XXHighEnd

Thought I would start this topic because the settings I found to improve MQA sound quality (SQ) have been mentioned in the topic "2.09 Sound Quality" but mostly that Topic is probably about Redbook 16/44.

I've been trying MQA but found sound quality to be rather poor compared to 16/44 and even conventional HiRes music, this with settings nominally same as my signature. I tried increasing Q1* and SQ gradually improved up to Q1=14*40. Still not great but at least some albums listenable. These setting however were not good for 16/44 which then sounded bland & uninteresting. Hence starting this topic specific to MQA.

Just started to try KS Mode:Special with more usual Q1=14*1 and this is promising & I will try some more. Peter tells me KS Mode:Special does not work with NOS1 DAC. My DAC is DDDAC1794 with cm6631a USB.

 Hope this encourages others to experiment with MQA a report back.

Cheers, Colin

Colin, so far it doesn't look like it. Personally I don't think it is a wise thing not to try it.
By now I have quite some MQA albums in my "Nice Stuff" Gallery, and the Demo Gallery also grows and grows. But it is a matter of finding them, and what your favorite genre is. And for example, I can not imagine that classical will improve from it (I never tried it, so this is just based onm experience).

Indeed it is so that NOS1 users can not use KS Special Modem, unless they have the very first NOS1 version (the one without USB interface). But if two are still round, I'd make up one already. But the other one regularly visits the forum, so he will ready this too ...
And just saying : KS Special Mode ever back was the best means of playback. That should still count for those not using a NOS1 DAC ...

Back to MQA - I guess it is difficult for people to even start thinking it might be for the better. I am positive it is, but it depends on the album. And let's be fair, once you found one for the better, you should be happy with it ?

Personally I would not approach this in the sense of "better sound quality". It is not like that and the dimension is a different one; it is merely about a "is working out better".
Gets into you better. Brings across messages better. Is more clear. Is NOT more refined.
And this latter would determine SQ. So when judged with that, we go wrong (IMO).

Is more firm. Has more punch. Is sounding more straight. Is sounding more honest. And yes, more real. And I hardly dare to say it : more how it ever was during the recording ... although I can't know that. But this is about the "more real" thing. You can sense it and you can see it.

One thing is quite crucial I'm afraid, and this is the NOS1 again. So if you'd know a but how MQA's filtering works (or destroys, name it as you like) then it actually requires the NOS1 including its filtering. Well, "requires" is a big word, but it suddenly brings out all the positives of MQA while any other DAC would not. But include the filtering, which lately for me is Arc Prediction, or Custom Filter - Highest (705600). I must experiment some more with this (it all goes way slowly), but the filtering is the key and the NOS1 does not filter anything. So it can all be tuned 100% in software.

For those who understand a bit more and now don't understand  :) : We are explicitly talking about "no MQA hardware" and use the decoder stage only, BUT replace the MQA hardware filtering (!) with our own in software. This is not because we lack the hardware (at least I myself don't) but because that part (named the Rendering part) is the bad boy of the MQA chain. So that's convenient.

Anyway, I really like people to try it, already because all the effort I put in it (all for free). So a little feedback would be nice, just like Colin just did. And if it is negative then this is OK because expected. But better would be if we could make something of it because it then will benefit ourselves.
Yes, it requires Tidal. But you can get a trial subscription for 30 days for free (the 19.99 subscription). And no matter you need to give your credit card, if you cancel then you're out of it (I tested that myself).

With settings I cant help because mine go all over the place, lately. Different settings matter in general but not specifically for MQA that I noticed.

Thanks !
Peter

I'll definitely do this when I return from my work travels next week.

Mani.

Hmm...

I have to say that I really like the sound of the few MQA albums I've tried so far.

I have an MQA and non-MQA version of an album that I'm sure is from the same master. With MQA decoding switched off in XX, the two versions of the album sound very different. The non-MQA version sounds softer and fuller. The MQA version sounds a bit on the thin side in comparison, but 'cleaner' with it. My preference is the MQA version with the MQA decoder in XX switched on.

