Article: MQA: A Review of controversies, concerns, and cautions

[The Computer Audiophile]

17 minutes ago, Archimago said:

 

For atypical DAC varieties (eg. not the typical XMOS + ESS chipsets), there's going to be a bit of custom work to implement MQA. Not sure for example at what cost or who would do the work for a device like the PS Audio's DirectStream DAC itself. Also might depend on how close to the chest MQA likes to keep their source code especially given the secret "authentication" algos :-).

 

Yeah, sounds like the Bridge II card will just do the software decode for the network streams; although I see reports of it upsampling up to 176.4/192kHz presumably with their minimum phase slow-roll filter varieties.

 

You nailed it. Cost are high to do this because of the time it takes and skills required. It took dCS several thousand hours and unprecedented access to MQA to implement it as an option in dCS DACs. 

 

[WiWavelength]

1 hour ago, Archimago said:

 

For atypical DAC varieties (eg. not the typical XMOS + ESS chipsets), there's going to be a bit of custom work to implement MQA. Not sure for example at what cost or who would do the work for a device like the PS Audio's DirectStream DAC itself. Also might depend on how close to the chest MQA likes to keep their source code especially given the secret "authentication" algos :-).

 

Yeah, sounds like the Bridge II card will just do the software decode for the network streams; although I see reports of it upsampling up to 176.4/192kHz presumably with their minimum phase slow-roll filter varieties.

 

 

A thought followed by a related question...

 

I have wondered if some of the myriad MQA upsampling filters -- most of which appear to be rarely, if ever utilized -- exist to counteract, per se, any un-MQA-like characteristics that could be introduced downstream in non-defeatable processing in FPGA or other heavily customized DACs.

 

Has anyone run frequency domain transforms on all of the MQA upsampling digital filters?  I would be curious to see their frequency responses.

 

AJ

[firedog]

2 hours ago, labjr said:

 

He didn't say anything about FPGA's  or their business model.

 

He said: 

 

"MQA requires changes to our D to A converters that we are not willing to make. Because we don't want to compromise their performance or their sound." 

 

 

 

First, "our D to A converter" IS an FPGA. That's the context. 

Second, the wider context, if you understand their use of FPGA and their business model, is exactly what I said. Here's one  other quote by Paul M. on the subject (there are many similar ones):

 

Quote

Ted and the engineers continue to improve the state of the art with digital filters on a regular basis – and one of the whole reasons we went to an FPGA based DAC in the first place was the freedom to improve the product’s performance over time. If we were restricted in what we could to future developments, held back because everything in the USB chain had to be 'fixed' by the MQA engineers first… that would be nuts.

 

I'm not going to bring you a full page of quotes of his on the subject. So yes, I interpreted what he said based on understanding the context and what  he's said over time about MQA. It doesn't make sense for a DAC business model based on regular SW updates that are essentially giving you a "new" DAC with changes to the filtering on a regular basis. 

[miguelito]

7 hours ago, Archimago said:

 

Looks like the MQA version soft decoded is a little better just by the numbers. My CD version is exactly the same as the Tidal HiFi.

 

Either way, at DR <6, this is nasty. 

The production is horrible. But the music is wonderful, so improvements make a big difference in my book. 

[mansr]

6 hours ago, WiWavelength said:

Has anyone run frequency domain transforms on all of the MQA upsampling digital filters?  I would be curious to see their frequency responses.

Frequency responses of the 16 4x upsampling filters on the Bluesound devices:

[miguelito]

9 hours ago, The Computer Audiophile said:

You nailed it. Cost are high to do this because of the time it takes and skills required. It took dCS several thousand hours and unprecedented access to MQA to implement it as an option in dCS DACs. 

 

Right.

 

In dCS DACs, the first unfold is done in the network card (currently Rossini, and will be done in the Network Bridge, and Vivaldi Upsampler). My understanding is MQA Ltd and dCS worked together to write the FPGA code to implement the MQA filter. I get the impression that there are complications to this code above and beyond implementing the filter itself, as the filter needs to read the MQA filter spec on the first-unfolded PCM stream and chose it's parameters accordingly. Probably easier said than done in an FPGA, especially if you have to spend some time every few samples to check for changes in the MQA filter spec.

