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

[Hayabusa]

GO's test was fatally flawed, since he never did his homework and learned how the system operates. The only thing he proved is that he is very bad at this stuff. Embarrassing, really.

Can you make the test?

 

[jensgk]

He is a "blogger." MQA provided the clear explanation of what was wrong with his test before he went public with his test. He ignored that and ran off with his testing and anti-MQA arguments he had read online, personalized the whole thing with MQA being mafia ready to go after him, etc. just like a blogger would do to sensationalize the topic. .............
I suggest backing off from this lest you want to be dragged into the same mud with him.

Your very emotional rant, and appeal to authority does not really change my stance.

A person can be very respected with academic work, and at the same time be a snake oil salesman. MQA marketing certainly smells a lot of snake oil.

I think @GoldenOne has done an excellent job exposing MQA and also very much Tidal. Was the analysis without faults? No, but I think most of the main points hold.

@amirm: I have a deep respect for you and your work, - just not in this case. Nobody is perfect.

A person with Reply ban power over you if you keep posting about your emotions than anything related to the topic.

@amirm It's your forum, you can do whatever you want, but things like that just don't look good.

 

[tmtomh]

Thank you very much! That gives me effectively the answer I was looking for, as you already explained that the MQA encoder output will be essentially the same size regardless of whether the input was 88kHz or something higher (unless I misinterpreted something).

Reposting my original question to @abdo123 below with full context, I was addressing his assertion that MQA files were smaller for equivalent perceptually relevant information, and you have answered the size part of that in the negative. The perceptual equivalence part is of course much more difficult to prove definitively as evidenced by the page count of this thread, but over the course of it I think there is enough to have reached an at least somewhat informed opinion.

Yes, thank you both. And to be clear, there is no such thing, in any meaningful sense, as an MQA file encoded from a source with a sample rate higher than 96kHz. Any file with a sample rate of 176.4k, 192k, 352.8k, 384k, and so on, will first be downsampled by the MQA encoder to either 88.2k or 96k (depending on which of those is an even divisor of the original PCM file's sample rate).

So to be clear, @ebslo is correct that the MQA encoder output will be essentially the same size regardless of whether the original file's sample rate was 352.8k, 176.4k, or 88.2k - because the MQA encoder is only folding 88.2k. Anything beyond that gets discarded entirely before the folding begins.

 

[amirm]

It will play.

Ah, you are right. I wasn't thinking.

 

[earlevel]

Just some tangential thoughts on storage, compression, and streaming while making lunch...

The truly important achievements in compression were those that were "enabling". At a time when streaming even CD-quality audio was prohibitive (few people had data service that would handle it), mp3 compression enabled an industry. Even for non-streaming it was huge—a costly hard-drive based iPod didn't hold many tunes uncompressed.

Compression just to have a smaller size is not enabling, but it does give consumers choice and has value.

Amir's point on the data savings for Tidal—yes, that translates into a substantial savings on the storage system. The negative is that it basically moves that cost to the consumer (for DAC, any additional cost or streaming service—they may get some back in data rates, but that would mostly be for mobile phone use, which seems to work against the idea of being able to hear the difference in hi-res audio. Also, I'm not sure if I understand correctly that maybe you can use software decoding with Tidal? possibly saving on the DAC bump for MQA?).

As noted this would be pitted against the lossless streamers, including Apple coming soon. One thing for certain is that compression is less and less important over time. Unlike video (SD/HD/5k/8k), audio needs don't change much, and the bandwidth requirements are modest relative to video, which also has a great need to be streamed—the same pipe that handles video give you audio for (relatively) free. This makes me doubt MQA will win out, above a niche for audiophiles who believe in ti, unless it can prove to enhance the listening experience. (And if it does, it will still suffer from pushback on why they don't just build the technique into players, and forget about the streaming service MQA.)

Just had a flashback...I remember a press release years ago, announcing Apple had bought a particular non-archival (meaning not slow) storage system in the size of 12 PB (for iTunes cloud). 12 PB—wow, that boggled my mind (just looked it up, 2011). In a twist of fate I changed industries about that time, and since a couple of year later have been supporting that specific product for the past 8 years. 12 PB seems pretty modest these days.

 

[tmtomh]

GO's test was fatally flawed, since he never did his homework and learned how the system operates. The only thing he proved is that he is very bad at this stuff. Embarrassing, really.

Your comment here, particularly the way you're clearly parroting others' claims but in a more generalized, nonspecific manner, illustrates only your own lack of understanding.

 

[Jimbob54]

Ah, you are right. I wasn't thinking.

I think that factor is one of the drivers for some of the fear. People like redbook 16/44.1. If undecoded mqa is "like" redbook but not it, even if perceptually lossless to redbook, its a perceived threat to what they understood to be "all you need"
If nothing else, mqa's clarity of message and signposting has been a failure in this regard.

 

[LeftCoastTim]

That is all it needs to show to make a case for not pre-filtering the content when it is distribute online. The CD is no more. I don't understand why we are stuck with its format as far as digital content.

I guess if CD didn't exist, and considering tradeoff between useful musical content, then I would propose 20 bits of dynamic range and 25KHz of bandwidth.

As for MQA, reducing the dynamic range by 2-3 db in the <20kHz range in exchange for expanded frequency response (and larger file size) doesn't seem worth the tradeoff to me. IMHO, extending Opus to handle up to 25kHz with some conservative threshold of hearing values >18kHz would be a far better trade off than what MQA is doing.

 

[amirm]

Your very emotional rant, and appeal to authority does not really change my stance.

A person can be very respected with academic work, and at the same time be a snake oil salesman. MQA marketing certainly smells a lot of snake oil.

