MQA disappointing

Sounds really stupid of a rather big company as Auralic when it’s about streaming and DAC’s!
But I think pretty much will happen with MQA, when the Big Elephants in the hifi/high-end business realizing that they can’t afford to not jumping on train, and they can’t stop it either!
So was it when the CDs came, it took awhile before you could buy a decent CD player! I remember that my first only had 1 bit, the first really good one was a Yamaha that I bought in 1988 and it had I until it crashed, around 2004/5! They said pretty much the same as now about MQA!
It’s not going to be around so long, it’s only a new thing but the usual music listener will not buy it and it’s wasn’t especially much music to buy either!
IBM told the same about the Pc, that shall the usual people have a computer in their home for and what shall they use to!

The one who lives will see :wink:

Sounds rather brilliant to me. Implement your own interpretation of the renderer and include it in your product. Don’t pay Bob! In implementing your own interpretation realize the downfalls of MQA and warn your users not to use it. Pretty strait forward and honest sounding to me, basically the complete opposite of MQA Ltd.

2 Likes

Okay Jeff,
I getting your point in doing that and agree with you as well.
As long it works and regarding Bob, so have he created it and maybe taken MQA as far as he can or wants!
So maybe by doing as Auralic now have done, other will take it further and making it better!

Thanks for making me to see the bright side of it :smile:

Cheers

That wasn’t their intent though was it? You talk about it as if this was their master plan! :rofl:

Do you mean before the MQA DRM police came calling? Auralic’s plan seemed pretty straight forward from the beginning. They knew how they wanted to implement it to fit well in their system and what they wanted to give their users. MQA said, no blue light for you, so they did it anyway.

My experience and a reference:

I find MQA to almost always sound better than the regular (CD) version (Tidal or lossless ripped CD). I haven’t compared much to true HD or DSD. I am listening with Focal Utopia headphones and Chord Hugo 2 (and a few other speaker based systems with pretty highend gear, none of my DACs are MQA DACs, so I’m just listening to the first unfold) and can hear significantly more “space” and information with most MQA versions than the non-MQA versions. Also I find MQA just more pleasing to listen to. I tend to want to just keep listening to most MQA presented recordings.

Here’s a recent interview of Bob Stuart (MQA developer) as to why you want to set Roon to minimum phase mode and also some other interesting things I didn’t know about MQA (and I know a fair amount). https://darko.audio/2018/05/playback-pioneers-2018-bob-stuart-mqa/

Getting the settings right is important.

2 Likes

If you are using Roon to do MQA decoding at 88.2/96k try up-sampling to the maximum capability of the DAC (select max sample in device settings, select power of 2 in DSP-SRC) then select smooth minimum phase filter. In my listening test this method produces the best SQ.

Do note using smooth minimum phase has its disadvantage. The filter is essentially quite weak to attenuate aliasing images, and this gets reflected back in the audio range causing distortion.

1 Like

Yes, there are trade offs, some argue that it’s other benefits are worth it.

But in this case, the aliasing problem is not serious because we have already folded up to 88 or 96k. So the aliasing is way up beyond the audible range.

As to the minimum phase recommendation, this statement comes directly from Bob Stuart as stated in the video I linked to. His statement is that this is what should best be used for MQA. He had some reasons he explained in the video.

That said, I do hear (on non-MQA recordings) some advantages to linear phase upsampling (I prefer “precise linear phase” over “smooth linear phase”, especially for natural acoustical recordings. My understanding of precise vs smooth is that the main reason to use smooth is if you don’t have the processing power to use precise upsampling. My listening tests although brief seem to support that.

As to the maximum upsampling (power of 2), it’s interesting, when I do this in Roon (which of course happens after the first unfolding decode), I find that my DAC which is MQA compatible, no longer shows full MQA rendering in its indicator lights. Perhaps this is just to indicate that the signal has been in some form tampered with. Not sure. In general I think upsampling in Roon is often good, but whether it’s best done in the DAC or not depends on the DAC. Lots of variables!

Not exactly true, some MQA albums are done from 44.1/48k masters, these albums suffered more aliasing artifacts than those 88.2/96k masters. Take a look at this excellent article and you know why audiophiles are very concern about this type of artifacts. Read the computeraudiophile on MQA, it gives a very detailed information about this format.

This only applied for those who don’t have a MQA DAC, and many of us don’t own it!

