MQA, DSP and "sound quality"

[pirad]

Hello everybody, it’s my first post here. I am not an engineer,
so I am asking for your understanding and leniency.
To the subject matter:
I tested* MQA fairly early, both with the dedicated MQA DAC (Mytek)
and the Tidal app, which offers 96/24 decode. (Also www.2L.no is a good source of
MQA and hi-res files). I also used extensively Dirac room correction installed on
the same computer as my file playing apps. It could correct the decoded MQA.
After a year of experimenting I pretty much lost my interest both in MQA
and room correction. The main problem with MQA is still the very limited
titles offer. Also, I grew the suspicion that the differences between redbook and
MQA files can be attributed more to remastering than to the codec. And I mean
„differences”, not „sound quality”; as in "different" , not "better".
Caveat: staying on the objectivist ground it would be good to agree on
the definition of "Sound Quality" at the reproduction end.
My adventure with DSP started with building Linkwitz's LX521 loudspeakers which
use miniDSP platform (4x10HD). It continued with room correction applications,
mostly Dirac. The initial observations were positive. But with time and the continuous
tests with sound professionals, my final conclusion was: different, not necessarily better.

I could say now that my quest has been to reach the heart of the recording, not to
improve it at the reproduction end . I believe the digital manipulation should be taking
place only once: at the DAC converting the digital file into analogue. This conversion
should be of the highest quality. If any DSP is introduced, it should be before DAC.
Downstream of DSP/DAC I prefer: passive attenuation, a wire with gain and top transducers.
And last but not least: good room acoustics.

* My tests are of two kinds:
electroacoustic- with REW and DATS software, and occasional oscilloscope ,
auditory- with an AB switch box and the bunch of usual suspects: musicians,
producers (some Grammy winners ), critics and golden ears. My system is very
resolving and by true audiophiles can be labeled as "clinical clean". Pros like it.

 

[Fitzcaraldo215]

YMMV, meaning "your mileage may vary", as is a typical quote from most US auto makers about whether or not their cars achieve the mileage obtained in EPA testing. And, yes, of course, different implementation setups of audio systems in different rooms as measured or as heard by different listeners also vary, considerably I would say.

I am not sure I see a question in your post. However, I do agree with you that the application of DSP room correction should ideally and if possible be done entirely with signal in the digital domain ahead of one, single and final d to a conversion by the DAC. Not adhering to this and adding additional a-d/d-a steps might be acceptable to some, but I have not found it difficult myself to avoid doing that. I see no reason to. But, note that DSP need not be in the same box as the DAC itself. I use Dirac Live in a PC ahead of my DAC.

MQA is predicated on that single, final d-a idea, although there is a Tidal computer app that partially unfolds and decodes the MQA, making it suitable for use with a conventional DAC. However, its full decoding and processing is complex. As such, it has remained incompatible with DSP EQ prior to the DAC. However, a recent quote from Bob Stuart promises that this may change in the near future with some new implementations now being developed. However, it remains to be seen if those are compatible with existing EQ tools, such as Dirac. Allegedly, some Merdidian active speakers now apply both DSP EQ and MQA full decoding.

Kal Rubinson's Stereophile blog recently reviewed some new Mytek DACs. He compared his listening experiences with/without MQA vs. with/without DSP EQ (Dirac Live). While he found some things to like about MQA+no DSP EQ with 2L hirez recordings, he strongly preferred DSP Room EQ with no MQA. This squares entirely with my own independent, subjective listening reactions.

So, the choice for us at this point is quite simple. We absolutely would not be without DSP EQ, and MQA on its own simply does not offer enough sonic upside to overcome that. I will also say that in my listening to MQA, I found it offered some small but noticeable advantage generally in single blind, not DBT, testing of native hirez recordings in the unfolded MQA vs. the non-MQA CD version, both from PC files. However, that advantage largely disappeared in comparing native hirez recordings to their fully unfolded MQA equivalents, both of course compared without any DSP EQ.

I do think DSP EQ is here to stay. So far, I do not think MQA will rise above niche status, if it survives at all.

