Isn't mqa just another kind of mp3?

Geepstar

New Member
Nov 9, 2020
1
0
1
49
I am just getting familiar with Tidal and mqa. And when I read that mqa is a way of packaging studio quality by removing silences I think... I've heard this before. But, after getting me a ifi zen and hearing the difference, I have come to the conclusion. No. MQA is not like MP3. It brings quite a lot of detail into music. High res. And I haven't even got a very expensive setup. My computer is connected with the ifi zen, and I have plugged that into my Zoom livetrak L-12. The livetrak feeds my Hedd Type07's and my M-Audio M3-8's and Yamaha active subwoofer. Without the zen the music sounded quite sharp allready, but after plugging the zen in.. I entered another galaxy. And I know I will have to maintain this hobby, because my absolute hearing demands more. ;-)
 
Last edited:

Ian B

Well-Known Member
Oct 19, 2020
191
116
98
41
MQA works in a very different way from MP3/MP4. It is technically a lossy format, but with HD files it compresses stuff above what you can hear aggressively, but leaves the audio band intact, and hides the ultra-high frequencies under the noise floor until they are decoded. The biggest difference is it has an algorithm that improves the timing accuracy of the audio beyond standard PCM, so it sounds different. Depending on your taste it might sound better. I find it is a bit of a mixed bag.
 

Yahooboy

Well-Known Member
Jul 23, 2013
6
3
298
I am just getting familiar with Tidal and mqa. And when I read that mqa is a way of packaging studio quality by removing silences I think... I've heard this before. But, after getting me a ifi zen and hearing the difference, I have come to the conclusion. No. MQA is not like MP3. It brings quite a lot of detail into music. High res. And I haven't even got a very expensive setup. My computer is connected with the ifi zen, and I have plugged that into my Zoom livetrak L-12. The livetrak feeds my Hedd Type07's and my M-Audio M3-8's and Yamaha active subwoofer. Without the zen the music sounded quite sharp allready, but after plugging the zen in.. I entered another galaxy. And I know I will have to maintain this hobby, because my absolute hearing demands more. ;-)
When You look a little deeper You will discover that it is indeed MP3 v2
Look at the measurements the exact same master. I think these graphs speak for themselves

Blue one is the Original DXD (24/352,8)
Red is MQA decoded (for instance Tidal app - 17/44-48 upsampled once)
Green is MQA renderer (depending on the capabilities of the DAC 17/44-48 upsampled to whatever the embedded metadata states)
mqa-mutilation-freq-domain.png.d829f4559b931fe938d79a6a13cd4a58.png
 
  • Like
Reactions: Eichenbaum

rbbert

Well-Known Member
Dec 12, 2010
3,820
239
1,000
Reno, NV
The big difference is that MP3 encoders and decoders are open source and free; MQA is proprietary
 

Ian B

Well-Known Member
Oct 19, 2020
191
116
98
41
When You look a little deeper You will discover that it is indeed MP3 v2
Look at the measurements the exact same master. I think these graphs speak for themselves

Blue one is the Original DXD (24/352,8)
Red is MQA decoded (for instance Tidal app - 17/44-48 upsampled once)
Green is MQA renderer (depending on the capabilities of the DAC 17/44-48 upsampled to whatever the embedded metadata states)
View attachment 71905
If you notice the frequency indications on the X axis you can see that MQA and PCM are close to identical in the range we can actually hear, below 20khz. They don't really deviate until around 40khz, and the earliest difference visible is 25khz at below -120db. Most DACs aren't even capable of resolving below -120db, and virtually no humans can hear 25khz. Put simply, none of the obvious differences on that graph are audible in terms of volume or frequency range.

I would say that the MQA line is what you would expect to see with 24 bit audio that is progressively compressed above the audible frequency range.

With mp3 you would have a maximum dynamic range below 96 db and a sharp cutoff between 15 and 18khz resulting in alias distortion in the audible range.

Now, having listened to MQA, I do find that it is indeed lossy in some ways. But the process is quite unique, and the sound character of MQA is unlike any other format/compression scheme.
 
Last edited:

Yahooboy

Well-Known Member
Jul 23, 2013
6
3
298
If you notice the frequency indications on the X axis you can see that MQA and PCM are close to identical in the range we can actually hear, below 20khz. They don't really deviate until around 40khz, and the earliest difference visible is 25khz at below -120db. Most DACs aren't even capable of resolving below -120db, and virtually no humans can hear 25khz. Put simply, none of the obvious differences on that graph are audible in terms of volume or frequency range.

