MQA is Vaporware

[firedog]

3 minutes ago, FredericV said:

It simply does not work like that

Exactly. It simply does not work as you think it does. 

[firedog]

8 minutes ago, mansr said:

In that case, why can't they simply complete their voodoo in the studio and avoid the need for special DACs entirely?

First, they claim there has to be deblurring on the users DAC end that is specific to each DAC for the full effect. 

And beyond that, I think you know the answer. For the same reason they aren't making a software MQA decoder available, when they have specifically said it is possible. They want the solution to be based in the users HW. 

[crenca]

5 minutes ago, firedog said:

Exactly. It simply does not work as you think it does. 

 

This reminds me of the exchange between The Computer Audiophile and esldude a while back (could have been this very thread).

 

I am curious firedog, even though you are technically correct as far as the way MQA is actually implemented (at least I think you are) is that how it is actually sold?  Do you just chalk the incongruity of how it actually works and how it is sold (as "end to end", "as the artist intends", etc. etc.) to the usually marketing fluff?

[firedog]

Just now, crenca said:

 

This reminds me of the exchange between The Computer Audiophile and esldude a while back (could have been this very thread).

 

I curious firedog, even though you are technically correct as far as the way MQA is actually implemented (at least I think you are) is that how it is actually sold?  Do you just chalk the incongruity of how it actually works and how it is sold (as "end to end", "as the artist intends", etc. etc.) to the usually marketing fluff?

Yep, at least in part. 

But even I don't think it is totally fluff. Conceptually, the idea that the digital recording and playback process is done according to a standard, and that the AD devices on the recording end, and the DA devices on the user end are configured to correct for the anomalies in the digital/analog processing on each end makes sense and should result in better sound. 

It's the way MQA is being marketed, and the possibility for DRM control of what we are allowed to listen to that concerns me. 

[mansr]

7 minutes ago, firedog said:

First, they claim there has to be deblurring on the users DAC end that is specific to each DAC for the full effect. 

And beyond that, I think you know the answer.

I know that there isn't actually anything DAC specific going on.

 

7 minutes ago, firedog said:

For the same reason they aren't making a software MQA decoder available, when they have specifically said it is possible. They want the solution to be based in the users HW. 

The reason they want it in the DAC is to extract royalties from one more party, and so they can sell it to the studios with a promise of end-to-end DRM.

[firedog]

Just now, mansr said:

I know that there isn't actually anything DAC specific going on.

 

The reason they want it in the DAC is to extract royalties from one more party, and so they can sell it to the studios with a promise of end-to-end DRM.

Again, I'm not disagreeing with you. Just disagreeing that a DAC with tube (or any other kind of output) violates the MQA definitions. 

[mansr]

3 minutes ago, firedog said:

Again, I'm not disagreeing with you. Just disagreeing that a DAC with tube (or any other kind of output) violates the MQA definitions. 

MQA is whatever they say it is, and they've apparently approved a tube piece, so as such you're trivially correct. It still doesn't make sense in the context of their marketing claims. Now I'm certain that if you just paid them their fees, they'd certify a tin can on a string too.

[tobes]

8 hours ago, mansr said:

I know that there isn't actually anything DAC specific going on.

 

PSAudio have stated that they didn't want to mess with the FPGA dac programming to accommodate MQA.

The PSAudio directstream dacs are MQA certified, but only (AFAIK) on their network bridge inputs - i.e. the network card/chips are doing the full MQA decode and then (one supposes) pass this on to the FPGA dac.

I'm curious how this complies with the dac 'deblurring' which is apparently central to the MQA process.

 

[Don Hills]

There's no magic to the deblurring. It can be done by any DAC with loadable filters. It would be child's play to someone like Mans. I believe the correct MQA filter coefficients were used to generate Archimago's test files. But the average audiophile may as well buy a DAC with it already built in.  

[mansr]

7 hours ago, tobes said:

PSAudio have stated that they didn't want to mess with the FPGA dac programming to accommodate MQA.

The PSAudio directstream dacs are MQA certified, but only (AFAIK) on their network bridge inputs - i.e. the network card/chips are doing the full MQA decode and then (one supposes) pass this on to the FPGA dac.

I'm curious how this complies with the dac 'deblurring' which is apparently central to the MQA process.

This is no different from, say, the Bluesound feeding decoded MQA to an off the shelf DAC chip. According to the marketing, they'd have measured the response of the DAC, whether a single chip or a complex FPGA-based design, and tweaked the decoder's filters to compensate for any (or at least some) anomalies. In reality, however, all DACs we've checked appear to use exactly the same filters, meaning there is no such tuning.

[tobes]

11 hours ago, mansr said:

 In reality, however, all DACs we've checked appear to use exactly the same filters, meaning there is no such tuning.

