MQA is Vaporware

[Cebolla]

15 hours ago, FredericV said:

I don't understand why the believers still believe there is a third unfold

 

 

Bob Stuart's multifold showing diagrams explain the belief. The first one slyly indicates that MQA has 'lossless' content in at least one of the 'unfolds', BTW:

http://bobtalks.co.uk/blog/science-mqa/mqa-playback/

[FredericV]

4 hours ago, Cebolla said:

 

Bob Stuart's multifold showing diagrams explain the belief. The first one slyly indicates that MQA has 'lossless' content in at least one of the 'unfolds', BTW:

http://bobtalks.co.uk/blog/science-mqa/mqa-playback/

 

He calls this science? 24 bit files stop at the red line, there is no space to encode C when you have a range of 0 -> -144db ....



To have C, you would need a 32 bit distribution file. Not going to happen anytime soon, furthermore files which claim to be 32 bit per sample, are usually floating point and used as internal format in the studio. MQA is not a floating point file. MQA is not an internal studio format, but a consumer distribution format with crypto DRM and entropy reduction aka lossy. There is no third unfold.

https://www.sounddevices.com/32-bit-float-files-explained/
 

[John Dyson]

4 hours ago, Cebolla said:

 

Bob Stuart's multifold showing diagrams explain the belief. The first one slyly indicates that MQA has 'lossless' content in at least one of the 'unfolds', BTW:

http://bobtalks.co.uk/blog/science-mqa/mqa-playback/

One thing to me -- VERY SUSPICIOUS -- appealing to the 'authority' of the 'mastering community'.   That is like appealing to Radio Shack for high equipment standards.   So much nonsense has come out of the mastering community, even 'elite' remastered stuff.

 

Anyone with any technical background AT ALL knows that pristine recordings sound pristine, UNTIL THEY ARE MASTERED.  I have heard the good stuff from time to time, and it often sounds NOTHING like what is received by consumers.

 

The consumer has been duped all along -- sending perfect (I mean, audibly perfect) recordings to CD is no rocket science.   It only requires proper handling of the signal from the original to the 44.1k/16bits.   I really dislike 44.1k, but it is capable of being MUCH BETTER than what most people hear as high-res remastered.

 

Please, get rid of the mastering, just play-out the master tapes -- proper levels, proper decoding and NO NORMALIZATION, and only the EQ required to get rid of BUGS.   No artistic endeavors, please.  This stuff is NOT a science project, and even CD technology is 30+yrs old.  There is nothing new here.

 

The only thing new is that a few of us have realized the precise *nature* of the poor quality of the distributed recordings.  We have known for quite some time that there is something wrong with the sound -- NOW WE KNOW WHY.   Ignorance is now voluntary.  We all know (if we look) the specific character of the damage -- and remastering without correcting the damage is just another way of taking money from people.  Remastering SHOULD be about supplying superior product, not about yet another distorted copy.

 

MQA FIXES NOTHING, but only changes/encrypts/folds or whatever the still damaged recording.  There is ZERO practical improvement in MQA.   The only possible improvements that I have seen recently is the 'time base correction' scheme for magnetic tape, and the ability to fairly accurately recover the recordings from the mismastered digital copies.

 

Everything else is specious -- might as well have 64bit double precision floating point inaccurate copies -- the result is still the same, and the quality is no higher.  The results are still inaccurate until the industry starts giving the customer what they have been paying for all along.  We have plenty of space and bandwidth to handle 32bit floating point if we wanted, but 24bit 96k is good enough -- no encryption, folding, compression, etc needed -- until the customer does it locally.