 

I can't imagine there's anything particularly harmful for PS Audio to implement an MQA filter in the FPGA. They might just not be willing to make the investment.

[The Computer Audiophile]

Investment, access to MQA ltd IP, and the skilled people to actually do it. 

 

Whether one likes it or not is another animal.

[firedog]

And access of MQA Ltd to their IP, which they may not want to give out.

 

AFAIK, they basically write firmware updates, listen, and as soon as they like what they hear they release it to the DAC owners. All at their offices. 

 

Putting MQA approval-code writing into this process would complicate it and slow it down, possibly quite a bit - not to mention the possibility it could also change the sound of the firmware they just wrote and decided they were happy with. 

[crenca]

21 minutes ago, The Computer Audiophile said:

Investment, access to MQA ltd IP, and the skilled people to actually do it. 

 

Whether one likes it or not is another animal.

 

Interesting.  Is there any other DAC manufacturer who has this kind of relationship with MQA that you know of?  Also, why dCS?  Often such things are forged personally, for example is the owner/principle of dCS a current or former employee/associate of Bob S?

[The Computer Audiophile]

1 minute ago, crenca said:

 

Interesting.  Is there any other DAC manufacturer who has this kind of relationship with MQA that you know of?  Also, why dCS?  Often such things are forged personally, for example is the owner/principle of dCS a current or former employee/associate of Bob S?

Not that I know of, but that doesn't mean much. 

 

I think there is definitely a Cambridge connection between Bob and dCS. 

[miguelito]

6 minutes ago, crenca said:

 

Interesting.  Is there any other DAC manufacturer who has this kind of relationship with MQA that you know of?  Also, why dCS?  Often such things are forged personally, for example is the owner/principle of dCS a current or former employee/associate of Bob S?

My understanding is that actually the office are very close in Cambridge. Surely there's been a pint or two (or 500) in between dev sprints.

[Archimago]

4 hours ago, miguelito said:

The production is horrible. But the music is wonderful, so improvements make a big difference in my book. 

 

Agree...

 

And I hope for the sake of all that's good left in music production that they still have high quality, uncompressed tracks in the vaults such that some day this could be properly redone free from being a "loudness war" casualty.

 

[crenca]

Just now, Archimago said:

 

Agree...

 

And I hope for the sake of all that's good left in music production that they still have high quality, uncompressed tracks in the vaults such that some day this could be properly redone free from being a "loudness war" casualty.

 

 

On this subject, rather than crosspost here:

 

https://www.stereophile.com/comment/573833#comment-573833

[miguelito]

55 minutes ago, Archimago said:

Agree...

 

And I hope for the sake of all that's good left in music production that they still have high quality, uncompressed tracks in the vaults such that some day this could be properly redone free from being a "loudness war" casualty.

That would be amazing, but I am not holding my breath...

[mitchco]

52 minutes ago, Archimago said:

 

Agree...

 

And I hope for the sake of all that's good left in music production that they still have high quality, uncompressed tracks in the vaults such that some day this could be properly redone free from being a "loudness war" casualty.

 

 

Unfortunately, for rock, pop, blues and most anything not classical, this will not be the case. As an ex-recording/mixing engineer, I can say that tracks are already compressed before even being recorded. Meaning, most mixing consoles have compressors/limiters on every channel strip. More often than not, compression and limiting is applied on the way to the multitrack recorder (analog or digital). More often than not, during mixdown, not only possible on individual tracks, but absolutely on the stereo master bus before hitting tape or digital. Finally, applied again during the mastering process.

 

Speaking with one of the engineers that did the mass majority of HDTracks transfers, more often than not, receives whatever is provided by the publishing company, and more often than not, it is a "safety" master is provided, which is an already compressed mixdown on tape or digital. The best the engineer can do is eq the master and add some stereo processing, depth, maybe add a bit of reverb...