I would say that the MQA line is what you would expect to see with 24 bit audio that is progressively compressed above the audible frequency range.

With mp3 you would have a maximum dynamic range below 96 db and a sharp cutoff between 15 and 18khz resulting in alias distortion in the audible range.

Now, having listened to MQA, I do find that it is indeed lossy in some ways. But the process is quite unique, and the sound character of MQA is unlike any other format/compression scheme.
Yup, the only lossy 17/44-48 "codec" posing as a lossless 24/up to 384 format. (Don't even mention the latest "upgrades to Tidal 16/44 MQA - actually even worse around actual 13 resolution)
 

Ian B

Well-Known Member
Oct 19, 2020
191
116
98
41
Yup, the only lossy 17/44-48 "codec" posing as a lossless 24/up to 384 format. (Don't even mention the latest "upgrades to Tidal 16/44 MQA - actually even worse around actual 13 resolution)
I don't really understand this. I'm not saying that MQA is identical to the PCM source, or that it doesn't lose something, but looking at that graph it clearly can manage 24 bit performance and doesn't show added distortion until what would be 20-21 bits. Also, because it isn't that same in the ultrasonic region doesn't mean it is equivalent to 44.1 or 48khz. Look at DSD, from this type of metric it actually performs worse than MQA, even though it sounds better.

My understanding is that they wanted to make a compression scheme that retained the sound of high bandwidth, high sample rate audio, while throwing out or heavily compressing the inaudible data, and it seems like they generally succeeded at that. They also created a unique filter kernel that corrects the ringing and energy smear of the source, which also seems to work. I'm not sure why it sounds darker on the top, or lacking in detail to my ears, but it doesn't have the obvious artifacts of mp3, and they did seem to preserve and enhance the characteristics of high sample rates in a smaller package.
 

Heckyman

Well-Known Member
Oct 22, 2020
12
8
68
52
Hi, essentially mp3 sounds like the original CD or hi-res master. MQA applies proprietary processing (and adds a hardware dependency) that intentionally alters the sound in such a way as to subjectively appeal to some people.

Personally MQA sounds worse than CD quality.
 
Last edited:
  • Like
Reactions: montesquieu

Yahooboy

Well-Known Member
Jul 23, 2013
6
3
298
I don't really understand this. I'm not saying that MQA is identical to the PCM source, or that it doesn't lose something, but looking at that graph it clearly can manage 24 bit performance and doesn't show added distortion until what would be 20-21 bits. Also, because it isn't that same in the ultrasonic region doesn't mean it is equivalent to 44.1 or 48khz. Look at DSD, from this type of metric it actually performs worse than MQA, even though it sounds better.

My understanding is that they wanted to make a compression scheme that retained the sound of high bandwidth, high sample rate audio, while throwing out or heavily compressing the inaudible data, and it seems like they generally succeeded at that. They also created a unique filter kernel that corrects the ringing and energy smear of the source, which also seems to work. I'm not sure why it sounds darker on the top, or lacking in detail to my ears, but it doesn't have the obvious artifacts of mp3, and they did seem to preserve and enhance the characteristics of high sample rates in a smaller package.
It isn't apparent from this one measurement. But when You look at a lot of different measurements. You will see that each and everyone of the MQA files have the same frequency responce above 22-24 kHz, with no relation to the original file. If You zoom in on this picture You will also note the strange phenomenon that the two MQA streams doesn't even match inside their covered frequency ranges.

Measurements where there is a gap in the frequency response also shows the same (high frequencies with no relation to the original signal) Bruno-Mars-Imaging.png
 

Mikem53

Well-Known Member
Oct 1, 2020
662
581
105
Hi, essentially mp3 sounds like the original CD or hi-res master. MQA applies proprietary processing (and adds a hardware dependency) that alters the sound in such a way as to subjectively appeal to some people.

Personally MQA sounds worse than CD quality

HDCD, SACD, DAT formats didn’t survive.. can MQA and DSD can be converted to PCM so it’s compatible with current DACs .
 

Al M.

VIP/Donor
Sep 10, 2013
8,684
4,473
963
Greater Boston
MQA works in a very different way from MP3/MP4. It is technically a lossy format, but with HD files it compresses stuff above what you can hear aggressively, but leaves the audio band intact, and hides the ultra-high frequencies under the noise floor until they are decoded. The biggest difference is it has an algorithm that improves the timing accuracy of the audio beyond standard PCM, so it sounds different. Depending on your taste it might sound better. I find it is a bit of a mixed bag.