 

John

 

[Cebolla]

52 minutes ago, FredericV said:

He calls this science? 24 bit files stop at the red line, there is no space to encode C when you have a range of 0 -> -144db ....
To have C, you would need a 32 bit distribution file. Not going to happen anytime soon, furthermore files which claim to be 32 bit per sample, are usually floating point and used as internal format in the studio. MQA is not a floating point file

 

Who needs the 'extra' 8 bits, when C is actually upsampling, with a questionable minimum phase filter, as close as possible to the so called 'original' sample rate within the capabiliies of the DAC? 🙂

 

Also, the 'lossless' claim is not lost on elements of the hi-fi press that blindly repeat marketing statements:

https://www.whathifi.com/advice/mp3-aac-wav-flac-all-the-audio-file-formats-explained

 

 

[John Dyson]

5 minutes ago, Cebolla said:

 

Who needs the 'extra' 8 bits, when C is actually upsampling, with a questionable minimum phase filter, as close as possible to the so called 'original' sample rate within the capabiliies of the DAC?

 

Also, the 'lossless' claim is not lost on elements of the hi-fi press that blindly repeat marketing statements:

https://www.whathifi.com/advice/mp3-aac-wav-flac-all-the-audio-file-formats-explained

 

 

Heh -- a minimum phase filter necessarily causes relative time delays vs. frequency.   It cannot be an accurate representation of the signal (well, unless the filter is very short, at that point -- makes no difference minimum vs. linear phase.)   A linear phase filter produces a proper constant time delayed representation of a rolloff -- if the rolloff effect is what is desired.

 

Now, for a audio special effect, a minimum phase filter is just fine.  The music might even sound better with a minimum phase filter, but accuracy is NOT one of its attributes.   Varying time of arrival (propagation time) vs frequency IS one of the minimum phase filter attributes.

 

 

John

 

[FredericV]

31 minutes ago, John Dyson said:

Now, for a audio special effect, a minimum phase filter is just fine.  The music might even sound better with a minimum phase filter, but accuracy is NOT one of its attributes.   Varying time of arrival (propagation time) vs frequency IS one of the minimum phase filter attributes.

 

When we researched MQA's & Ayre's filter, which is minimum phase with one cycle of postringing, everything sounded too tight. For EDM and euro dance hits from the 90's, this would be a nice gimmick, but as the post-ringing has been artificially shortened, something is missing. Voices are not natural with this filter, something is truly missing. Having a more natural post-tail also gives our neurological system some extra post-info after transients or point source events, which helps a lot. The reason being post-echo's are sustained a little bit longer, which helps our neurologic system. The only thing we cannot handle well, is having audio before the actual event occurs, so pre-ringing with linear phase, smears the time domain. So I actually oppose all of MQA's time domain hocus pocus claims as they sound wrong to my ears. The way MQA tries to solve this pre-ringing is wrong.

To my ears, the ultimate filter is Archimago's Intermediate Phase filter: phase correct in the audible spectrum, and almost no pre-ringing and mainly post-ringing, but with a longer tail than MQA's fake content generator aka it's leaky filter which causes fake ultrasonics, which uses the shortest possible upsampling post-ringing tail, but with all side effects associated with such filter..

So some time ago we decided to implement this filter and give it away as a free software update.Anyone can google the filter specifications and modify their upsampler, and this filter sounds in every way superior than MQA's second unfold upsampler.

[Confused]

24 minutes ago, FredericV said:

 

When we researched MQA's & Ayre's filter, which is minimum phase with one cycle of postringing, everything sounded too tight. For EDM and euro dance hits from the 90's, this would be a nice gimmick, but as the post-ringing has been artificially shortened, something is missing. Voices are not natural with this filter, something is truly missing. Having a more natural post-tail also gives our neurological system some extra post-info after transients or point source events, which helps a lot. The reason being post-echo's are sustained a little bit longer, which helps our neurologic system. The only thing we cannot handle well, is having audio before the actual event occurs, so pre-ringing with linear phase, smears the time domain. So I actually oppose all of MQA's time domain hocus pocus claims as they sound wrong to my ears. The way MQA tries to solve this pre-ringing is wrong.