This was not what I was expecting at all!

Mani.

Well I have to say that I am thoroughly enjoying exploring MQA-encoded music with the NOS1 at the moment. There tends to be a clarity to the sound that I really like. To my ears it's a bit like comparing AP to AI filtering - AP tends to lose the LF bloom but is simply more accurate-sounding and more life-like.

Guys, it's obvious to me that Peter must have spent a lot of time and effort sorting MQA out for XX. I think we owe it to him to at least try it and give our feedback.

And those with a NOS1 may be pleasantly surprised...

Mani.

Nice post Mani:- Well I have to say that I am thoroughly enjoying exploring MQA-encoded music with the NOS1 at the moment. There tends to be a clarity to the sound that I really like. To my ears it's a bit like comparing AP to AI filtering - AP tends to lose the LF bloom but is simply more accurate-sounding and more life-like.

Guys, it's obvious to me that Peter must have spent a lot of time and effort sorting MQA out for XX. I think we owe it to him to at least try it and give our feedback.

I have been working with MQA a few weeks now & I quite agree with your SQ comments & complements to Peter.
----------
My DAC is NOS but not NOS1(a) it needs different settings to get good SQ from MQA but these a fine with redbook as well.

Main change is to KS Mode = Special, *Q1 must = 1 to avoid sound brake up / distortion. Q4 =1 or 2 improves SQ but only for first tracks played, next tracks in playlist result in sound brake up. Q4=3 working nicely at moment.

If you have a DAC other than NOS1(a) give KS Mode = Special a try, of course you may need other Q setting than mine.

Colin


 

Hey Peter, in one of his interviews, Bob Stuart says:

Some people say, ‘I can do a better filter than MQA’, but I say, well actually, you can’t because the encoder and the decoder together make the perfect match… and it’s not the same on every recording. The filter that’s chosen on a 2L recording will be different from one on a jazz recording.

If so, does this mean that no matter which filter we choose in XX, it can never be optimally matched? Or that if it is, it is totally coincidental?

Just trying to understand why I generally like the sound of MQA played back through XX and the NOS1.

Mani.

Cheers Colin.

I thought the whole 'MQA deblurring' thing was just marketing speak. But my ears are telling me that there seems to be something actually to it.

I look forward to hearing more thoughts about the sound of MQA from others...

Mani.

Hi Mani,

The answer to this is too extensive, I think. For example, I take it that your quote is from quite recently ? If so, there's only one person saying that, as far as I know ... me.
So if we can start from there (and maybe a link ?), then I know what to talk about for starters.

And hey, thank you very much !
Peter

The quote is from an interview Bob gave in May this year at the Munich High End Show (I attended his MQA presentation the day before this interview): https://www.youtube.com/watch?v=3r_VRxcwODI

I find the interviewer quite frustrating - he keeps on interrupting whenever Bob seems to be getting into interesting stuff. By the sound of his accent the interviewer's Dutch, so we'll forgive him ;)

A few things that Bob said that caught my ear:

“What looks like high frequency on a FFT is actually temporal microstructure in the audio itself.”

“Our hearing is in fact incredibly acutely sensitive to microstructure in the mid-range and to typically quiet sounds.”

“The de-blurring process is like cleaning your window. If the window is dirty and we clean it, that is lossy… because we took the dirt away. But we made it better.”

“The passage of sound through the whole digital chain should be like sound travelling through air.”

Mani.

Maybe not. It is exactly what you say. What a most annoying guy.

So, Mani ...


Please keep in mind that this is strictly my own viewpoint (and finding / judgment / reasoning / truth) :

There's a small thinking error in this, and this is the fact that if the DAC isn't doing anything, there also does not need to be a match. And now we're right into the reason of the existence of the NOS1 : it is doing nothing BUT all needs to be done in software now.
And when talking to the MQA guys, I don't think they ever understood what I was telling them - explaining about the "perfect match" with what our XXHighEnd software does and with what our NOS1 D/A converter does not.