The theoretical enhancement of timing accuracy, with removing the "blurring" of transients, is debatable. In practical terms it would require knowing the exact analog-to-digital converter(s) (ADCs) used for recording, so that their timing errors can be corrected, but there is no evidence that this knowledge is available and/or acted upon. In fact, it seems that they perform bulk conversion of files, without any real knowledge of the ADCs and recording process for each individual file. This negates the whole "correcting" purpose of MQA.

Also, MQA may actually worsen transient smearing:

To my ears, standard PCM digital, including Redbook, has no obvious transient problem, as judged against unamplified live music and high-quality vinyl rigs of my friends. Of course this assessment, apart from personal perception, also depends on the quality of your digital and your system as a whole. Adequate transient resolution in playback is not easy to achieve.

One reason why MQA may sound different could be aliasing problems:
 
  • Like
Reactions: rbbert

Ian B

Well-Known Member
Oct 19, 2020
191
116
98
41
The theoretical enhancement of timing accuracy, with removing the "blurring" of transients, is debatable. In practical terms it would require knowing the exact analog-to-digital converter(s) (ADCs) used for recording, so that their timing errors can be corrected, but there is no evidence that this knowledge is available and/or acted upon. In fact, it seems that they perform bulk conversion of files, without any real knowledge of the ADCs and recording process for each individual file. This negates the whole "correcting" purpose of MQA.

Also, MQA may actually worsen transient smearing:

To my ears, standard PCM digital, including Redbook, has no obvious transient problem, as judged against unamplified live music and high-quality vinyl rigs of my friends. Of course this assessment, apart from personal perception, also depends on the quality of your digital and your system as a whole. Adequate transient resolution in playback is not easy to achieve.

One reason why MQA may sound different could be aliasing problems:
Sometimes it's really a matter of language, for example what you interpret "transient problem" to mean. I come from a lot of experience with tape and DSD and to my ears the subtle timing of Redbook is pretty atrocious. If you listen to the same recording in DSD the attack and release are much faster and more effortless, the envelope of sound is more fluid and natural, and the spacial localization is vastly better. PCM doesn't even approach this until 24/192 and above, and usually requires some unique filter designs.

Now, I'm trying to be clear: I don't prefer MQA, and after a few samples I scarcely listen to it. To me the problem with it is that it lacks precision, and the top end is dark compared to regular PCM. I can't speak for how they process the majority of releases for Tidal etc, but I do know that they have a database of mastering converters in order to apply the processing. This is more complicated than it seems though, because that likely does not take into account the tracking and mixing converters used, so I would take some of it with a grain of salt.

But I do have to put the timing processing into a separate bucket from their data compression and rights protection schemes. Their filter kernel is actually a pretty elegant design, and from the examples I've heard it did get PCM sounding a bit closer to DSD in terms of fluidity and speed. If you know what aliasing sounds like, that's not the major element in the sound of MQA, although there probably is a little bit in there. It's not a big problem to me as I usually prefer DACs with slow rolloff filters, and lots of us like the sound of NOS or non-digital filtering DACs which all have aliasing.

A few years ago 2L re-released a Mozart album, one of the first DXD recordings done, as an MQA "Remix". This is a PCM file, but what they did was apply the MQA processing to the individual instrument recordings. So, no data compression or anything, just the timing/filter algorithm. I also happen to have the original DXD recording, and the MQA remix one sounds better. It seems like a pretty useful technology, and does help with what seems to be a real problem with PCM recordings. If there was a way to record, mix and master in a new non-Nyquist digital format I think there would be real sonic gains. Unfortunately, it was invented by Meridian who have been all about right protection schemes from the days of the MLP on the old DVD-Audio discs (which, 15 years later I still can't rip), and is tied with data compression.
 
  • Like
Reactions: Heckyman

Al M.

VIP/Donor
Sep 10, 2013
8,684
4,473
963
Greater Boston
Sometimes it's really a matter of language, for example what you interpret "transient problem" to mean. I come from a lot of experience with tape and DSD and to my ears the subtle timing of Redbook is pretty atrocious. If you listen to the same recording in DSD the attack and release are much faster and more effortless, the envelope of sound is more fluid and natural, and the spacial localization is vastly better. PCM doesn't even approach this until 24/192 and above, and usually requires some unique filter designs.