To my ears, the ultimate filter is Archimago's Intermediate Phase filter: phase correct in the audible spectrum, and almost no pre-ringing and mainly post-ringing, but with a longer tail than MQA's fake content generator aka it's leaky filter which causes fake ultrasonics, which uses the shortest possible upsampling post-ringing tail, but with all side effects associated with such filter..

So some time ago we decided to implement this filter and give it away as a free software update.Anyone can google the filter specifications and modify their upsampler, and this filter sounds in every way superior than MQA's second unfold upsampler.

A little off topic here (more like MQA isn't Vaporware), but have you tried the MQA filters in HQPlayer?

[John Dyson]

17 minutes ago, FredericV said:

 

When we researched MQA's & Ayre's filter, which is minimum phase with one cycle of postringing, everything sounded too tight. For EDM and euro dance hits from the 90's, this would be a nice gimmick, but as the post-ringing has been artificially shortened, something is missing. Voices are not natural with this filter, something is truly missing. Having a more natural post-tail also gives our neurological system some extra post-info after transients or point source events, which helps a lot. The reason being post-echo's are sustained a little bit longer, which helps our neurologic system. The only thing we cannot handle well, is having audio before the actual event occurs, so pre-ringing with linear phase, smears the time domain. So I actually oppose all of MQA's time domain hocus pocus claims as they sound wrong to my ears. The way MQA tries to solve this pre-ringing is wrong.

To my ears, the ultimate filter is Archimago's Intermediate Phase filter: phase correct in the audible spectrum, and almost no pre-ringing and mainly post-ringing, but with a longer tail than MQA's fake content generator aka it's leaky filter which causes fake ultrasonics, which uses the shortest possible upsampling post-ringing tail, but with all side effects associated with such filter..

So some time ago we decided to implement this filter and give it away as a free software update.Anyone can google the filter specifications and modify their upsampler, and this filter sounds in every way superior than MQA's second unfold upsampler.

 

I have explained this before, but I know that I don't have a wide audience.   The 'ringing' associated with FIR filters is NOT ringing, but is Gibbs effect.  It is basically the left-overs (kind of a residue effect) from missing frequency components.  This is NOT energy storage like the ringing in a high-Q filter.   Gibbs effect is a 'residue' from missing components that almost seem like sine waves, but as a composite are NOT.  They are individually a mixed set of MISSING sine waves that don't cancel and re-enforce anymore, because they have been chopped-off/missing.  They leave little droppings in synthesized waveforms like square waves, but it isn't that they exist, it is that the square wave is missing the chopped off high frequency components.

 

When the 'Gibbs effect' moves around vs filter type, it happens because of the time delay differences VS frequency for a linear phase (constant delay) vs any other filter.

 

So, if a filter seems to sound different because of those little 'wobbly things' moving around, the wobbly things aren't why it really sounds different.  The different filters sound different because of the varying delay vs. frequency -- therefore times of arrival at your ears are different.

 

The times of arrival that are important are not so much associated with the missing frequency components in the Gibbs effect (pseudo-ringing or more accurately, wobbles :-)), but are because of the differing times of arrival at more audible frequencies.

 

One might ask -- what are the varying delay lengths? -- well, the maximum possible delay is the total number of taps in the filter.   For linear phase, the time delay is all constant -- essentially perfect phase response over the frequency range.  For linear phase filters, the delay through the filter is typically 1/2 of the number of taps.   For not-linear-phase, the time delay varies, so that perhaps low frequencies or high frequencies might arrive faster than the other end of the spectrum.

 

Whether linear phase or not-linear phase sounds better -- that might be the observation where some intermediate phase/delay variation seems to sound superior -- but has little to do with the 'wobbles' moving around.   Longer filters can create more varying delays vs frequency than shorter filters -- this is probably one reason for some mistaken reasoning about the longer filters.  Of course, a longer not-linear phase filter can sound different than a shorter near-equivalent -- because the relative propagation delays through the filters can vary.

