MQA Timing Blur question

[hudesigns]

Hi Peter,

I have been reading to gain some knowledge regarding MQA decoding. I am intuitively attracted to the idea about MQA's ability to address "timing blur" issue in DA conversion in general.

I am wondering if this correcting "timing blur" process happens in software decoding"? If so, we are already listening to the "time-corrected" playback via XXHE with MQA encoded albums.

Or, is it that MQA core only carries additional info of how to correct "timing blur" and then provides MQA DAC specific instructions to achieve the final "time corrected" resolution in a particular hardware?

It is interesting to me to do an AB comparison regarding the specific "time blur" issue.

Best,
Zheng

[PeterSt]

Dear Zheng,

I am intuitively attracted to the idea about MQA's ability to address "timing blur" issue in DA conversion in general.

This is (has been) for me no different and the only reason to exploit it. So Yes, assumed it can work for real. And there we have right away the undoable part : how to prove it.
So all we can do is listen and agree over something which is very hard to agree upon in the base (but we still could).

Side note for you and others : I don't think there is a single "Remaster" as such around anywhere that I like over the original. There also is hardly Hires around that I like better than the Redbook version, but this is mainly because this Hires in native version hardly exists (DVD 5.1 a lot but making Redbook out of that will be a mess).
Still MQA almost always is OK for me and this thus while a Remaster and such is not OK at all. Add to this that MQA for sure is/sounds different and there it starts to intrigue. Also, it has to be something "a kind of inherently good" or else I would not be able to like it in the first place.

I am wondering if this correcting "timing blur" process happens in software decoding"?

No. It happens in the ENcoding of the file. Thus, we have an album with its tracks (which are files as you and me know them) and they are processed through a software and this applies the "Deblurring" (mind you, FWIW).

If so, we are already listening to the "time-corrected" playback via XXHE with MQA encoded albums.

That would be correct anyway, Yes. So what we obtain from Tidal already is "deblurred" and all it really requires is the decoding of MQA. Let's say that only then the real "deblur" comes forward but which is hard to prove because playing without decoding (MQA) is a mess again (sounds super bad to me) and so can't be judged for any "deblur" already active in there. But we don't cae because we decode anyway (in XXHighEnd).
What's crucial in your question(s) above is that we do NOT need the MQA DAC to apply this "debluring" because it already happened. All the MQA DAC adds is a proprietary set of digital filters which any self respecting DAC has inside its cabinet. Here though it is presented as the 2nd and 3rd unfold which is only a nice story. What happens for real is upsampling through those filters and let's say that we can do better with out own filters OR we just like those in the MQA DAC. But this is so with any DAC, actually. Of course unless the DAC has no digital (and analog) filter at all - then it is NOS and it is up to you to again apply filters in software or just nothing because you like NOS as is.

Or, is it that MQA core only carries additional info of how to correct "timing blur" and then provides MQA DAC specific instructions to achieve the final "time corrected" resolution in a particular hardware?

This is the vague part (or at least it is for me and it can be investigated further) because it is indirectly claimed that this is so. With "indirectly" I mean that it has been found that the files may pass on different filter data (which to set out of a possible 32 or whatever) per album (or track for that matter). Now watch out because this is about a kind of opposite of time blur. Think lof this sequence :

1. File is "deblurred" and now is as transient as can be (and too much of it); This is not done the same for each recording as it can't be the same (some automatic analysis should be going on in the encoding software);
2. Each DAC is different for it's native D/A section (without filtering which will be overruled by the MQA filtering) and now the BLURRING is applied best matching the DAC of concern.

Blurring : This is the filtering again which is required. So it is not explicit blurring as such - it is re-applying the filtering which was taken out to some extent at the "deblurring" process.

Remember, this latter part is highly speculative as nobody really knows. It is about logical conclusions of what is and can not be.

It is interesting to me to do an AB comparison regarding the specific "time blur" issue.

I say you can't execute this really.

a. When you do so you need that MQA DAC and you first need to like it for all its aspects (chance is small unless your current DAC can be upgraded with an MQA module and then still ...
b. You must be able to switch off the last (Rendering) stage and this eliminates the filtering all together and you have theoretical rubbish (but it would be no different from playing through the Tidal Desktop player directly to your current DAC).

So you can't compare in a good fashion.
What you can do is A. find a native album (not MQA encoded) of the very same master as an MQA version you have (a toughest job to begin with) and compare the two through the MQA DAC which fully decodes (thus also applies the Rendering stage).