 

[amirm]

What was it about Dirac that you did not like?

With room equalization, you have full choice of the sound in your room. The target curve defines the ratios of bass to mid and treble for example. All of that can be tune and may be necessary as Room EQ can often give the impression of too little bass.

You may also want to experiment with setting a limit for Dirac correction. I have mine at 200 Hz for example so only bass frequencies are corrected.

 

[Rodney Gold]

I can tell you that I would never choose mqa over my trinnov or DIRAC room correction..EVER!!!

Amirm I have set up Dirac via minidsp and the trinnov side by side and can instantly switch between them by changing inputs .. level matched as best as possible..2 totally different sounds ..trinnov being markedly superior except in bass slam where I find Dirac has a slight edge...but I would take the trinnov any day..

 

[pirad]

I did not really pose a question in my above post. Just wanted to share my experiences with MQA and DSP.
I don't believe that "full" decoding of MQA is that much more processor heavy, I think it's purely a marketing decision.
I experimented with room correction quite a lot and found Dirac the best of the lot, for sure the easiest to implement.
My listening room has pretty good acoustics and the Dirac tests did not bring any significant improvements. Equalizing
frequency response and the phase adjustment did not really result in better sound. It was just different, a matter of taste.
I tried various limits with the sliders and and in the end would not go above 500Hz. The bass is not a problem with my
distributed subs system Earl Geddes style. Dirac only got confused and messed up things there. So in the end I found the
whole DRC business redundant in this case. However it can be helpful to some extent with poor rooms and lesser systems.
I used Dirac for some time in my living room on more typical speakers and there was an improvement. But even there
I dropped it after changing the system. One thing about Dirac is the "after correction" curve is simulated. It's worth
checking the results with REW or so, it's never so rosy as the simulation.

 

[Rodney Gold]

I have used sigtech, tact, lyngdorf room perfect , behringer, z-s7s, acourate drc , brutefir , krk ergo , meridian bass control , D speaker etc .. more DRC/DEQ devices I could list and I too find dirac the best of that bunch
I ran a swarm of 4 SVS sb13 subs with my older G1 models ala geddes/toole and used a minidsp 4x10 had to set them up and ran dirac to polish up the whole system , it worked real well

 

[amirm]

One thing about Dirac is the "after correction" curve is simulated. It's worth
checking the results with REW or so, it's never so rosy as the simulation.

Definitely. They all cheat this way.

 

[pirad]

I have used sigtech, tact, lyngdorf room perfect , behringer, z-s7s, acourate drc , brutefir , krk ergo , meridian bass control , D speaker etc .. more DRC/DEQ devices I could list and I too find dirac the best of that bunch
I ran a swarm of 4 SVS sb13 subs with my older G1 models ala geddes/toole and used a minidsp 4x10 had to set them up and ran dirac to polish up the whole system , it worked real well

How did you run the swarm of 4 subs? As mono or stereo and at what XO frequency? If stereo, how did you position them? If mono, how Dirac worked its magic?

 

[Sal1950]

I still cant' believe the audiophile community is so willing to lay down and let the distribution of music be taken over by Meridian and the big labels without a bigger fight. At first we were called paranoid but as time gone by even the pro-mqa crowd is beginning to admit that is the end game. Say good bye to lossless file distribution and accept all the problems that will be associated. I'm sure MQA will come along with some proprietary DSP scheme, all locked up tight in a box under their corporate umbrella control.
Who cares what MQA really sounds like? I don't know what the food tastes like at Cook County Jail either. But I know for sure I wouldn't want to live there.
BLAH!