 

An infinitely (well -- practically infinitely) long linear phase filter with the similar frequency response of a short filter will sound very similar except for the longer constant time delay.  I mix and match linear phase filters all of the time, and they add together very nicely with no cancellation effects.

 

As soon as you use a filter which does not have constant delay, then adding together the resulting signals, even if the filters are of the same lengths, will have all kinds of cancellation, comb effects, etc.   Any filter except for linear phase starts getting into complexities that aren't good -- except for maybe one last filter, where there is no further processing.   Similar issues happen when mixing together IIR and analog filters.   Linear phase filters open up a very wide array of very accurate signal processing options.   Not-linear phase filters can do fancy things also, but are much more mathematically complex to deal with -- best avoided, except for special sound effects.

 

John

 

[#Yoda#]

On 6/25/2020 at 6:56 PM, FredericV said:

When we researched MQA's & Ayre's filter, which is minimum phase with one cycle of postringing, everything sounded too tight. For EDM and euro dance hits from the 90's, this would be a nice gimmick, but as the post-ringing has been artificially shortened, something is missing. Voices are not natural with this filter, something is truly missing.

This may well be true for the MQA filters, but many Ayre DAC users will disagree with you here because of other listening experiences with the Ayre minimum phase filters implemented. Perhaps it is often due to the speaker used rather than the DAC filter that e.g. voices are not reproduced true to life. 

 

Not least because of extensive listening comparisons of different filter types, Charley Hansen continued to choose Ayres special minimum phase filter e.g. for the DAC of the QX-5 Twenty as can be seen from this quote from an John Atkinson article on Stereophile.com: "In an e-mail, Hansen told me that, during the development of Ayre's digital products, "We have custom filters whereby we can load whatever coefficients we want into [the FPGA] . . . an external switch allows us to load different filter coefficients and hear the audible changes they make. This is a great test, as it allows for everything else to be held completely constant. I spent four solid months doing nothing but listen[ing] to the effects of various filters—corner frequency, stop-band response, minimum-phase vs linear-phase, apodizing, sharpness at the 'knee,' windowing functions, interpolation rates (eg, 2x vs 4x vs 8x vs 16x), dithering functions . . . every single factor I could think of." There is no MQA decoding, however, as Ayre is still evaluating the codec."

 

Regarding the last sentence one more remark: I know that Charley already had a firm opinion about MQA and its minimum phase filters at that time, which he later expressed impressively in this thread and others. 

 

[Currawong]

The ringing seen before and after an impulse response are the result of what is an illegal, out-of-band signal being filtered. They do not appear when an in-band signal is put through the filter. Archimago's post, buried in all the MQA threads, of the effects of the digital filters on actual in-band sine waves is more relevant. I'll edit this post if I can find it.

[Fokus]

On 6/25/2020 at 5:33 PM, FredericV said:

He calls this science? 24 bit files stop at the red line, there is no space to encode C when you have a range of 0 -> -144db ....

 

That is quite wrong.

 

The 24 bit equivalent noise level of -144 dB holds for the noise integrated over the full available bandwidth (i.e. Fs/2). Conversely, the noise density at each frequency is at a much lower level. It is this density that is plotted in the FFT.

 

For 16 bit the noise line in an FFT is also not at -96dB, but much lower, and depending on the FFT's bin size.

 

[Fokus]

8 hours ago, Currawong said:

The ringing seen before and after an impulse response are the result of what is an illegal, out-of-band signal being filtered.

 

An impulse stimulus is indeed illegal, and thus misleading, but only for the specific case of a DAC, or more generally whenever the sampling rate is increased starting from a given signal in the digital domain.

 

It is not illegal for the general case where the sampling rate is reduced, with as a limit case the analogue to digital conversion and its anti-aliasing filter.

[sphinxsix]

 

I just got to know that CDJapan has informed about cancelling of number of planned MQA/UHQCD releases. Don't cry, please.

 

[The Computer Audiophile]

3 minutes ago, sphinxsix said:

 

I just got to know that CDJapan has informed about cancelling of number of planned MQA/UHQCD releases. Don't cry, please.