 

Finally, virtually all music has passed through the device before it even reaches your ears. The simple reality is that we have all heard the sound of this device since 1967 – that’s 50 years ago folks. In fact, if you are listening to mainstream music right now, it is likely that you are also listening to the sound of this device. If you look at how the  Universal Audio 1176LN Peak Limiter works, it can substantially alter the music waveform in a way that there is no coming back, regardless of anything MQA can or cannot do.

 

Not to paint the bleakest picture :), the likelihood of uncompressed tracks being in vaults, is about 0. Given the ubiquity of "studios in a box" and access to everyone, proper professional recording, mixing and mastering is becoming (or has become) a lost art...

 

If the consumer audiophile industry wanted to really do something, demanding better recordings, mixes and masters is where we would see real audible sonic improvements. Sorry Arch, did not mean to single you out, just a PSA.

[miguelito]

1 hour ago, crenca said:

On this subject, rather than crosspost here:

 

https://www.stereophile.com/comment/573833#comment-573833

I am a huge Moby fan. Really the hugest Moby fan. By far.

 

Suggest you guys listen to "Wait For Me" (both regular and ambient versions), "Destroyed", "Play B-sides" (Flying Foxes in particular). Jeezus...

[crenca]

2 minutes ago, miguelito said:

I am a huge Moby fan. Really the hugest Moby fan. By far.

 

Suggest you guys listen to "Wait For Me" (both regular and ambient versions), "Destroyed", "Play B-sides" (Flying Foxes in particular). Jeezus...

 

I hit edit on my post and it went into some kind of moderation mode, so my link is dead for now at least.  I talked about how instead of decently mastered 16/44 "audiophile releases" (Tony's question) we got...MQA!

 

I can't say I am much of a Moby fan, but I have slowly been collecting a bit of the better sounding and interesting eletronica.  I notice that "Wait for Me" is not on Tidal for some reason...

[miguelito]

3 hours ago, crenca said:

I hit edit on my post and it went into some kind of moderation mode, so my link is dead for now at least.  I talked about how instead of decently mastered 16/44 "audiophile releases" (Tony's question) we got...MQA!

There are so many fantastically sounding 16/44 recordings!

 

Quote

I can't say I am much of a Moby fan, but I have slowly been collecting a bit of the better sounding and interesting eletronica.  I notice that "Wait for Me" is not on Tidal for some reason...

Wow! That's weird. I own every single thing Moby has released, some in multiple versions, as well as things were he was a producer (eg Voodoo Child). His latest album (came out a few weeks ago) is fabulous.

[Archimago]

3 hours ago, mitchco said:

 

Unfortunately, for rock, pop, blues and most anything not classical, this will not be the case. As an ex-recording/mixing engineer, I can say that tracks are already compressed before even being recorded. Meaning, most mixing consoles have compressors/limiters on every channel strip. More often than not, compression and limiting is applied on the way to the multitrack recorder (analog or digital). More often than not, during mixdown, not only possible on individual tracks, but absolutely on the stereo master bus before hitting tape or digital. Finally, applied again during the mastering process.

 

Speaking with one of the engineers that did the mass majority of HDTracks transfers, more often than not, receives whatever is provided by the publishing company, and more often than not, it is a "safety" master is provided, which is an already compressed mixdown on tape or digital. The best the engineer can do is eq the master and add some stereo processing, depth, maybe add a bit of reverb...

 

Finally, virtually all music has passed through the device before it even reaches your ears. The simple reality is that we have all heard the sound of this device since 1967 – that’s 50 years ago folks. In fact, if you are listening to mainstream music right now, it is likely that you are also listening to the sound of this device. If you look at how the  Universal Audio 1176LN Peak Limiter works, it can substantially alter the music waveform in a way that there is no coming back, regardless of anything MQA can or cannot do.

 

Not to paint the bleakest picture :), the likelihood of uncompressed tracks being in vaults, is about 0. Given the ubiquity of "studios in a box" and access to everyone, proper professional recording, mixing and mastering is becoming (or has become) a lost art...

 

If the consumer audiophile industry wanted to really do something, demanding better recordings, mixes and masters is where we would see real audible sonic improvements. Sorry Arch, did not mean to single you out, just a PSA.