So it is one big trick which is applied now ...

1. Use the MQA encoded file (or stream);
2. Decode that to 24/96 (or any rate which is put in there for the first unfold and which can be 16/44 up to 24/96);
3. Apply a filter which suits the MQA deblurred data best;
4. Do nothing further (NOS DAC).

Of course it is about #3 in combination with #4 (which latter is "nothing" but totally crucial).
Of course it is also about what #3 is contentual, which should be a filter which not blurs the data again (no ringing).

So sure I can (at least theoretically) make a better filter, if I only use a DAC which does not destroy it;
The only match required is the match between the MQA's mastering console output (which is a computer program) and my software which I can make myself in ONE version only (but several *unlimited) to try) and which in itself does not need to match anything any more.
It would even be so that where we output to say 24/352.8 the least (assumed all DAC's will be doing that shortly), no filter in that DAC will have a chance of being destructive and we're set with any DAC. This is no different from people using XXHighEnd, some filter setting and like it for the better. So there too, the DAC's filter is overruled (at least to some extent and only when NOS *and* "analog filterless" it is by guarantee.

Solved.

There is way, way more of importance in the small paragraph above, because we can have a normal digital volume control, we can use DSP and we can, yep, have our own filtering. All of this is not really possible with MQA hardware involved, or it must have been made on a DAC (brand) per DAC basis. This is idiot of course and the sheer reason why I set it up like this. Well, sort of, because I wasn't given the opportunity to solve the problem of the requirement (I put to MQA) to control all this from software as we are used to, which requirement was not met, so it appeared.

Then a few more things from that interview :

Lossy means : can't reconstruct the original.
Here is where Mr Stuart has it all wrong (and the reason for the solution above) :

- A DAC is lossy. No, not when it is NOS/Filterless.
- A volume control is lossy. No, not the one in XXHighEnd.
- Filter in FAC chip is lossy. No, not when we have no filter.
- Each time you play it back it is different because of the filter. Funny, because my own invention/claim (9 years or so ago). But no, because you first must have a "rolling" filter for that but we use a genuine interpolator.

So you see, based on a few thinking errors we can achieve the same, or better. And if we'd see that virtually the NOS1 is our MQA renderer, then of course we made a specially matching filter for that (hey, I created Arc Prediction because any DAC hardware was in sight and only was in the drawing board - in NOS form).

So what needs to match ?
The deblurred file with a not-blurring DAC but WITH decent filter.
As I said, I told them, but they never understood (or they thought I wouldn't get there - also fine).

There's also this :
Nobody that I can see anywhere knows that the deblurring is in the file when we obtain it from Tidal, thus before decoding. I reasoned that out a 100 times because it couldn't be elsewhere. Now, I watched ONE video where Bob Stuart is interviewed (I really never looked at a single one before) and he mentions it in there at least two times.
"... deblurring in the studio ..."
"Without decoder MQA is already better than CD, because it has been deblurred already."

And this is exactly why I already know the benefit of "just 44.1 (or 48)" in MQA being better then CD or even more better because no possibility for fake upsampling etc., which was promised to me to be there, and which by now IS there (for a longer time, but not when I put the question back in March or April, 2017).

Try this one : https://listen.tidal.com/album/80896952
And promise yourself to only try this one after 20 minutes or so, if you can stop listening to the first one anyway : https://listen.tidal.com/album/80430588
(copy paste in the General field in Search).

One more new-ish hint : Select 8x Sampling Rate in XXHighEnd and the High Custom filter for 352800.
This sounds good if not better to me, but merely is related to MQA's I found failing at 16x (768000) but with a little hint that more could be going on as more fails. This is a different subject for now.

Have fun !
Peter

PS: Thank you very much, Mani.

Hi Peter,

Do you have any other MQA comparisons?  I really did not enjoy the Yello album and could not bring myself to listen to the non-MQA version.  That was the first MQA that I have tried.

