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

[amirm]

Apple's business practices have nothing to do with MQA. This is whataboutism.

No it is not. Whataboutism tries to shift topic to some other thing. Apple's conduct here is not some other thing. They have used their "format" (OS) to take away choice from you and to tax every developer for software distribution on their device unlike previous practice (on Windows, Linux, MacOS) to have none of that. This is what you claim with MQA, right? That you are not a user of high-res audio but fear that it would take over the world as far as baseline lossless audio as well. And charge everyone. Well, Apple has done that but you don't seem to want to go there. That tells me you have an emotional need to fight MQA and not any kind of principled reason to go after it.

Indeed, same thing is said about whataboutism as I explained above. From the wiki: https://en.wikipedia.org/wiki/Whataboutism

"Some commentators have defended the usage of whataboutism and tu quoque in certain contexts. Whataboutism can provide necessary context into whether or not a particular line of critique is relevant or fair. "

You fail the bolded section if you are OK with Apple, Blu-ray, Netflix, heck the entire scheme Tidal uses to for subscription business.

 

[q3cpma]

No it is not. Whataboutism tries to shift topic to some other thing. Apple's conduct here is not some other thing. They have used their "format" (OS) to take away choice from you and to tax every developer for software distribution on their device unlike previous practice (on Windows, Linux, MacOS) to have none of that. This is what you claim with MQA, right? That you are not a user of high-res audio but fear that it would take over the world as far as baseline lossless audio as well. And charge everyone. Well, Apple has done that but you don't seem to want to go there. That tells me you have an emotional need to fight MQA and not any kind of principled reason to go after it.

Indeed, same thing is said about whataboutism as I explained above. From the wiki: https://en.wikipedia.org/wiki/Whataboutism

"Some commentators have defended the usage of whataboutism and tu quoque in certain contexts. Whataboutism can provide necessary context into whether or not a particular line of critique is relevant or fair. "

You fail the bolded section if you are OK with Apple, Blu-ray, Netflix, heck the entire scheme Tidal uses to for subscription business.

But you're not forced to buy Apple nor are you unaware that you're entering the Apple ecosystem (whatever its problems) when buying Apple. MQA, on the other hand, isn't just another competing walled garden, it's trying sneakingly and insidiously to build its walls around the entire music industry and permeate it.

 

[KeithPhantom]

read the MQA papers. The process of dithering of data coming from upper folds is CLEARLY explained, even in annoyingly detailed terms. BTW: ultrasonics are not relevant because they are not listenable, instead because they are important in time coherence. According MQA and the dozens of papers of neuroscience they cite, as important as the audible band.

True, read about dithering and noise shaping. But you can do the same with regular FLAC. And that about neuroscience, not just cite your sources for this claim, also explain how are we able to even sense these air pressures at such high frequencies. Also explain the time coherence that mainly affects 20 kHz and up, where humans are pretty insensitive. For what I know, phase shifts due to filters only affect the area surrounding the corner frequency of the filter. Don’t forget to include the time dilation between each shifted frequency.

They replace that noise floor with the slice of relevant data captured from upper folds

What is “relevant”? How do you know this inclusion of data is even statistically significant? What do you gain from that extra data? Just want to know more.

real information": In this context, it is called "music

In frequency terms? What is “music”? If I’d ever consider valid test tones as music, why doesn’t MQA encode it correctly?

The remaining bits in Redbook are wasted space, from that MUSIC recording perspective. If you were registering other unknown profile data, your assumption that the whole space would need to be preserved would be right. MQA deals with music, not with the Perseverance photos sent from Mars.

Just dither. Dither a regular PCM file and you can fit more DR than you think you need in just 16 bits. I still don’t see what you mean by “music” and “information” in frequency terms, so I don’t really get your points.

They can't be lossless when unfolded: to be "lossless" in these Taliban terms, both noise floors must be identical, and by definition that noise floor doesn't exist anymore in MQA, as it was replaced with useful information.