Done ?
No. Because tell me why you are using XXHighEnd ?
Only when you currently use that with preamp and no digital attenuation and also no filtering active and thus only use it for its "environmental" (OS) influence, you are good to go and let that part of the chain 100% in-tact. But if you use it slightly differently then then you need to set all to what I just told anyway and things will sound different because of that. If you like the net result - OK. But of not ...
So you see, it is a LOT of hoopla for the unknown result at this moment WHILE you can apply the most crucial part already : the MQA decoding (includes the "deblur") *and* apply the filter of your liking (within limits of course because they must be available in the playback oftware of your choice).

Best regards,
Peter

[Rmalits]

Hi there,

that's a very interesting topic for me. I enjoy it more and more listening to good MQA albums with XXHE and my NOS1a-3G. For me the majority of the MQA albums sound significantly better (bigger sound stage, clearer highs and more precise bass) than the redbook and highres versions. I believe to receive well the better timing and/or debluring of MQA in my ears, but that's subjective for sure.

Ok, it depends a lot on the mastering the tracks come from.
In this way I think, very new albums are good for A-B testing, like e.g. Melody Gardot's "Live In Europe" what was released 2018.
We did a-b-c testing of the track "03-Rain" with three versions:
a) redbook from Tidal
b) highres 48/24 downloaded from Qobus
c) MQA 48/24 from Tidal

These versions are most likely from the same mastering, as the track lengths are exactly the same.

The result: c) clearly outperforms a) and b).
There is not so much difference between a) and b) but a pretty big gap to the MQA version c). Especially the basses are much more precise and the highs too with MQA. Some blind testing confirmed these results. My girl friend Alexandra prefers MQA:  "The bass is much nicer...".

Peter, it's great that you implemented the MQA decoding!
Thanks a lot for that!

Kind regards
Richard

[PeterSt]

Thank you, Richard.
I will try that album tonight (if I don't forget) and judge whether you and Alexandra are right.
Haha

Again, thanks !
Peter

[PeterSt]

OK, I like the cover to begin with ...

[Rmalits]

Yeah, I like the cover too!
The recording quality differs though from track to track. So don't losten to the bad ones too much, like the second track, what was recorded in Vienna with pretty strong hum on it 

[hudesigns]

This sounds fascinating to me! I love to hear the difference. But the strange thing is that I can't find this album in MQA format on Tidal. Is it because I am located in the states? Maybe you guys can send me a link?
Thanks,
Zheng

[Rmalits]

Hi Zheng,

hopefully this link will help:
https://listen.tidal.com/album/84487108

If not, let me know and we will find another solution.

Kind regards
Richard

[hudesigns]

Thanks, Richard. But Tidal says "The content is no longer available." after I clicked the link. You have another way? Maybe dropbox?
Best,
Zheng

[hudesigns]

Hi Peter,
First I must say that your reply #1 is both fascinating and complicated. I think I am beginning to feel more clarified and appreciate the process of "deblur". But I am still digesting... Here are some of my observations/questions:

1. If deblur is realised when MQA track is decoded (by XXHE for example), does rendering process in a MQA DAC adds anything more (maybe "blurring" to correct over-deblurring, as you suggested) to the deblur information? I am asking this because I do hear something distinctively different when my MQA enabled iFi DAC lights up the white light which indicates a MQA track is being processed -- a darker background behind female vocal which adds more air and expressiveness, and bass is tighter and more involved.

No. It happens in the ENcoding of the file. Thus, we have an album with its tracks (which are files as you and me know them) and they are processed through a software and this applies the "Deblurring"

-------

2. This is interesting observation: white light (MQA) on iFi DAC2 only lights up in these 2 conditions in my current setup: A) XXHE decodes, volume 0dB, 96=96 or 88=88, filters like Arc Prediction or custom ON. Or B) Tidal decodes, Full control of DAC, Passthrough OFF. If you switch off XXHE decode or Tidal decode, white light shuts off (meaning non-MQA track)

What it tells me is that iFi DAC2 in current form only performs rendering. That is why no 192 or higher playback is possible yet as result of "unfolding".

3. This leads to my question that what is "rendering" entails? Add filters of course? Is it an integral part of the whole time domain process? IF so, without a true MQA DAC, we are only partially enjoy the benefit of deblur? Or better yet--by carefully using appropriate filters settings in XXHE, we can achieve the whole glory of most transient time-corrected results as intended by original studio MQA encoder?!
This is certainly encouraging but I am still yet to find filters settings that will make non-MQA DAC present a similar result when the "while light" lits up on MQA DAC. But again it is subtle and can be very subjective.

What's crucial in your question(s) above is that we do NOT need the MQA DAC to apply this "debluring" because it already happened. All the MQA DAC adds is a proprietary set of digital filters which any self respecting DAC has inside its cabinet.