So, at the very least, dac compensation is subject to the implementation method of the MQA(?). This seems to be a very hit and miss way to rollout the product and doesn't provide much confidence in the whole 'Master Quality' mantra.

 

 

[loop7]

20 hours ago, tobes said:

PSAudio have stated that they didn't want to mess with the FPGA dac programming to accommodate MQA.

The PSAudio directstream dacs are MQA certified, but only (AFAIK) on their network bridge inputs - i.e. the network card/chips are doing the full MQA decode and then (one supposes) pass this on to the FPGA dac.

I'm curious how this complies with the dac 'deblurring' which is apparently central to the MQA process.

If I remember correctly, PS Audio was also uncomfortable with the level of access MQA required into their engineering design IP for MQA implementation. 

[mansr]

1 hour ago, tobes said:

So, at the very least, dac compensation is subject to the implementation method of the MQA(?). This seems to be a very hit and miss way to rollout the product and doesn't provide much confidence in the whole 'Master Quality' mantra.

It's Master Quality money going into Bob's pockets. That's all he cares about.

[firedog]

22 hours ago, Don Hills said:

There's no magic to the deblurring. It can be done by any DAC with loadable filters. It would be child's play to someone like Mans. I believe the correct MQA filter coefficients were used to generate Archimago's test files. But the average audiophile may as well buy a DAC with it already built in.  

Well the claim by people who's ability to write about is restricted by NDA is that there is more going on than the filtering and that Archi's tests aren't doing anything like testing MQA vs non MQA.....

[mansr]

16 minutes ago, firedog said:

Well the claim by people who's ability to write about is restricted by NDA is that there is more going on than the filtering and that Archi's tests aren't doing anything like testing MQA vs non MQA.....

Those people are not being entirely honest.

[Charles Hansen]

13 hours ago, firedog said:

Well the claim by people who's ability to write about is restricted by NDA is that there is more going on than the filtering and that Archi's tests aren't doing anything like testing MQA vs non MQA.....

 

Hi Firedog,

 

Who is saying that? One should remember that almost all of the people who have signed up are those that don't know how to make their own digital filters. I've been surprised at some of the claims made by those using MQA. I don't think they are lying, I think they truly don't understand the technology. There are less than a dozen  companies making their own digital filters.

 

Thanks,

Charles Hansen

[firedog]

1 hour ago, Charles Hansen said:

 

Hi Firedog,

 

Who is saying that? One should remember that almost all of the people who have signed up are those that don't know how to make their own digital filters. I've been surprised at some of the claims made by those using MQA. I don't think they are lying, I think they truly don't understand the technology. There are less than a dozen  companies making their own digital filters.

 

Thanks,

Charles Hansen

 

http://www.pinkfishmedia.net/forum/member.php?u=14119

I believe the poster in the link is John Westlake, known audio/dac designer.

[mansr]

6 minutes ago, firedog said:

http://www.pinkfishmedia.net/forum/member.php?u=14119

I believe the poster in the link is John Westlake, known audio/dac designer.

Requires registration.

[rickca]

16 minutes ago, firedog said:

I believe the poster in the link is John Westlake

John Westlake is the designer of the new Pro-Ject Pre Box S2 Digital which has MQA hardware decoding.

[Charles Hansen]

1 hour ago, firedog said:

http://www.pinkfishmedia.net/forum/member.php?u=14119

I believe the poster in the link is John Westlake, known audio/dac designer.

 

Hi Firedog,

 

As Mansr noted, it requires registration. Would you mind copying his post here? Thanks!

[Don Hills]

17 hours ago, mansr said:

Those people are not being entirely honest.

 

Mans, I haven't seen anything about the coefficients of the filters used to split (and later rejoin) the audio into the 0-24 and 24-48 KHz bands for "origami" folding. Are they also leaky and "time optimised"? I would expect the performance of these filters to be more critical than the anti-alias / anti image filters. 

[Charles Hansen]

3 hours ago, firedog said:

http://www.pinkfishmedia.net/forum/member.php?u=14119

I believe the poster in the link is John Westlake, known audio/dac designer.