Which noise floor doesn’t exist? All quantized files have a noise energy equivalent to the error between the sample and the continuous value. You can use techniques to shift the noise outside the passband or to average it, but not to make it disappear. Explain why MQA affect samples where reconstruction isn’t needed.

 

[symphara]

So say integrated. I was just referring to the word “amplifier”.

Why would I do that? It's much easier to ignore you. Take care.

 

[symphara]

MQA, on the other hand, isn't just another competing walled garden, it's trying sneakingly and insidiously to build its walls around the entire music industry and permeate it.

Hard to believe this claim, considering how small the target audience is, how small the streaming service promoting it is, and just how irrelevant the whole thing is. By all means, don't support it. I don't. In fact when I cancelled Tidal I told them in the comment box that I don't want MQA but some normal high-resolution file. PCM, ideally. But I don't think MQA is quite La Piovra.

 

[q3cpma]

Hard to believe this claim, considering how small the target audience is, how small the streaming service promoting it is, and just how irrelevant the whole thing is. By all means, don't support it. I don't. In fact when I cancelled Tidal I told them in the comment box that I don't want MQA but some normal high-resolution file. PCM, ideally. But I don't think MQA is quite La Piovra.

Keyword is "trying". And it kind of works when DAC makers start to feel obligated to pay for their racket.

 

[GoldenOne]

Hard to believe this claim, considering how small the target audience is, how small the streaming service promoting it is, and just how irrelevant the whole thing is. By all means, don't support it. I don't. In fact when I cancelled Tidal I told them in the comment box that I don't want MQA but some normal high-resolution file. PCM, ideally. But I don't think MQA is quite La Piovra.

I'd suggest reading mqa's financial reports. Both the numbers and their stated situation info.

They are currently losing millions per year and have been since creation.
Their business plan is to grow until MQA is ubiquitous in the audio industry. At which point they will be profitable both because they'll be everywhere, and also will be able to pretty much make any demands of manufacturers they like.

 

[abdo123]

They are currently losing millions per year and have been since creation.

how can a codec or a format lose millions since inception?

I mean what is there to maintain about intellectual property?

 

[Rottmannash]

It is really sad that we are having to second-guess whether these major music streaming services are truly bit-perfect and lossless. If Amazon HD is not bit-perfect for at least 16bit/44kHz content (still no empirical data), Canadians have literally no other options left now that Tidal has started compromising its Hi-Fi content. I spent the entire morning trying to use various VPN tools to register for Qobuz with no success. I got to the stage where it asked for payment info and then it refused both my credit card and PayPal.

First World problems, I know.

Not suggesting you do this but theoretically..... if one had a friend in the US who could establish a Qobuz account and pay the year in full (which is a savings vs month to month) said friend could "loan" you their login and password. Then your VPN strategy might work.

 

[symphara]

I'd suggest reading mqa's financial reports. Both the numbers and their stated situation info.

They are currently losing millions per year and have been since creation.
Their business plan is to grow until MQA is ubiquitous in the audio industry. At which point they will be profitable both because they'll be everywhere, and also will be able to pretty much make any demands of manufacturers they like.

Wait, so they're losing millions per year and their business plan is to grow? It sounds like they'll grow right into bankruptcy.

Forgive me, I think your dive into MQA is interesting and there's much to learn from this thread on the technical side, especially for a audio-technically-neophyte such as myself, but I don't buy the threat thing.

As a CD collector, this is depressing to me: https://www.statista.com/chart/12950/cd-sales-in-the-us/. And the thing that would worry me would be MQA-CD. I just don't think that the MQA-CD is going to happen, because the CD thing might not be happening anymore and I'll end up buying tracks off the net soon.

On the other hand, Tidal is too small a fish in the streaming sea for MQA to have an impact, and I don't see the others embracing MQA.

Now if Spotify were to announce they're moving to MQA, I'd grab my pitchfork. However I believe the chances of that are rather miniscule.