Excuse my very long mumbling

Best,
ZH

[PeterSt]

Zheng,

I am not able to read your post which talks about white lights "thus MQA" etc. etc. ...
A white light should be "no MQA" and further it is blue (Authenticated plus signed off) or green (Authenticated, not signed off).

Maybe I will retry later, but essentially we can not communicate over this in this fashion. Sorry.

I'd also have to explicitly notice that when it really is so as you implicitly tell (white = MQA) then the iFi does not comply to MQA's rules and it can't (officially) be on market. Even the blue and green colors are stated by exact (RGB) color code.
So I don't know what's wrong here, but something is.

Peter

[PeterSt]

Zheng, without really reading your post (I really can't), I think the essence of what you want to know is this :

Taken that the deblurring happens at the front end (like on the Tidal servers) there's now too few reconstruction filtering going on and when indeed no filtering is further applied, you'd have it worse than normal Redbook. Mind you, to compare we can only use a NOS/filterless DAC (and your iFi is not that AFAIK). In addition, in XXHighEnd you should set upsampling to 1x and set NOT Decode HDCD/MQA.
This now sounds indeed horrible (to me). And this is (quite) contrary to what MQA ltd makes us believe. But remember, a normal DAC should be not NOS and should apply a reconstruction filter itself. This I thus now eliminated.

So it would be correct that in this situation there's too much of "deblur" and wave shapes are too steep. It just won't be right.

An other option would be to let XXHighEnd decode MQA, still set upsampling to 1x, disengage all filters (no red led in each of the filter buttons) and listen again. Now you'd have the Tidal Desktop player situation, with the notice that it still would be best to use a NOS DAC for best (genuine) comparison. Assumed the native sampling rate is 88.2 or 96 (not 44.1 or 48), you'd have a higher sampling rate requiring less "reconstruction" with still the non-optimal situation of better reconstruction being possible (like upsampling/filtering to 176.4 and higher).
I say it again : this last upsampling steps (AKA 2nd etc. unfold) are just that, also for MQA. But when the MQA DAC is avoided (or shut off for the MQA decoding and rendering) you'd still have the too steep (too much "deblurred") wave form shapes. Thus, apply your own filtering (set upsample to more than 2x in XXHighEnd + select a filtering means) and all is better (if all is right). Or, engage the Rendering part of the MQA DAC and all is also better but different because of a different filter (undoubtedly).



I know, this is all the most vague plus it is my own reasoning (but at least I have one, contrary to what ever I see elsewhere). IOW, what ever I reason it should be consistent and such reasoning is also the only way to make things consistent. MQA ltd doen't write it out anywhere, is silent everywhere and makes it actually voodoo or hoax otherwise.
Still it is able to sound (very) different, and still it does not incur for me shutting such MQA albums off within seconds, like I do with about all existing Remasters of any random anything. So for this stupid theory alone, it should "be" something (and not a hoax).

Ehh, thanks for listening ?
Regards,
Peter

[Rmalits]

Hi Zheng,

there another new album from March 2018, which is very good for comparing Redbook with MQA. Here in Europe both versions are available on Tidal:

Artist: Nakhane
Album: We Will Not Die

Again, in my ears the MQA version outperforms Redbook in an pretty impressive way: better soundstage, clearer highs, more natural sound.

My dropbox is pretty full, so I will find another way.

Kind regards
Richard

[hudesigns]

Hi Richard,
This is so enticing! But again Tidal USA does not have this album. Could you send me a link to try again? Looks like I need to look into VPN to load up Tidal UK or something...

Keep on sending me new recommendations of music.

Best,
Zheng

[hudesigns]

Peter, you're a good man!
First I feel guilty of spending your time and energy from your crazy busy life (judging from insane amount of stuff you have to deal with...) Your points are helpful and will guide me to carefully tread through the treacherous path of MQA AB tests. Couple of points to clarify:

1. Yes, iFi's implementation of LED scheme is totally unconventional and has nothing to do with MQA's proposed scheme (blue/green). In fact iFi's schemes are not even consistent between their own products -- their OFFICIAL MQA light is magenta with no mention of how to indicate MQA's blue/green lights. But on my iFi iDAC2 it is white because there's no magenta color LED on some legacy products. In short, iFi has one LED to indicate whether its a MQA stream or not.

2. My iFi iDAC2 DOES have a switch setting called "bit perfect" which according to iFi is NON Upsampling. But it still needs software MQA to decode first (either by Tidal desktop or XXHE as a 88 or 96 stream)

3. As a comparison I want to try to let XXHE decode MQA with all filters OFF and volume at 0 dBFS.

4. Can you tell me which are XXHE filter buttons. I know Arc Prediction or Custom is one. Anything else?

Keep it short, 

Best,
Zheng