 

Oh no. People will be stuck with unaltered lossless audio. What a shame :~)

[sphinxsix]

37 minutes ago, The Computer Audiophile said:

Oh no. People will be stuck with unaltered lossless audio. What a shame :~)

Schiit happens! 

[sphinxsix]

No-one said our audio journey would be nice'n'easy after all..   BTW, just thinking.. shouldn't audio formats that didn't make it, get immortalized let's say in the form of e.g. monuments.? What could an MQA monument look like..?  

[sandyk]

2 hours ago, sphinxsix said:

 

I just got to know that CDJapan has informed about cancelling of number of planned MQA/UHQCD releases. Don't cry, please.

 

 Don't get too excited yet. CD Japan needs to cancel ALL planned MQA releases.

[KeenObserver]

Anyone know the status of MQA Ltd?  Trying to figure out their last two postings on Companies House.  Looks like Reinet Investments updated their address.  In regards to Director Jbara the part after new information is blank.

[The Computer Audiophile]

The best way to find out about MQA ltd, I believe, is to check for people moving on to new jobs. If the ship is sinking, people usually jump off. 

[mav52]

1 hour ago, The Computer Audiophile said:

The best way to find out about MQA ltd, I believe, is to check for people moving on to new jobs. If the ship is sinking, people usually jump off. 

 

Agree but some companies still add MQA to their DACS and streamers. I guess the want to suck a few dollars off the consumers.

[MikeyFresh]

1 hour ago, mav52 said:

Agree but some companies still add MQA to their DACS and streamers. I guess the want to suck a few dollars off the consumers.

 

Yes and that always the case, most of these companies have engineers that know what's what. This was always a case of ticking off "features" for marketing purposes, and ultimately, shifting boxes in the name of $$$, not superior digital audio performance or value to the consumer, quite the opposite actually.

[Allan F]

On 7/10/2020 at 3:11 PM, sphinxsix said:

No-one said our audio journey would be nice'n'easy after all..   BTW, just thinking.. shouldn't audio formats that didn't make it, get immortalized let's say in the form of e.g. monuments.? What could an MQA monument look like..?  

 

I note that you are from The Netherlands. Nothing personal, but If you were aware of how sensitive the topic of monuments currently is in the US, I don't think that you would be referring to monuments in an audio forum.

[sphinxsix]

10 hours ago, Allan F said:

 

I note that you are from The Netherlands. Nothing personal, but If you were aware of how sensitive the topic of monuments currently is in the US, I don't think that you would be referring to monuments in an audio forum.

 

I'm afraid I will have to disappoint you, my friend - I would have anyway   The reason for that is very simple - if I decided not to touch on topics which are 'sensitive' ones in some particular place or circles, I would probably not be able to speak, or post on any topic at all.

Including MQA.. 

I don't think my attitude towards racial issues requires correction, I also don't think any of my friends of 'colour' would say it does 

IMO political correctness can often be compared to an attempt of getting rid of the symptoms of a disease instead of curing its actual source which in many cases can be a painful and long lasting process..

But we would have to go way out of topic to discuss it here 

Cheers!

 

[bambadoo]

14 hours ago, The Computer Audiophile said:

The best way to find out about MQA ltd, I believe, is to check for people moving on to new jobs. If the ship is sinking, people usually jump off. 

You mean from Wimp/Tidal to  MQA to lets say Roon(which comes from Soloos/Meridian) ?

 

[Allan F]

5 hours ago, sphinxsix said:

I don't think my attitude towards racial issues requires correction, I also don't think any of my friends of 'colour' would say it does.

 

No one was trying to correct your attitude towards racial issues or remotely suggesting that it needed correction. You misinterpreted my comment and became unnecessarily defensive. However, showing deference to sensitive issues has nothing to do with political correctness, notwithstanding your apparent view to the contrary. But, no worries.