 

[mieswall]

True, read about dithering and noise shaping. But you can do the same with regular FLAC. And that about neuroscience, not just cite your sources for this claim, also explain how are we able to even sense these air pressures at such high frequencies. Also explain the time coherence that mainly affects 20 kHz and up, where humans are pretty insensitive. For what I know, phase shifts due to filters only affect the area surrounding the corner frequency of the filter. Don’t forget to include the time dilation between each shifted frequency.

What is “relevant”? How do you know this inclusion of data is even statistically significant? What do you gain from that extra data? Just want to know more.

In frequency terms? What is “music”? If I’d ever consider valid test tones as music, why doesn’t MQA encode it correctly?

Just dither. Dither a regular PCM file and you can fit more DR than you think you need in just 16 bits. I still don’t see what you mean by “music” and “information” in frequency terms, so I don’t really get your points.

Which noise floor doesn’t exist? All quantized files have a noise energy equivalent to the error between the sample and the continuous value. You can use techniques to shift the noise outside the passband or to average it, but not to make it disappear. Explain why MQA affect samples where reconstruction isn’t needed.

Sorry, my letter heading below is because my limitations, that alude to your paragraphs in same order:

A- Well, that's an interesting shit. Now we are discussing if capturing ultrasonic is important at all, not the "lossyness" detected in these tests. That's a step forward, and I would be very interested in the refutals. But I'm not able to talk in behalf on Meridian or MQA in this.

B- The same relevance as described in the audible band. Ultrasonic band also has noise floor that you don't need to recover; and the slice of data captured by ADC there, instead of all the noise and artifacts added by the capturing process; data that is even smaller, and so, much more compressible to "fold" it in previous bands (folds) of the origami process: the chunks of 0-48 / 48-96 (less data) / 96-192 (even less) / 192-352 (tiny) of a the DXD source. And, again, if we engage in if that information is statistically relevant we would have, at last!, moved the discussion from this nonsense of lossless to a more substantial point. Perhaps it is relevant, perhaps not: wouldn't it be more useful for us all to be diccussing about that instead of the desire of a lossnessy that is out of context in this subject?

C- "Dither a regular PCM and file and you get more DR". But the point is that Redbook doesn't do that. Meridian chooses one way to regain that wasted space. Of course creative audio engineers may find others. But, again, we are making is a step forward in the discussion: now we would not be questioning what MQA does, but instead if it does it the best way.

D- The noise floor of the incoming signal in the 0-24Khz region (and please remind that MQA is 24-bit depth, giving more headroom that even the best reproduction chain is able to resolve, among other considerations). It was just discarded and replaced like said in B above.

 

[samshaver]

This is amazing!!! Thanks for sharing!!!

 

[ebslo]

C- "real information": In this context, it is called "music": the ever smaller space occupied by music the further you rise in frequency, as captured in ADC. Most bits in lower frequencies, progressively less the higher in frequency you move. The remaining bits in Redbook are wasted space, from that MUSIC recording perspective.

You've said this (or similar) several times, but I can't reconcile what it sounds like you are saying with the fact that MQA files are playable by non-MQA decoders. Other sources say they put the extra information in low-order bits below the noise floor, but it sounds like you are saying they convert to frequency domain and store stuff in the high-order bits of the high-frequency bins which are statistically usually 0. But then it couldn't be played by non-MQA decoders and MQA CD's couldn't exist. What am I missing?

 

[GoldenOne]

how can a codec or a format lose millions since inception?

I mean what is there to maintain about intellectual property?

£490k turnover
£4.5 million in administrative expenses. £2.07 million of which were wages.


Bob Stuart took home nearly half a million

https://find-and-update.company-information.service.gov.uk/company/09123512/filing-history

 

[symphara]

PS: I bought speakers in January and none of the major dealers in an European capital had a CD player in the show room. It was astonishing to me. I was told it's a strictly import-to-order item, since very few people order them, they keep no stock. Worrying about MQA-CD, which would be the thing I'd care about, seems completely moot.