"There are two issues here, and with their PR campaign MQA Ltd. has done a great job of focusing our attention on one—sound quality—and not the other: the hazards of a format monopoly. If MQA succeeds, I predict that it will lock in for a decade or two, or even longer. That will mean that all high-resolution files from the major labels during those decades will be formatted in MQA. No alternatives.
Some smaller distributors will undoubtedly continue to offer old-fashioned uncompressed masters for sale in a variety of formats, perhaps even via a boutique non-MQA streaming service. But with the major labels committed to MQA, such efforts will remain at the margins.
Which brings us back to sound quality. A possible format monopoly is all the more reason we should be absolutely sure that MQA is a format whose sound quality we can accept for the long term. But without the ability to even evaluate the format's compression scheme separate from its deblurring component, I don't believe that, over the long term, MQA is in the best interests of audiophiles. I just hope it's not too late.—Jon Iverson
Read more at https://www.stereophile.com/content/mqa-benefits-and-costs#0fHMokoJOhumizPT.99

 

[Fitzcaraldo215]

Definitely. They all cheat this way.

I don't think it is fair to call it cheating. The reality is that a separate test tone sweep must be done if the corrected response after calibration is desired. With Dirac, that would mean resituating the mike through the multiple positions it requires. That probably would not duplicate the positions originally used for the calibration and filter calculation, potentially creating a false discrepancy. I think they are wise to avoid the post-calibration measurement issue.

Doing a follow up measurement with REW may be useful, though it probably will never, as a single point measurement, completely agree with the predicted, spatially averaged response.

I don't think post calibration measurement is as simple as it seems, unless you do everything, pre- and post- from a single, unvarying mike point. But, is that a better approach in room acoustics than multi-point averaging? Maybe that is a learning experience for another day.

 

[amirm]

They need to be very clear that what is shown is a simulation, not real. None do that. It is not good to mislead the people that way.

 

[oivavoi]

How did you run the swarm of 4 subs? As mono or stereo and at what XO frequency? If stereo, how did you position them? If mono, how Dirac worked its magic?

Interesting posts, Pirad! Sounds like you have really explored this thoroughly. But back at you: what xo frequency do you use? Have you tried different frequencies? Mono, I guess? I'm looking into trying the multi sub approach myself, but not sure about how high they can/should be crossed without drawing attention to themselves. This also has some bearing on which subs to choose.

 

[Fitzcaraldo215]

They need to be very clear that what is shown is a simulation, not real. None do that. It is not good to mislead the people that way.

I was never in doubt, myself.

 

[pirad]

I don't think it is fair to call it cheating. The reality is that a separate test tone sweep must be done if the corrected response after calibration is desired. With Dirac, that would mean resituating the mike through the multiple positions it requires. That probably would not duplicate the positions originally used for the calibration and filter calculation, potentially creating a false discrepancy. I think they are wise to avoid the post-calibration measurement issue.

Doing a follow up measurement with REW may be useful, though it probably will never, as a single point measurement, completely agree with the predicted, spatially averaged response.

I don't think post calibration measurement is as simple as it seems, unless you do everything, pre- and post- from a single, unvarying mike point. But, is that a better approach in room acoustics than multi-point averaging? Maybe that is a learning experience for another day.

There is no need for accuracy in positioning of the 9 measurements.
The purpose is to average response around some area: armchair, sofa, theater rows. Usually I don’t even use a mic stand, just sit in the middle and extend my arms with the mic. Then do the same 9 easy moves with REW. It works.
Dirac looks for a compromise , otherwise you would need to hold your head in a vice when listening. You can actually measure real calibration results with Dirac, there is a workaround. But REW is better.

 

[pirad]

Interesting posts, Pirad! Sounds like you have really explored this thoroughly. But back at you: what xo frequency do you use? Have you tried different frequencies? Mono, I guess? I'm looking into trying the multi sub approach myself, but not sure about how high they can/should be crossed without drawing attention to themselves. This also has some bearing on which subs to choose.

I am using Audiokinesis Swarm. Normally it comes with one mono Dayton sub amp, I got two for flexibility. Theoretically up to 80Hz there is no need for stereo.Some people claim to be able to locate sounds from 60Hz. That’s interesting given that almost all music has lows mixed as mono. In my setup mains work full range and provide all stereo cues. The four subs are scattered along the walls in a random manner. I tested them up to 200Hz LPF both in mono and stereo. In my experience the sources cannot be located up to 130-150Hz so mono is enough. If your mains work full range, mind you. Distributed bass is amazing. No need for EQ in my opinion.