 

Dang Mitch. That's freaking depressing, man! 

 

[mitchco]

1 hour ago, Archimago said:

 

Dang Mitch. That's freaking depressing, man! 

 

 

Sorry Arch, probably revealed a little too much on how the sausage is made... I am illustrating how absolutely pointless MQA is, and for that matter most of HDTracks as well. Check this less than 2 minute video out:

 

 

Even with YouTube compression and over crappy laptop speakers, most folks can hear a clear audible difference with uncompressed versus compressed drums.

 

That's the point - MQA does nothing to solve what is easily the most audible problem and scourge of the consumer audiophile industry for the last 25 years, especially if one listens to anything other than classical or jazz on speciality labels.

 

Now had MQA structured a deal with the 3 majors to actually "remaster" from the original multi-track masters and guaranteed that Master Quality Authenticated meant a DR of 10 or greater then take my money!! I would even pay extra! Instead it is some woo about fixing obscure ADC problems that no-one seems to reliably hear even a perceptible audible difference, so ppfffttttt!!!

 

Oh yeah, wait till folks find out about the additional compression/limiting being applied to their streaming audio... ooops. 

[billg]

Nice clip Mitchco!

My belief is we need better masters and access to them. I don't think tinkering with the streaming process is where the effort is really needed. As Paul McGowan said, we can stream 4k HD on demand.

 

I've sometimes wondered whether it would be possible to "unmaster" recordings. If the engineer kept sufficient notes on the processes and equipment it might be possible to reverse engineer a "RAW" music file. Probably dreaming though.

[miguelito]

Prepare to get depressed...

 

https://www.uaudio.com/hardware/1176ln.html

[miguelito]

6 hours ago, mitchco said:

Unfortunately, for rock, pop, blues and most anything not classical, this will not be the case. As an ex-recording/mixing engineer, I can say that tracks are already compressed before even being recorded.

I have always wondered if some DSP could be applied to improve a massively compressed track. I understand that anything that clipped would be lost forever, but maybe those bits that are "compressed" but not clipped could be improved. I would pay very very good money for such a filter.

[mitchco]

Arch has a nice write-up here on doing just that:

http://archimago.blogspot.ca/2017/08/musings-increasing-dynamic-range-of.html

 

When I was recording, I did my best not to apply compression going to multi-track tape. A lot depends on the musicians being able to play whatever instrument/vocals consistently throughout a song. If you have even been in a band, even a school band from yesteryear, the dynamics go up and down with the flow of the emotion of the music being played. In a studio environment, most artists/bands are surprised how clear and dynamic they sound over headphones and sometimes have a hard time adjusting as small playing changes yield big dynamic changes in volume. So when going to tape and getting a good signal to noise ratio. some players require compression to be able to either not have enough SNR or too much and distort the tape.

 

Every multi-track tape usually had a track sheet that had notes on fader levels, eq, compression, effects, etc., per track, so one could relatively easily reproduce the mix or get real close. Further, most mixing consoles had automation where a number of these parameters are remembered by the computer.

 

That's what I am talking about with respect to "remastering". Not just taking the already mixed two track tape and messing with it. Rather going back to the original 2" multi-track tape or digital and now with modern DSP, meticulously restoring the sound quality to something that sounds like high fidelity and not having the life squashed out of the music due to too mush dynamic range compression. If that (ever) happened, I would be happy to repurchase my entire catalog...

 

[labjr]

18 hours ago, firedog said:

 

First, "our D to A converter" IS an FPGA. That's the context. 

Second, the wider context, if you understand their use of FPGA and their business model, is exactly what I said. Here's one  other quote by Paul M. on the subject (there are many similar ones):

 

 

I'm not going to bring you a full page of quotes of his on the subject. So yes, I interpreted what he said based on understanding the context and what  he's said over time about MQA. It doesn't make sense for a DAC business model based on regular SW updates that are essentially giving you a "new" DAC with changes to the filtering on a regular basis. 

I just quoted exactly what he said was the reason they don't have MQA in their DACs.