Why you ask, because of pre ringing due to the steep filters. That’s why.
No they don’t, they sound bloody good in my experience…
1 Like
Yes they do, they sound bloody bad in my experience…
3 Likes
You know, you could all be right at the same time. You all have different ears and different prejudices and different neural pathways for processing auditory sensations.
2 Likes
Yeah but what you really need is the original source to compare to. That can be hard to pin down, including that Warner may have processed different files than the ones that were replaced.
Agreed. They are worse.
f.i. the Tracy Chapman’s album “Tracy Chapman” in MQA can be compared with the original PCM version as it is available on Qobuz.
(MQA lovers bring up that album all the time and “wow it’s sounds better” but they didn’t even take an effort to compare apparently; this is also the reason why I picked it).
The tracks are marked “LP version” on Qobuz; that are the original PCMs they used to make the MQA from. The track lengths expressed in number of samples are exactly the same for the MQAs as for those PCMs… (It’s VERY unlikely it’s a different master when all track lengths are the same).
So MQA-believers DON’T start saying “the comparison is invalid becaus it must be a different master”, IT IS NOT.
I ripped the 1st track in both MQA and PCM locally and played them bit perfectly through Foobar2000 to my MQA-DAC. Just to make sure software (like Roon/Audirvana/native Tidal app/native Qobuz app) are not using any DSP that might affect the results. So original PCM/MQA, bit perfect playback of those, using the same (full MQA) DAC, same amp, same speakers/headphones.
(and yes the MQA signaling is accepted on the DAC for the MQA if you wondered, emblem, etc)
The MQA of Tracy Chapman’s album sounded pretty good was my first impression; but that was BEFORE actually comparing. Then I started A/B-ing the MQA with the original PCM, and the fog disappeared: The PCMs are clearly much crisper, the MQA suddenly sounds like that same PCM is played through some piece of cardboard. I don’t know how to describe it but it seems blur + reverb is added. So nothing “deblurring” on the contrary. Some might like that reverbing but not me. If I want it I can install my old Equalizer between my DAC and the amp and press the “reverb” function on it.
Conclusion:
I hate to say this but I am one of those who spent 3000$ on an MQA compatible DAC.
I’m happy with the DAC, but NOT with MQA. If I knew this I would have spent my money on
even a better non-MQA DAC. I did comparisons for a month hoping to find a single MQA
that sounds better, and I didn’t, and now I cannot return the DAC to the vendor any more because the 14 days trial period has ended.
MQA adds nothing for me. I never heard it was better.
I did hear it was worse sometimes, especially for 16bit MQAs (including MQA-CD; those are actually even worse when they go >44.1kHz original sample rates as they miss the 3rd byte to store their folds)
6 Likes
Sounds like you’ve done some great testing.
I just want to clarify that you are comparing files ripped from an MQA CD? That there are multiple versions (like SACD hybrids) on the CD? (Apologies, as I have never played an MQA CD) And those were also compared with the versions on Tidal and Qobuz?
Anything that you can add about the equipment used for this would help others repeat your test and confirm your findings.
This is so important, because, as you say, many may spend thousands of dollars on a DAC that actually sounds worse with MQA.
1 Like
No I don’t even want to start ripping tracks from an MQA-CD those are even lower quality:
MQA-CD is a 16bit medium, it cannot carry 24bit MQAs.
Every MQA on Tidal that is 48kHz or more is 24bit.
For MQA-CD they
- strip off the lower byte of a 24bit MQA (losing the folded higher frequencies!) OR
- they compress the sound data even more when encoding
to make room to fold SOME of the high frequency data into those 16bits
(which results in even lower quality of course).
I don’t like MQA, and I certainly don’t like MQA-CD.
f.i. a 24b 352.8kHz recording => 24b 44.1kHZ FLAC on Tidal; but a 16b 44.1kHz on MQA-CD.
Huge difference. Dire Straits MQA-CD in 16b 44.1kHZ => 24b 352.8kHz ? No Thanks !
I compared ripped PCM streams (original PCM and MQAed PCM) from Tidal and Qobuz.
