MQA Listening Impressions

[Miska]

23 minutes ago, manisandher said:

OK, hopefully this will be the most insightful non-MQA vs. MQA comparison I've managed to come up with.

 

non-MQA _ HQPlayer _ poly-sinc-xtr-mp to 705.6 _ NS5:

 

MQA _ XXHighEnd _ MQA first unfold to 88.2 _ Arc Prediction to 705.6

 

Hmmh, sorry Mani, I don't understand what is this supposed to demonstrate...

 

So this is like MQA track played without unfolding through HQPlayer? Compared to unfolded MQA track played through XXHE?

 

And curious why did you choose such filter in HQPlayer? Btw, are you on 3.18.1 or newer to avoid the problem (created by buggy compiler) that was in minimum-phase filter initialization in some earlier Windows versions?

 

23 minutes ago, manisandher said:

I'm capturing the analogue output of my DAC with a Tascam DA-3000 running at 24/192

 

You could use DSD128 instead because the ADC (TI PCM4202) is 1-bit DSD ADC. PCM outputs are on-chip conversion from DSD128...

 

This way the recorded result is not subject to decimation filter effects.

 

23 minutes ago, manisandher said:

But even with all this, hopefully you'll get a sense of what I'm hearing in the MQA file. It's simply 'cleaner' sounding. And what I'm really interested in is how this 'cleaning' has been achieved with the MQA processing.

 

Hires vs RedBook and nothing to do with MQA? If you'd use the 88.2k version as comparison point it would probably sound cleaner than the MQA version. Now due to unfolding you got rid of the strong images of XXHE leaky digital filter between 22.05 and 44.1 kHz! Of course it sounds cleaner! But it has nothing to do with MQA.

 

Now compare the unfolded MQA played through HQPlayer instead of XXHE!

 

But for comparing MQA against non-MQA you need to use non-MQA 88.2 kHz version of this track and compare that against the MQA 88.2 kHz version.

 

I'm not going to even try listening this, because this is even further from any valid comparison, now it maxes out all possible variables so everything is as different as possible.

[manisandher]

6 minutes ago, Miska said:

So this is like MQA track played without unfolding through HQPlayer? Compared to unfolded MQA track played through XXHE?

 

No.

 

This is a regular 16/44.1 played via HQPlayer (to use the best anti-image filter available) vs. a 24/44.1 MQA played via XXHighEnd (to do the unfolding to 88.2 and then use the best time-domain filter available).

 

I think this is incredibly insightful, and gives a sense of what MQA can do to the sound. No probs if you don't agree.

 

Mani.

[manisandher]

11 minutes ago, Miska said:

But for comparing MQA against non-MQA you need to use non-MQA 88.2 kHz version of this track and compare that against the MQA 88.2 kHz version.

 

I want to listen to the music I like in the best way I can. If there's only a choice between a regular 16/44.1 vs. a 24/44.1 MQA, right now I'll take the MQA (in most cases).

 

If a 24/88.2 were available, I'd definitely want to take a listen, but wouldn't be prepared to pay the earth for it. Right now, I'm getting a bunch of MQA files for no more money than my regular Tidal subscription.

 

Mani.

[Miska]

1 minute ago, manisandher said:

This is a regular 16/44.1 played via HQPlayer (to use the best anti-image filter available) vs. a 24/44.1 MQA played via XXHighEnd (to do the unfolding to 88.2 and then use the best time-domain filter available).

 

I think this is incredibly insightful, and gives a sense of what MQA can do to the sound. No probs if you don't agree.

 

OK, so you think comparing RedBook and hires played in different way is insightful about MQA? No it is not. MQA version is quality degraded hires while RedBook is made from the hires using some unknown method (at least with quite a bit of noise shaping seemingly).

 

Sorry, I think it is just bullshit that this comparison would tell anything about MQA. You have other much bigger factors in play!

[Miska]

3 minutes ago, manisandher said:

I want to listen to the music I like in the best way I can. If there's only a choice between a regular 16/44.1 vs. a 24/44.1 MQA, right now I'll take the MQA (in most cases).

 

If a 24/88.2 were available, I'd definitely want to take a listen, but wouldn't be prepared to pay the earth for it. Right now, I'm getting a bunch of MQA files for no more money than my regular Tidal subscription.

 

But that doesn't tell a freaking shit about quality of MQA! So just admit that you like hires (although a degraded one) more than RedBook. Fine, that is quite expected. It would be even much better without the MQA part that is totally unnecessary from technical point of view.