 

Hello All,

 

I think I have found the post that John Westlake made:

 

JohnW Trade: Lakewest   Join Date: May 2010 Posts: 12,250

Quote: Originally Posted by Julf That was the original story, but as soon as Tidal became their main vehicle, all mentions of the added levels seem to have ended. We have seen no evidence of the supposed additional processing in hardware. Julf, I don't want to wade into the merits of MQA or not - but IMO the tests are flawed because they do not take into consideration the final "critical" rendering stage of MQA which strives to correct for the time domain issues of the DAC and "normalize" the entire replay chain. The first stage unfolding of MQA compressed data is not sonically relevant (which is the subject of the tests) - its the second DAC "Rendering" stage that's sonically most significant. Having a far few products pass through the MQA certification process I can testify to the fact that there a SIGNIFICANT engineering effort made by the MQA team to optimise the "renderer" for each DAC variant - which can also be extended to compensate for the DAC's Analogue LPF if required. Every MQA certified DAC has the hardware rendering stage. From the single direct DSD to MQA conversion "test file" I've heard, I still prefer the DSD track so while MQA is a step in the right direction, its sad that Native DSD is not more prevalent as its IMO sonically the closets "digital" format to the source material. MQA does try to "advance" the lowest common denominator - without the efforts of MQA I see nothing to stop us sliding into even lower Download audio quality - with files of unknown origin etc.

~~~~~~~~~~

 

It seems pretty clear from this that we have two conflicting sets of information. The first is that John Westlake designed a DAC for Pro-Ject that uses ESS DAC chips. He was told a story by MQA about all the "engineering efforts made by the MQA team".

 

However it seems that the story is just that -- a story. When Mansr and Archimago dumped the filter responses of MQA "renderers" using completely different DAC chips (Burr-Brown in the Explorer2 and ESS in the DragonFly), the results were identical. In other words no difference at all between the filters whatsoever.

 

Based on the evidence, I can reach no other conclusion than that John Westlake is simply repeating (yet another) story told by MQA that happens to be untrue - NDA or no NDA.

 

Hope that helps,

Charles Hansen

 

EDIT: PS - Please note that even John Westlake prefers DSD to MQA. He just sees it as a step up from the standard brickwall filters used by most DACs. No need for a proprietary format with licensing fees, royalties, and DRM to use better sounding digital filters that make all of your music sound better and not just the MQA files.

[Charles Hansen]

5 hours ago, firedog said:

 

http://www.pinkfishmedia.net/forum/member.php?u=14119

I believe the poster in the link is John Westlake, known audio/dac designer.

 

Hi Firedog,

 

As noted John Westlake has designed many DACs, starting with one for Pink Triangle many years ago. However I can find no evidence that he has ever designed his own custom digital filter. This situation is very reminiscent of when HDCD was popular in the mid-'90s. Very few of the high-end manufacturers understood digital at all. There were only two companies on the planet making their own custom digital filters - Wadia and Theta. All of the rest were happy as clams to buy digital filter chips from a third party. There were only a few choices back then, and the HDCD "decoder" chip included one in it. At the time, Robert Harley (then with Stereophile) was claiming that in his experience all of the best sounding DACs used the HDCD digital filter.

 

Many, many, many companies were happy as clams to be able to purchase "the best" digital filter without having to know anything more than how to connect it. But that was over 20 years ago. Today there is little excuse for this - except for mid-fi products, or those trying to meet a very low price point. To implement a custom digital filter requires at the very least a little knowledge. Almost all of the ESS DAC chips allow for custom coefficients to be programmed (which is what was done for the Pono Player) and that keeps the cost down.

 

To go to the next step up in performance requires more computational horsepower than is in the DAC chip, so either a DSP chip or an FPGA must be used. Both of these add significantly to the cost of the unit, plus require an understanding of how digital filters work. And all custom digital filter solutions require that someone has done a thorough investigation of what the available variables affect in terms of sound quality. There are still only around a dozen companies that have taken the trouble to do so.

 

Almost everything about the Pro-Ject S2 DAC is simply well-implemented off-the-shelf solutions - from the XMOS USB receiver to the MQA custom filter to the choice of 7 other filters built into the ESS DAC chips used to the ESS headphone amp chip to the ESS DAC power supply chip. The only innovation I see in that unit is a circuit to clean up the incoming power from the USB cable that powers the unit. That is certainly a good idea, although many already do something similar. Perhaps he has done it better, but he also states on the Pinkfish forum thread that the sound is significantly improved by using a better external power supply. So that innovation is limited at best, it seems to me.

 