 

[Rodney Gold]

How did you run the swarm of 4 subs? As mono or stereo and at what XO frequency? If stereo, how did you position them? If mono, how Dirac worked its magic?

I had 2 pairs in stereo positioned midway each sidewall..my speakers I played full range and played the sub with a fairly low crossover .. maybe 40hz or so..I cant recall ...as my speakers were full range down to 30hz or so
I used duke le jeunes setup method and with the 4x10 varied delays , polarity , levels etc till I got the best room smoothing as per REW measurements ... That whole setup had as its head , a minidsp dirac box and I then ran dirac on the whole lot so it saw my speakers and integrated subs as a single system

 

[pirad]

I had 2 pairs in stereo positioned midway each sidewall..my speakers I played full range and played the sub with a fairly low crossover .. maybe 40hz or so..I cant recall ...as my speakers were full range down to 30hz or so
I used duke le jeunes setup method and with the 4x10 varied delays , polarity , levels etc till I got the best room smoothing as per REW measurements ... That whole setup had as its head , a minidsp dirac box and I then ran dirac on the whole lot so it saw my speakers and integrated subs as a single system

So you put two subs next to each other on each side wall? Or stacked them?
Effectively you did not distribute the bass sources, just two points?
Which Duke’s setup method you used? Can you link to Audiokinesis site?
I use his Swarm.

 

[mitchco]

They need to be very clear that what is shown is a simulation, not real. None do that. It is not good to mislead the people that way.

Well, I have tested both Acourate and Audiolense simulations, both in the frequency and time domains, which were "identical" using REW as the third party measurement software to validate. The "proof" is in the articles I wrote on CA for both products. I have extensively tested Acourate in my book, even to the point of measuring in REW, exporting the measurement, importing into Acourate and overlaying the simulation with the actual measurement - virtually identical. Same goes vice versa, and certainly within in the technical uncertainty of these acoustic measurement programs.

Sorry, but to say "none do that" is a disservice to your forum readership who are trying to learn maybe something new. You have not provided any scientific proof or objective measurements to back up your claim.

 

[Fitzcaraldo215]

There is no need for accuracy in positioning of the 9 measurements.
The purpose is to average response around some area: armchair, sofa, theater rows. Usually I don’t even use a mic stand, just sit in the middle and extend my arms with the mic. Then do the same 9 easy moves with REW. It works.
Dirac looks for a compromise , otherwise you would need to hold your head in a vice when listening. You can actually measure real calibration results with Dirac, there is a workaround. But REW is better.

Oh, no? I am afraid that I must disagree based on my own experience. Mike position affects measured acoustic response. Measured individual acoustic responses affect the spatial average. So, if mike positioning is not the same before and after calibration, I have no idea what in the hell you have got in the way of a comparison. The after measurement is quite pointless unless the mike positions duplicate the before positions fairly exactly.

No need for head in a vice thanks to spatial averaging. That is its purpose.

Sorry, but your casual, arm's length, hand held mike technique might very well make clear why you did not get satisfactory results with Dirac Live. It would not work with anything else, either. It seems you do not understand the technology and proper measurement. So, how can you expect the miracle of a misapplied, misunderstood tool making some positive impact on your sound?

 

[Kal Rubinson]

Oh, no? I am afraid that I must disagree based on my own experience. Mike position affects measured acoustic response. Measured individual acoustic responses affect the spatial average. So, if mike positioning is not the same before and after calibration, I have no idea what in the hell you have got in the way of a comparison. The after measurement is quite pointless unless the mike positions duplicate the before positions fairly exactly.

I've always felt this way, too, but I am beginning to look at it another way. If the purpose of the multiple spaced measurement mic placements is to supply enough information to create a relatively uniform response in the volume of space that they encompass. Now, if one measures anywhere in that space with any number of positions, the result will probably not replicate the "predicted" result but it should show a significant amelioration when A/B-ed with EQ bypassed. If it doesn't show that, the process failed.