 

From ethical point of view I rather listen to RedBook than MQA.

 

[manisandher]

2 minutes ago, Miska said:

So just admit that you like hires (although a degraded one) more than RedBook. Fine, that is quite expected.

 

In most cases, yes, I prefer the MQA to the redbook, where they're from the same master. If this is the only choice I'm given, I'm going to take the best SQ.

 

How long have we waited for hirez to become ubiquitous? It hasn't happened. ECM has started releasing some of their material in MQA, available from Tidal for no extra cost (right now... and I hope that continues!). Sorry, but I'm going to take it, and enjoy the music I love, with better SQ than 16/44.1.

 

Mani.

[PeterSt]

40 minutes ago, Miska said:

Now compare the unfolded MQA played through HQPlayer instead of XXHE!

 

Mani can't do this. HQPlayer does not decode MQA ...

[PeterSt]

44 minutes ago, Miska said:

But that doesn't tell a freaking shit about quality of MQA! So just admit that you like hires (although a degraded one) more than RedBook.

 

Miska, this is nothing about Hires. This is what you like to make of it. I know how Hires sounds (when executed well). This is not about that. Even the contrary :

 

Quote

The MQA version sounded again similar, more grainy

 

You mentioned that at least three times in this thread. Why ? because it is your general consensus about MQA.

It is my consensus too.

Ha, another unexpected one. I can tell you : it even is so that when I listened to MQA for two hours in a row, I start to long for "resolution". And almost always I end my listening session of the day with stuff which satisfies that. And I always wonder whether I am more happy with that or not.

 

Quote

and suffocated with reverb tails collapsed.

 

You said that a couple of times too. Again I agree; MQA sounds more dry. Maybe no wonder with the less ringing, right ? And then to think that I say this relative to MY playback (filters), not yours. So surely you will notice that.

 

But now this one :

Over and over again when I listened to the Rush versions yesterday, one thing was very apparent : the clear hall / reverb  in both of the very different MQA versions. Of course we can blame the master(s), but isn't this what many are saying ?

So go figure : more dry and more reverb at the same time.

Btw, I wasn't working with Mani's uploaded tracks; I worked with the Tidal and my own albums (the MFSL) for the judgments.

[semente]

1 hour ago, PeterSt said:

 

Miska, this is nothing about Hires. This is what you like to make of it. I know how Hires sounds (when executed well). This is not about that. Even the contrary :

 

 

You mentioned that at least three times in this thread. Why ? because it is your general consensus about MQA.

It is my consensus too.

Ha, another unexpected one. I can tell you : it even is so that when I listened to MQA for two hours in a row, I start to long for "resolution". And almost always I end my listening session of the day with stuff which satisfies that. And I always wonder whether I am more happy with that or not.

 

 

You said that a couple of times too. Again I agree; MQA sounds more dry. Maybe no wonder with the less ringing, right ? And then to think that I say this relative to MY playback (filters), not yours. So surely you will notice that.

 

But now this one :

Over and over again when I listened to the Rush versions yesterday, one thing was very apparent : the clear hall / reverb  in both of the very different MQA versions. Of course we can blame the master(s), but isn't this what many are saying ?

So go figure : more dry and more reverb at the same time.

Btw, I wasn't working with Mani's uploaded tracks; I worked with the Tidal and my own albums (the MFSL) for the judgments.

 

 

I'm with @Miska here: comparing how a (or several) particular recording(s) sounds in Redbook, MQA and HR versions doesn't tell you anything about MQA sound quality per se (and if you were to do it with all your recordings would make this topic the longest in history).

 

 

eClassical.com has a couple of demo tracks available (from BIS recordings).

But because the masters are slightly but possibly audibly different (and in my understanding biased towards HR sounding better) there is no way we can achieve an honest and meaningful conclusion - in other words they're worthless.

This is the "La Mer" track:

 

 

[manisandher]

3 hours ago, Miska said:

But for comparing MQA against non-MQA you need to use non-MQA 88.2 kHz version of this track and compare that against the MQA 88.2 kHz version.

 

[Highlight mine]

 

@Miska, I've found a non-MQA 24/88.2 version of this track on HDTracks - same master as the MQA!

 

I'd like to use the best anti-alias filter in HQPlayer during replay - which one do you suggest I use to take it up to 705.6? And NS5 too?

 

(I'd be happy to use XXHighEnd and its Arc Prediction filter, but am preempting the backlash at having used a time-optimized, i.e. 'leaky', filter.)