It certainly seems to be a good solid design and competent for the money, but there is only so much anybody can do at that price point - €350. Every penny counts when making that kind of product, even at the very high volumes that Pro-Ject is able to reach, with their vast European marketing network. But we have not reached the point where people are turning lead into gold or water into wine. If that's all the money you have and it provides the features you want, it is likely a very good choice. But it does not even offer an Ethernet input for direct streaming, so one either needs to attach a computer or dedicated streamer (such as the micro-Rendu) to even take advantage of the MQA feature. All of a sudden you are looking at something that is well over $1000 in order to listen to the only MQA source currently available - Tidal.

 

Just keeping things in perspective here. In that regard, it is not that much different from the Meridian Explorer2, currently for sale at $200 on Amazon, or even the AudioQuest DragonFly Black ($100) or Red ($200). It adds two S/PDIF inputs (co-ax and Toslink), has a nicer form factor (IMO), and has separate headphone and line outs (single-ended only). If those features are important to you, it may be worth roughly twice the price of the Meridian, or quadruple or double the price of the DragonFlies (I don't know what the US retail of the Pro-Ject is yet). It's a big world, with lots of products at all kinds of price points with all kinds of feature sets.

 

Despite Mr. Westlake's experience designing complete DACs, it doesn't appear that he has a thorough understanding of designing digital filters. As I've noted before in this thread, this seems to be very common. One can attribute the anti-MQA attitude of those who design custom digital filters either to greed on the part of those manufacturers (as they want to retain their proprietary performance advantage) or to greater insight as to what MQA is actually doing to the music. (I'm not sure why one would be predisposed to select the first choice as obviously there is also a greed motive for MQA.) One would have to think that all of the anti-MQA digital audio designers are greedy and selfish, while MQA and all of the MQA adopters are wonderful, charitable organizations to come to that conclusion.

 

Hope this helps,

Charles Hansen

[Charles Hansen]

3 hours ago, Don Hills said:

Mans, I haven't seen anything about the coefficients of the filters used to split (and later rejoin) the audio into the 0-24 and 24-48 KHz bands for "origami" folding. Are they also leaky and "time optimised"? I would expect the performance of these filters to be more critical than the anti-alias / anti image filters. 

 

Hi Don,

 

In the Pinkfish thread where John Westlake (designer of an MQA-enabled DAC) posts, he claims the exact opposite - that the "real secret" are the reconstruction filters used in the DAC:

 

http://www.pinkfishmedia.net/forum/showthread.php?t=206723

 

(No registration required to view.)

 

In this post:  http://www.pinkfishmedia.net/forum/showpost.php?p=3210264&postcount=69   Mr. Westlake shows his lack of understanding as he apparently does not realize that the ESS DAC chips used actually allow the digital filter to be completely bypassed.

 

Further underscoring his lack of understanding is here:  http://www.pinkfishmedia.net/forum/showpost.php?p=3210276&postcount=71  where he regurgitates the MQA talking points despite the fact that Mansr and Archimago have measured them and shown them not to be true. In that same post he shows further lack of understanding by claiming that "many DAC's are only 24 but [sic] input limiting the mathematical precession [sic] of the process - hardware rendering avoids all these issues" as apparently he doesn't realize that MQA limits the audio resolution to only about 17 bits (MQA's noise-shaped dithering results in the noise floor varying throughout the audio band) so there is very little point in using filters with more than 24 bits of precision.

 

Finally in this post:  http://www.pinkfishmedia.net/forum/showpost.php?p=3210290&postcount=76  Mr. Westlake believes that he understands the advantages of MQA but that MQA has simply not properly explained them to the public, which I find to be somewhat ironic.

 

Hope this helps,

Charles Hansen

[firedog]

46 minutes ago, Charles Hansen said:

 

Hi Don,

 

In the Pinkfish thread where John Westlake (designer of an MQA-enabled DAC) posts, he claims the exact opposite - that the "real secret" are the reconstruction filters used in the DAC:

 

http://www.pinkfishmedia.net/forum/showthread.php?t=206723

 

(No registration required to view.)

 

In this post:  http://www.pinkfishmedia.net/forum/showpost.php?p=3210264&postcount=69   Mr. Westlake shows his lack of understanding as he apparently does not realize that the ESS DAC chips used actually allow the digital filter to be completely bypassed.

 

Further underscoring his lack of understanding is here:  http://www.pinkfishmedia.net/forum/showpost.php?p=3210276&postcount=71  where he regurgitates the MQA talking points despite the fact that Mansr and Archimago have measured them and shown them not to be true. In that same post he shows further lack of understanding by claiming that "many DAC's are only 24 but [sic] input limiting the mathematical precession [sic] of the process - hardware rendering avoids all these issues" as apparently he doesn't realize that MQA limits the audio resolution to only about 17 bits (MQA's noise-shaped dithering results in the noise floor varying throughout the audio band) so there is very little point in using filters with more than 24 bits of precision.

 

Finally in this post:  http://www.pinkfishmedia.net/forum/showpost.php?p=3210290&postcount=76  Mr. Westlake believes that he understands the advantages of MQA but that MQA has simply not properly explained them to the public, which I find to be somewhat ironic.

 

Hope this helps,

Charles Hansen

And do have any idea what he means by this:

http://www.pinkfishmedia.net/forum/showpost.php?p=3211170&postcount=115

Quote

As I also mentioned the Renderer also has other processing - not just the Time domain processing.