The purpose of this up-sampling to maximum capability of the DAC does one very important step. If the DAC is fed with the highest sample rate the internal digital filter of the DAC will behave like a pass through so we use Roon smooth minimum phase filter instead. This also mimic like ‘MQA renderer’.

And since you own a MQA DAC, you can easily make this comparison and hear if there’s any different in the SQ.

Blockquote
This only applied for those who don’t have a MQA DAC, and many of us don’t own it.

I’m not sure your above statement is true (but I’m not 100% sure). The reason I say this is that Roon support has stated that they can do the first unfold and then allow Roon processing (DSP, etc. which I assume would include upsampling) and then when the signal gets sent to an MQA DAC it should still show as MQA fully decoded. Not all MQA player software (or hardware) can do this, . So my statement was indeed specifically about using an MQA DAC. Certainly if the DAC is not an MQA DAC then the there would not be any indicators on the DAC for MQA support or correct decoding/rendering at all.

Blockquote The purpose of this up-sampling to maximum capability of the DAC does one very important step. If the DAC is fed with the highest sample rate the internal digital filter will behave like a pass through so we use Roon smooth minimum phase filter instead. This also mimic like ‘MQA renderer’.

I understand the reasoning for using Roon upsampling, so as not to use DAC internal processing. But my comments in my original post are mainly regarding decoding MQA with MQA DACs. Even then it makes sense, and this is why I noted the MQA indicator light issue when doing the upsampling in Roon. Hence my original post. For all other DACs, this upsampling approach certainly makes sense to me (although some highend DACs may best the Roon processing). Again, lots of variables.

What I’m trying to say those who don’t have an MQA DAC, can use Roon MQA decoding and combine Roon’s up-sampling + smooth minimum phase filter (this process mimic that of MQA renderer). In my test using Mytek Brookyn DAC, a fully certified MQA DAC, I can’t tell the different between using a fully decoded MQA DAC and Roon’s MQA decoding and up-sampling + smooth minimum phase filter.

This process open up for people who want to listen MQA to near it best capability while don’t have the luxury to own a MQA DAC.

1 Like

Oh, I know about all the concerns.
But in the case of unfolding in Roon, even 44/48 is turned into 88/96.
Per MQA rules.

All modern non MQA DAC over-sample 44.1/48k to 352.4/384k (8x over-sampling) internally, this effectively remove any form of aliasing. 88.2/96k is too low to supress any form of aliasing artifacts. Don’t forget aliasing artifacts does not just exist above the ultrasonic range, the image can get reflected back in the audio range and modulate with the music contents causing undesirable distortion.

Thanks for clarifying this issue! This makes a lot of sense.

I do have one question though, is it specifically “smooth minimum phase filter” for upsampling or is Roon’s “precise minimum phase filter” perhaps even better? Roon documents have alluded to the precise upsampling being a better way to upsample but it does require more processing demand from the server. I don’t remember that Bob Stuart indicated “smooth minimum phase upsampling”, my remembrance is that he said to just use minimum phase upsampling for an MQA unfolded signal (but I’m not 100% sure). Thanks for any further clarification on this!

MQA uses slow-roll off minimum phase, similar to Roon’s smooth minimum phase filter. However, MQA has total 16 different variant of slow-roll off minimum phase filters, some are catered for different sample rates and some has slightly different post ringing shown here

Precise minimum phase filter (also know as sharp cut off minimum phase filter) has no pre-ringing but a large series of post ringing. Try to avoid using this, instead smooth or slow roll-off minimum phase has no more than 1 cycle of post ringing.

Hi dear Roon Fellows.

Interesting articles and discussion about smooth minimum phase filters in MQA, and other things about it.
When I was sleeping because of different clock there we are living!
But it’s okay, I don’t feeling what’s I been dismissed from the MQA discussion.
It was more interesting to read than the newspaper :wink:

Love & Respect

Does anyone watching Seinfeld on TV?
The episode about “The soap nazi” if they shall buy his soap that’s So Good, but if they don’t follow his rules, he shots No Soap for you!

If you don’t follow “Bob’ rules MQA says, No Blue Light For You”!!!
I think it’s both funny and scary! Because no one shall have that power to use, as he wants! Even if “Bob” has created MQA! It’s Dictators Rules!!!

But’s it’s my thoughts! So soon does my Blue Light disappears :smile:

Like this ?

I have to edit this post later. For some reason I can’t upload pictures anymore from my iPad. I’m getting an error

uninitialized constant EXIFR::JPEG