 

Mani.

[Miska]

5 minutes ago, manisandher said:

 

@Miska, I've found a non-MQA 24/88.2 version of this track on HDTracks - same master as the MQA!

 

I'd like to use the best anti-alias filter in HQPlayer during replay - which one do you suggest I use to take it up to 705.6? And NS5 too?

 

 

Good! There's no clearly "best", what people prefer is another thing. "poly-sinc-short-mp" is probably good compromise because it is somewhere in the middle between time and frequency domain optimized and minimum-phase variant is probably good because that is also what MQA uses. Just make sure you have HQPlayer 3.18.2 or 3.19.0.

 

TPDF dither is just fine, remember to have "DAC Bits" set to 24. (in case the driver doesn't tell correct value)

 

Don't set HQPlayer's volume higher than -3 dBFS.

 

After recording you could run both recordings through normalization to minimize level difference between the two. (because HQPlayer will be a bit lower with that volume setting)

 

[Miska]

2 hours ago, PeterSt said:

Mani can't do this. HQPlayer does not decode MQA ...

 

HQPlayer doesn't need to do decoding, HQPlayer Embedded can take digital (and analog) inputs from other devices. I've been playing streams decoded by the Tidal application and files decoded by Audirvana+ through it. I think I'll also get Bluesound NODE2 for testing purposes too.

 

I can try to make some DXD test recordings at some point, original hires, MQA decoded and rendered by DAC and MQA decoded in software upsampled by HQPlayer.

 

[mansr]

12 minutes ago, Miska said:

HQPlayer doesn't need to do decoding, HQPlayer Embedded can take digital (and analog) inputs from other devices. I've been playing streams decoded by the Tidal application and files decoded by Audirvana+ through it. I think I'll also get Bluesound NODE2 for testing purposes too.

You can also use my wrapper around the Bluesound MQA library.

[manisandher]

New level matched 24/192 captures of the output of the DAC posted in OP.

 

Mani.

[bhobba]

10 hours ago, watercourse said:

Do these people that you are speaking for have their own way of listening and comparing MQA tracks, or only via your system, during limited-time sessions?

 

I do quite a bit of listening sessions with other audiophiles.  At these we compare quite a bit of stuff including MQA.

 

Just today I was with a very experienced audiophile checking out a new DAC and we listened on his system, via Audivarna, HQPlayer (it was by far the best which annoyed my acquaintance no end - but that is another story) and the Tidal app.

 

Again we checked, via the Tidal app, a number of albums at 44.1 and 96 (eg Dianna Krall Wallflower).  We checked it on a ripped copy on my disk as well.  The other person was pretty upset when I revealed his deluxe version of Wallflower in DSD when I ran it through my 'truth' program was just up-sampled 44.1.  The order was -  worse to best - Tidal 44.1, my ripped copy played through HQPlayer, and Tidal MQA - with the MQA he was just rapt saying - wow - listen to the bass - I haven't heard that before with this.  It was my view as well BTW.  We didn't have the Bridge connected on the DS to hear the full unfolding.

 

The system was:

1.  An old Mac-Book with an old version of Audirvana.

2.  My Windows machine I use for nothing but music with HQplayer I used to upsample to DXD

3.  Two DAC's - a Direct Stream and the DAC we were checking out built by a local person.

4.  BHK pre 

5.  ISO Regen with specially made linear supply made by the guy that makes the DAC.

6.  BHK 250W amp with specially selected NOS valves done by a guy who is a valve nut.   Its the best BHK I have heard.

7.  ML1 speakers which are very good mini monitors.

 

It is at sessions like this with varying equipment and listeners that I have done this comparison. 

 

Thanks

Bill

[bhobba]

10 hours ago, PeterSt said:

This reminds me of an other maybe not coincidental matter : Bill (bhobba) is a NOS guy - and then even without any reconstruction filtering. Is that still so, Bill ?

 

No.

 

You are  thinking of my liking for a DAC built around those old Phillips Double Crown chips called the Killer. 

 

Now opinions on that DAC vary - some say its simply real - others think its just suitable for a certain vocal type of music where it is very addictive.   My view is a bit in between - on the music I tend to listen to - Diana Krall and what not it sounds wonderful - but ever so slightly euphonic - still I love it - my personal favorite, although I never recommend to anyone to just get that DAC - on certain types of music (eg Leonard Cohen - In My Secret Life) to me it sounds not quite right.   The guy that makes the Killer says its simply reproducing what its feed and it what Leonard Cohen sounds like - we are good friends but differ on that one.