Cheers,

Anthony

Hi Anthony,

This is personal of course (though hard to imagine that a single soul exists who could not be intrigued by how such an album like the Yello does not work out <- maybe it just doesn't in your system ?).

Anyway, see attached for a list of MQA's which I put in my Nice Stuff Gallery. This is not always so for the reason of having compared them with Redbook and found them better (which comparison is not really what I do) but because I listen to the album and experience an "impact" (a coming across) in a sense that is new for me for the album of concern. So for example, don't ask me how often I have played RAM (McCartney), but still the MQA version ended up in my Nice Stuff Gallery (which means the whole album should be of interest a next time). In my Demo folder I have a lot of individual MQA tracks which would be more distinct, but harder to show to you which tracks those are.

Btw never play MQA when the system is too cold as it will be too harsh (there's even logic in that).

The Neil Young's you see are not available any more (and they are easily "the best").

Cheers,
Peter

Peter, just delete these links if you think it's too risky to share files...

Here are two files from the same master (see waveforms and spectra below):

https://drive.google.com/open?id=1DkADQPvNQwGhpzy_9fW9jG7UJsuMEUQ2

https://drive.google.com/open?id=1asq0eZed6aiEUB-ULV2jRtDnIHqsr1kD

The music may not be to everyone's tastes, but it should be pretty inoffensive to the ear.

The clarity in the MQA file should be immediately apparent. There should also be a difference in the punchiness of the bass once the track gets going.

There is no question in my mind which one I prefer...

Mani.

PS. Peter, I've got the Yello albums you suggested - I'll take a listen in a short while.

This album with Diana Krall i think has a reference soudquality:

MQA: https://tidal.com/album/77644037

FLAC: https://tidal.com/album/73364268

I prefer the sound from MQA but it is not very easy to hear the difference in my setup.

I think the soundquality with XX 2.09, NOS1 and new settings is best ever

Thanks Peter, the winter is not so long with this SQ  :)

Tore

I've tried listening to MQA albums, and so far still not sure about how to evaluate, or optimize settings.  I listened to both the 16/44 and MQA version of The Eagles "The Eagles"; Witchy Woman, etc.  With the same settings (my signature), I find the 16/44 version fuller, more dynamic, and engaging.  The MQA files show as 24/48, but since there is no information on the original recording it's hard to tell what original resolution is being used for the MQA version.  Will keep experimenting. 

Hey Ramesh,

We must think a bit differebtly here; the original resolution will interest you a hoot, in this (Eagles) case because back at the times I was 15 ... what resolution. Right ?
What counts for everyone beginning with this is whether you have the decoding working, and you have not. See the screenshot below;

In the top right you see your "Witchy Woman" album. It is labelled "192" which means that in decoded fashion you will (in software) receive it as 24/96. Not more, because software only "unfolds" one time, and this is to 24/96. This is good because we don't want more. So we do NOT want the indicated 192 as it is us ourselves making it (even) 768. So in goes 96, out goes 768.
Now compare with the Ed Sheeran album (Divide) and see it labelled "44". *This* one will be received as 24/44.1 and there is no first unfold. But you should also not use these to test whether your decoding works. So for that use "88"'s the least and they should output 88 (or 96 when 96 based).

I am sure this is not easy to bring across as it also involves a "malfunctioning" PC on your side. Upgrade it to 2.09 or otherwise you can't see any of this. AND :
If you did and these labels do not show, then something else is amiss which could be harder to solve. But first things first ...
Btw, after upgrading to 2.09, ditch the old "Preps" from your TT (and further) folder. Otherwise their status could be fixed and you don't want that.

Regards,
Peter

PS: Coincidentally all of the Eagles albums were my first MQA's which I unconditionally liked over the normal Redbook versions. So all you need to do is somehow agree with me. And when you do, you're set. Haha.

For me, this setting takes redbook past MQA!

Mani.