 

[AdamG]

Wait, so they're losing millions per year and their business plan is to grow? It sounds like they'll grow right into bankruptcy.

Forgive me, I think your dive into MQA is interesting and there's much to learn from this thread on the technical side, especially for a audio-technically-neophyte such as myself, but I don't buy the threat thing.

As a CD collector, this is depressing to me: https://www.statista.com/chart/12950/cd-sales-in-the-us/. And the thing that would worry me would be MQA-CD. I just don't think that the MQA-CD is going to happen, because the CD thing might not be happening anymore and I'll end up buying tracks off the net soon.

On the other hand, Tidal is too small a fish in the streaming sea for MQA to have an impact, and I don't see the others embracing MQA.

Now if Spotify were to announce they're moving to MQA, I'd grab my pitchfork. However I believe the chances of that are rather miniscule.

MQA-CD’s are already a thing: https://www.google.com/search?q=MQA...HXXPCFgQ_AUoAXoECAIQAQ&biw=1366&bih=882&dpr=2

 

[symphara]

MQA-CD’s are already a thing: https://www.google.com/search?q=MQA...HXXPCFgQ_AUoAXoECAIQAQ&biw=1366&bih=882&dpr=2

Some exist, yes, nobody's denying that. But it's absurd to worry about MQA-CD taking over a dying market. It's up there with the worry that Meridian-patented alloy horseshoes are taking over the horseshoe market.

 

[voodooless]

C- "Dither a regular PCM and file and you get more DR". But the point is that Redbook doesn't do that.

Redbook doesn’t need to do that, the mastering process does, and yes, this has been done to CD’s for decades.

 

[voodooless]

£4.5 million in administrative expenses.

Any specifications on what those are?

Also the turnover seems kind of low? If for instance a MQA license for a DAC would cost $5, that would only be 90k units, which seems low if you look at the amount of products available. And that is they would only get money from that. Tidal had what, 3 million payed users in 2019? Let’s say 20% pay for the MQA version, so let’s make it 500k users that listen to MQA. I cannot imagine that Tidal pays less than $1 per year for the license? Then there is the licenses for tools, encoders etc.. probably not cheap.

I rather think they spend loads of money to push the technology in some way or another.

 

[KeithPhantom]

Well, that's an interesting shit. Now we are discussing if capturing ultrasonic is important at all, not the "lossyness" detected in these tests. That a step forward, and I would be very interested in the refutals. But I'm not able to talk in behalf on Meridian, or MQA in this.

The test for losslessness is important, but it isn’t what I was looking for in my first point. I just want evidence that ultrasonics are important. A quick search in Google Scholar and my school’s library yielded opposite evidence to your claim.

The same relevance as described in the audible band.

I’m not accusing you of anything you just to make that clear. But I just want evidence supporting your claim. I was looking for it, and found counter evidence against your claim. Found something about ultrasonics and bone conduction, but these have to be at an unnatural level to even be sensed through a nonconventional method. The rest agreed with the null, the effective experimental limit of the human hearing is 20 kHz.

Dither a regular PCM and file and you get more DR". But the point is that Redbook doesn't do that and all modern DAWs allow it. Meridian chooses one way to regain that wasted space. Of course creative audio engineers may find others. But, again, we are making is a step forward in the discussion: now we should not be questioning that MQA does, but instead if it does it the best way.

Automatically, it doesn’t. But the option is there and there’s nothing in RedBook that prevents you from doing it. Also, list what are the effective cases of needing more dynamic range than the 93.2 dB effective in non-ideal or/and experimental conditions.

The noise floor of the incoming signal in the 0-24Khz region (and please remind that MQA is 24-bit depth, giving more headroom that even the best reproduction chain is available to resolve, among other considerations).

Not always 24-bit, since interpolations from 13 bits exist. Also again, PCM can do this and better since you can dither 24-bit to exceed the 144 dB limit. Again, when do you need this much headroom for home audio reproduction.