 

However those chips are getting harder and harder to come across, so the person that makes it is moving on to some other chip that does 386k but wont tell me what it is - its hush hush right now.  However it will asynchronously re-clock with some ungodly accurate clock called a Killer clock.   I haven't heard one but those that have rave about it.

 

And no - that was not the new DAC I heard today - that was an entirely different beast- it used an ultra modern AKM chip but with a very interesting output stage - just a Jensen transformer.

 

IMHO it MUDERED a Direct Stream which I also own - but after hearing this DAC I may get rid of it.

 

I own a number of DAC's but will likely rationalize sometime.

 

Thanks

Bill

[manisandher]

1 hour ago, manisandher said:

New level matched 24/192 captures of the output of the DAC posted in OP.

 

I'll get the ball rolling...

 

Firstly, I have to say that the 24/192 captures I linked to in the OP unfortunately fall short of the SQ of the original files being played back directly. However, I believe they are good enough to highlight the differences in SQ between the 3 files, and to give a sense of what I'm hearing coming out of my DAC.

 

Let's put the 16/44.1 out of its misery right off. However it may have been derived from its master, it's lost a lot in the process. The other 2 are a quantum leap better all round.

 

The MQA vs. the 24/88.2? Hmm...

 

There is a 'clarity' to the MQA file that the 24/88.2 doesn't have. This gives it more ambience and generally more life - more 'get up and go'. But I sometimes detect a slight 'edge' to instruments and the vocals that makes it sound slightly more 'processed'... which of course it is. But nevertheless, my ranking is:

 

1. MQA 24/44.1

2. 24/88.2

3. 16/44.1

 

As I said, you should be able to hear my substantiation in the 24/192 captures.

 

Mani.

[bhobba]

16 minutes ago, manisandher said:

There is a 'clarity' to the MQA file that the 24/88.2 doesn't have. This gives it more ambience and generally more life - more 'get up and go'. But I sometimes detect a slight 'edge' to instruments and the vocals that makes it sound slightly more 'processed'... which of course it is. But nevertheless, my ranking is:

 

Bingo - that's what I hear -  I describe it as too clean.  Do you hear a certain thinness as well?  That's my impression.  Compared to 88/24 - that's a hard one for me - I may give the nod to MQA at that resolution.  But 176/24 up-sampled to DXD via HQplayer - for me that's better than MQA.   I wish I could get an MQA version of my DXD master - I believe the master will blow away the MQA.  But then again the bandwidth requirement for DXD preclude its current use for streaming and that's where I think MQA will win hands down.

 

Thanks

Bill

[manisandher]

4 minutes ago, bhobba said:

But 176/24 up-sampled to DXD via HQplayer - for me that's better than MQA.

 

Bill, I'm up-sampling to 705.6/24 in HQPlayer! I still have a slight preference for the MQA, and believe it's reflected in the 24/192 captures.

 

Mani.

[matthias]

Mani,

why you did not use Arc Prediction for all the three files?

Thanks

 

Matt

[mansr]

Is "arc prediction" some form of spline interpolation or what? Doesn't seem like a good idea.

[firedog]

13 hours ago, bhobba said:

BTW I have tried the 2l test, and yes for me the MQA does sound a bit better than the DXD.   But I suspect it wasn't really what we call in Aus a fair dinkum comparison ie they selected stuff that MQA was better on.

The other issue is that for 2L they did a ""bespoke" MQA version, whereas in general they are doing mass transfers to MQA using generic (not A/D specific) de-blurring. So your generic MQA album isn't turned into MQA at the same level those 2L files were. 

[Dr Tone]

Just a suggestion:  The samples should be posted without being labelled and without comment on what Mani perceives as the difference.  Wait a day or 2 after members have voiced their opinion on preferred sample and why and then divulge which is which.

 

You can still describe the pedigree of the samples just not the specifics of which is which.

[matthias]

Mani,

please do further comparisons with ONE variable only, MQA vs. non-MQA, and do not use different filters.

Thanks

 

Matt

[watercourse]

I’ll be on work travel the next couple of days so I will be away from my system, but this morning I listened to The New Pornographers latest album. I started listening to the first track of the MQA version on Tidal, but was bothered by some treble harshness. I then switched over about halfway through the first track to the non-MQA version, and didn’t switch back to compare with the MQA because the sound was much more enjoyable to me.

This is not an atypical experience for me.