definition and is about data loss.. so no. CD has all all the 16bit 44.1Khz encoded data present.

If you mean that data is lost if the original 24 bit 48kKhz file (probably the most common format to record in even today) when encoding a CD - yes it is lost but that is not what "lossy" means as a technical term .

The problems come when people want ot express themselves with normal, everyday language ... but use words their audience assume to mean, or shift meaning to, a familiar technical term.

Unfortunately, you can't really talk about data loss or information loss in a process to an audience that strongly relates the term 'lossy' to data formats .. You cannot, unfortunately , get most people to separate their own habitual language from the language of the person who is talking.

You have to play games, work around it, perhaps treat others as being inflexible in their thinking .. a shame, but that is majority human nature, people in general are not flexible in thinking.


The MQA guy appeared to often use different meanings of 'lossy' to speak to a non-technical consumer audience, only for the technical audience to slam it's use in the context of MQA.

Meridian have long used folddown techniques to remove signal information that the brain is less able to hear, in order to allow reproduction of signal that is more audible psychoacoustically. If you look at that from a data perspective .. the data is irrevocably changed, therefore cannot be recreates intact and is therefore lost ... even if the human ear couldn't hear it . Therefore their folddown techniques have always been lossy. It is inherent in the process. They know that, i know that .. therefore they could never have been using 'lossless' in a true data sense.

A marketing mistake to use it at all, ever ... when they meant something else by it.
 
Last edited:
Status
Not open for further replies.
Top Bottom