For Tracy Chapman both are 16b 44.1kHz (PCM and MQA)
2 Likes
In a DAW unless you change the sample rate and/or the start and end markers for the export, all exports will be of the exact same length. Sample count cannot be relied upon at all. If I were to revisit a master in my DAW to make some tweaks (compression, EQ or whatever) and re-export - it will still have an identical number of samples.
1 Like
Unfortunately you have a problem as so much music is only ever available as 16 bit audio due to the nature of early digital recordings… even Dire Straits I believe. If recordings come from analog transfers these would have been generated at 96/24 and above. Modern Digital is much better at 24/48 and higher for MQA. The sound here is just incredible…
The PCM’s extracted from the non-MQA flacs that are still on Tidal are exactly the same
as the PCM’s extracted from the flacs on Qobuz. And I said it is VERY unlikely (not impossible)… however, I will be glad if you can show me two different masters having the exact same track lengths. 
I understand MQA was usefull years ago for 24bit >48kHz recordings, because of bandwidth limitations, but that’s no longer an issue. Furthermore 95% of the music available online is 16bit 44.1k… then why convert those into same bitdepth, same sample rate MQA PCMs in the first place?
Now you need extra hardware to get only some of the original quality back.
MQA cannot create frequencies that were not recorded in the first place.
If a recording was 44.1k… 22.05kHz max freq was recorded… no folding, nothing.
No offense.
Simple, you remove the pre ringing and finish with a much nicer and faithful audio experience. It is even improved without a decoder.
Roon can remove pre-ringing with it’s minimum phase filter and upsampling. No need for end to end licensing and new hardware.
If you like the sound of pre and post ringing removed and don’t want to do it in software try an Ayre DAC, their listen mode filter does a great job. Again, no need for end to end licensing.
4 Likes
You are probably right about them being the same - the Qobuz version is an old Elektra with a product number, and the Tidal shows Warner-Elektra with a slightly modified product number.
What I get in comparing the first track of the Tidal MQA vs Qobuz is different to your interpretation and is very typical of what I hear in these older CDs. In the Qobuz PCM, the vocal at the center stands out while the surrounding instruments are both recessed and “smudged”, in other words not to the fore and not well resolved. In the MQA version all the instruments are better resolved, more forward, and thus more evenly balanced with the vocal, so the voice is no longer as prominent as it was. There is also a bit of timbral shift in the MQA towards a slightly more bassy sound but that could be either a result of the overall balance shift or just of the two different filter pathways for PCM vs MQA. My preference is for the MQA version because of its better overall resolution but you may prefer the PCM. I don’t get “cardboard” out of it though.
1 Like
Well look, the MQA path does corrections for both the A/D and D/A converters, including (I think) correction for modulator noise if needed, and the MQA downsample and upsample filters are on both sides of the chain. And there is careful control of the levels of dither noise (subtractive dither) and so on. Just substituting a single Roon or Ayre upsampling filter is not even close to the same thing.
Oddly, the downsampling and upsampling is no where to be seen in the MQA marketing material. Truly one of the many reasons why MQA is disappointing.
Ayre DACs are exceptional sounding, I highly recommend a person give them a listen, especially if you like what MQA does to music.
2 Likes
Try their JAES published papers which do a somewhat better job of describing the whole pathway. You really do have to spend a lot of time reading all of their materials unfortunately.
They may well be the same, however you cited sample count as evidence of that, and as someone who has been working in a DAW day in day out in my past I think that is meaningless as evidence.
Indeed, they never use the word downsampling or upsampling
but they do it all the time. In fact there are only few situations where real unfolds are happening.
I included a scheme with all kinds of things that happen when core decoding by software, rendering, decoding+rendering, playing on non MQA dacs and on MQA dacs (example is for DAC that is able to play 384kHZ). Apparently MQA upsamples everything and it is limited by hardware only (bob talks : mqa playback) Took me some time to find out.
I marked the unfolds as u1 (1st unfold) and u2 (2nd unfold)
Mind : This is valid for a DeltaSigma DAC that is capable of going to 384kHz.
You’ll see that ANY mqa plays as 352.8kHZ or 384kHz in the end
Even 44.1kHz MQAs… this is something most users don’t see.
2 Likes