There was no appeal to authority. I was asked why I was not putting OP on pedestal, I explained that he is not deserving of such. And if we want to show respect, Bob Stuart deserves it far more than OP. You can't be a part-time vegetarian here, constantly accusing of Bob of being snake oil while wanting to elevate who OP is.

Really, I am not interested in constantly hearing such rhetoric. It is not information. You still have a beef with something to do with MQA which we have not addressed, let's see it. Otherwise, let's leave the accusatory comments at the door.

MQA has provided a solution to high-res delivery and companies have brought it to market. In no way do they look like an expensive cable which does nothing whatsoever different than ordinary cable.

 

[DimitryZ]

On a parallel thread at PFM, JimAudiomisc, a storied engineer, professor and writer has turned his considerable analytic skills to MQA. This is a first installment of his work, focusing on MQA content played on non-MQA equipment - a topic of discussion here as well.

http://www.audiomisc.co.uk/MQA/investigated/MostlyQuiteHarmless.html

 

[amirm]

I guess if CD didn't exist, and considering tradeoff between useful musical content, then I would propose 20 bits of dynamic range and 25KHz of bandwidth.

Agreed.

As for MQA, reducing the dynamic range by 2-3 db in the <20kHz range in exchange for expanded frequency response (and larger file size) doesn't seem worth the tradeoff to me. IMHO, extending Opus to handle up to 25kHz with some conservative threshold of hearing values >18kHz would be a far better trade off than what MQA is doing.

??? MQA once decoded gives you more than 16 bits, not less. The less than 16 bits is for backward compatible baseline layer without decoding. This is an independent enhancement to higher sample rate.

 

[DimitryZ]

Can you make the test?

Nope, I am an aerospace engineer by education. But had he done his homework and placed his tones into the Shannon Triangle that MQA actually encodes, he would have had at least interesting results, instead of meaningless.

 

[LeftCoastTim]

MQA once decoded gives you more than 16 bits, not less. The less than 16 bits is for backward compatible baseline layer without decoding. This is an independent enhancement to higher sample rate.

I don't know that sounds like you are saying you can fit more bits than the available bandwidth with clever encoding. I might have to invoke Shannon there.

 

[tmtomh]

Agreed.


??? MQA once decoded gives you more than 16 bits, not less. The less than 16 bits is for backward compatible baseline layer without decoding. This is an independent enhancement to higher sample rate.

Every high-res format gives you more than 16 bits. The difference is that 24-bit PCM gives you 24 bits, while 24-bit MQA gives you more than 16 bits but a good deal less than 24. A 24/48 PCM FLAC file will have greater bit depth/lower noise floor than any MQA file, and will provide just as much high frequency extension in the audible range as any MQA file. Such a file also will be equal or smaller in size than an MQA file.

 

[Hayabusa]

Nope, I am an aerospace engineer by education. But had he done his homework and placed his tones into the Shannon Triangle that MQA actually encodes, he would have had at least interesting results, instead of meaningless.

ok, so there must be someone here that can do it on this forum. Maybe we should bundle some knowledgeble persons to create the ultimate test file that does test MQA in the correct way.
Anyone?

 

[KSTR]

ok, so there must be someone here that can do it on this forum. Maybe we should bundle some knowledgeble persons to create the ultimate test file that does test MQA in the correct way.
Anyone?

Best would be if we had access to an encoder directly, as well as a full-featured software decoder.
At any rate we still could meaningfully discuss what such a test file should contain and what the analysis would be, after defining the goals of the test. That's not going to be easy but it is not impossible.

 

[tmtomh]

ok, so there must be someone here that can do it on this forum. Maybe we should bundle some knowledgeble persons to create the ultimate test file that does test MQA in the correct way.
Anyone?

There are a lot of people here who can do it. The trouble is, they'd need access to the MQA encoder to do it.

 

[danadam]

Reposting my original question to @abdo123 below with full context, I was addressing his assertion that MQA files were smaller for equivalent perceptually relevant information, and you have answered the size part of that in the negative. The perceptual equivalence part is of course much more difficult to prove definitively as evidenced by the page count of this thread, but over the course of it I think there is enough to have reached an at least somewhat informed opinion.

It may interest you, that for the same file that Amir tested, if the original file (24 bit 88 kHz) is converted to 20 bit, keeping the sampling rate unchanged, you get essentially the same size as mqa:

45474K - 2L-145_01_stereo.mqa.flac
45195K - 2L-45_stereo_01_FLAC_88k_24b.converted.flac
71914K - 2L-45_stereo_01_FLAC_88k_24b.flac

 

[amirm]

This to me is THE PROBLEM. If Tidal had left up the original versions of the Warner Catalog that they replaced with MQA versions then I could ABX the two versions and decide for myself if I could hear any difference and if I could if I had any preference. The fact that the original versions were removed after the MQA versions were put up combined with MQA's refusal to publish "before and after MQA demonstration tracks", while "proving" nothing certainly does not pass an ignorant layman's "smell test".

Why not start with the 2L files? They have been around for quite a while and still are. They were also used in the study Stuart/Craven published on audibility of filters.

 

[amirm]

It may interest you, that for the same file that Amir tested, if the original file (24 bit 88 kHz) is converted to 20 bit, keeping the sampling rate unchanged, you get essentially the same size as mqa:

45474K - 2L-145_01_stereo.mqa.flac
45195K - 2L-45_stereo_01_FLAC_88k_24b.converted.flac
71914K - 2L-45_stereo_01_FLAC_88k_24b.flac

Now you have to re-encode the MQA to target 20 bits, not 24.