Now, I'm trying to be clear: I don't prefer MQA, and after a few samples I scarcely listen to it. To me the problem with it is that it lacks precision, and the top end is dark compared to regular PCM. I can't speak for how they process the majority of releases for Tidal etc, but I do know that they have a database of mastering converters in order to apply the processing. This is more complicated than it seems though, because that likely does not take into account the tracking and mixing converters used, so I would take some of it with a grain of salt.

But I do have to put the timing processing into a separate bucket from their data compression and rights protection schemes. Their filter kernel is actually a pretty elegant design, and from the examples I've heard it did get PCM sounding a bit closer to DSD in terms of fluidity and speed. If you know what aliasing sounds like, that's not the major element in the sound of MQA, although there probably is a little bit in there. It's not a big problem to me as I usually prefer DACs with slow rolloff filters, and lots of us like the sound of NOS or non-digital filtering DACs which all have aliasing.

A few years ago 2L re-released a Mozart album, one of the first DXD recordings done, as an MQA "Remix". This is a PCM file, but what they did was apply the MQA processing to the individual instrument recordings. So, no data compression or anything, just the timing/filter algorithm. I also happen to have the original DXD recording, and the MQA remix one sounds better. It seems like a pretty useful technology, and does help with what seems to be a real problem with PCM recordings. If there was a way to record, mix and master in a new non-Nyquist digital format I think there would be real sonic gains. Unfortunately, it was invented by Meridian who have been all about right protection schemes from the days of the MLP on the old DVD-Audio discs (which, 15 years later I still can't rip), and is tied with data compression.

That may depend on the digital system. I have a Schiit Yggdrasil 2 DAC, which has a digital filter optimized for timing. It is proprietary, thus not off-the-shelf, and runs on high-performance SHARC DSPs.

A friend and I once compared on his high-resolution system an originally analog sourced recording of a string quartet on both his vinyl rig and from a 16/44 file on the same DAC that I have. Both sounded very similar, and transients as well as flow showed hardly any difference, if any.

Attacks and release on piano can be incredibly fast and dynamic from my Redbook digital replay, very noticeable on modern material that goes to extremes on that (Stockhausen, Wolfgang Rihm), and which I have also heard live, so I have a reference. On the other end of the spectrum, very delicate transients, like some where the bow almost bounces like a feather on the strings in string quartet playing, are also portrayed convincingly. My vinyl loving friends have always admired my soundstage, so it is on a high level as well.

One vinyl loving friend has also recently commented on the great flow and 'atmosphere' of music in my system.

Thus if you say:
"...to my ears the subtle timing of Redbook is pretty atrocious. If you listen to the same recording in DSD the attack and release are much faster and more effortless, the envelope of sound is more fluid and natural, and the spacial localization is vastly better",

I am left wondering about the quality of your digital playback.
 

rbbert

Well-Known Member
Dec 12, 2010
3,820
239
1,000
Reno, NV
the old DVD-Audio discs (which, 15 years later I still can't rip), and is tied with data compression.
Dolby TrueHD and DVD-A (both use MLP) are routinely ripped into stereo and 5.1 PCM (as is DTS-HD) and have been for years. One does have to buy the software, AFAIK none of it is freeware (as it is for CD and ordinary DVD rips)

There are measurements showing that DSD records transients better than even hi-res PCM, although 24/192 comes close (DXD matches it, according to Grimm)
 
  • Like
Reactions: Ian B

Ian B

Well-Known Member
Oct 19, 2020
191
116
98
41
That may depend on the digital system. I have a Schiit Yggdrasil 2 DAC, which has a digital filter optimized for timing. It is proprietary, thus not off-the-shelf, and runs on high-performance SHARC DSPs.

A friend and I once compared on his high-resolution system an originally analog sourced recording of a string quartet on both his vinyl rig and from a 16/44 file on the same DAC that I have. Both sounded very similar, and transients as well as flow showed hardly any difference, if any.

Attacks and release on piano can be incredibly fast and dynamic from my Redbook digital replay, very noticeable on modern material that goes to extremes on that (Stockhausen, Wolfgang Rihm), and which I have also heard live, so I have a reference. On the other end of the spectrum, very delicate transients, like some where the bow almost bounces like a feather on the strings in string quartet playing, are also portrayed convincingly. My vinyl loving friends have always admired my soundstage, so it is on a high level as well.

One vinyl loving friend has also recently commented on the great flow and 'atmosphere' of music in my system.

Thus if you say:
"...to my ears the subtle timing of Redbook is pretty atrocious. If you listen to the same recording in DSD the attack and release are much faster and more effortless, the envelope of sound is more fluid and natural, and the spacial localization is vastly better",

I am left wondering about the quality of your digital playback.
I'm glad you are enjoying the sound of your digital system. There is no problem with that. That said, what I'm talking about is an issue that has been know for a long time. The filter-related timing issues with PCM are known issues and are the reason why these things exist: higher PCM sample rates (including DXD), DSD, minimum-phase filters, slow rolloff filters, apodizing filters, adaptive filters, upsampling, and non-digital filtering DACs. They are all trying to tackle the same problem as MQA. If you get a chance with other DACs, experiment with these features/formats and see how they change leading edge transients, especially on things like tom fills.

My current digital system uses a special transient correcting adaptive filter and upsamples everything to 1 bit 45mhz and is about as good as I've heard, although I like some of the more rough non-digital NOS schemes as well. It still sounds best with DSD, and a little below that DXD. At a certain point there is only so much DAC engineering can do to address a difficulty with the recording format itself.
 

Al M.

VIP/Donor
Sep 10, 2013
8,684
4,473
963
Greater Boston
I'm glad you are enjoying the sound of your digital system. There is no problem with that. That said, what I'm talking about is an issue that has been know for a long time. The filter-related timing issues with PCM are known issues and are the reason why these things exist: higher PCM sample rates (including DXD), DSD, minimum-phase filters, slow rolloff filters, apodizing filters, adaptive filters, upsampling, and non-digital filtering DACs. They are all trying to tackle the same problem as MQA. If you get a chance with other DACs, experiment with these features/formats and see how they change leading edge transients, especially on things like tom fills.

My current digital system uses a special transient correcting adaptive filter and upsamples everything to 1 bit 45mhz and is about as good as I've heard, although I like some of the more rough non-digital NOS schemes as well. It still sounds best with DSD, and a little below that DXD. At a certain point there is only so much DAC engineering can do to address a difficulty with the recording format itself.

Yes, the theoretical issue has been debated for a long time, but it turns out that with better decoding even standard Redbook digital is capable of vastly more than people had been giving it credit for during several decades, including myself.

Obviously we can argue all day long, but we cannot transfer our own experiences of what we have heard to one another.

Once the pandemic is over and you find yourself one day in Massachusetts, you are welcome to visit me and hear my system. Then you can experience for yourself what I am talking about. And yes, the system also excels on drums, as several audiophile friends have heard -- including one who is a drummer himself.
 

rbbert

Well-Known Member
Dec 12, 2010
3,820
239
1,000
Reno, NV
No matter how optimized the playback, deficits in recording can't really be overcome, they can only be ameliorated. If MQA filtering has the potential to do that (which it originally appeared was the case, although the subsequent batch processing for most MQA titles gave the lie to that), then listeners and consumers would all benefit if it were separated from the lossy compression and further developed.
 

Ian B

Well-Known Member
Oct 19, 2020
191
116
98
41
No matter how optimized the playback, deficits in recording can't really be overcome, they can only be ameliorated. If MQA filtering has the potential to do that (which it originally appeared was the case, although the subsequent batch processing for most MQA titles gave the lie to that), then listeners and consumers would all benefit if it were separated from the lossy compression and further developed.
Agreed.
 

treitz3

Super Moderator
Staff member
Dec 25, 2011
5,459
961
1,290
The tube lair in beautiful Rock Hill, SC
Hello all and good evening to you. I hope all is well with you and yours.

This will be my one and only statement regarding MQA because in reality? I simply do not care of the how, what, where, when and why. All I care about is the end result as to what hits my ears.

The end result is something that is not worthy (to me) of putting on my big rig. It just isn't worth my time. To me, it's one of those, "been there, done that" moments.

With that said, I wish all of you a happy holiday season.

Tom
 

About us

  • What’s Best Forum is THE forum for high end audio, product reviews, advice and sharing experiences on the best of everything else. This is THE place where audiophiles and audio companies discuss vintage, contemporary and new audio products, music servers, music streamers, computer audio, digital-to-analog converters, turntables, phono stages, cartridges, reel-to-reel tape machines, speakers, headphones and tube and solid-state amplification. Founded in 2010 What’s Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals, we enjoy learning about new things and meeting new people, and participating in spirited debates.

Quick Navigation

User Menu

Steve Williams
Site Founder | Site Owner | Administrator
Ron Resnick
Site Co-Owner | Administrator
Julian (The Fixer)
Website Build | Marketing Managersing