OK but so what, if they don’t want to listen and you think it’s better then it is their loss. There are no winners or losers just arguments which go on for pages and pages.
I hope this forum is above this, listen to the music, enjoy it whether MQA or not, life with music is good!
If you guys read what I said, it is about the first unfold and my impression that successive unfolds are upsampling plain and simple. I also said that if such upsampling is aided by the unfolded stream setting some upsampling parameters then that is interesting and smart.
There seems to be a lot of sensitivity about any criticism of MQA here by some folk. I have yet to hear an actual argument that is not a personal attack.
I like the sound of many MQA albums (Joni Mitchell, The Smiths for example).
You have seriously misunderstood my posts. I am not arguing the SQ or merit of MQA per se.
My points are solely about software unfolding versus hardware unfolding. I repeat until we hear the software unfolding in ROON and A3+ or other versus hardware unfolding we cannot make any point on their respective merits i.e. the need for specific hardware to experience MQA.
Cynically hardware manufacturers who have invested in MQA have much to protect.
Didn’t you say tha you’d read all the literature that is available on this matter ?
The above statement, repeated several times above, is utterly wrong.
As I said above, if the MQA sample isn’t more than 44 or 88 or 96 then it doesn’t unfold to more than that. If it was sampled a 188, 192, 384 etc then it unfolds as that. There is no upsampling in that …period.
Enough !!!
What I find curious is MQA encoding for informed decoding of the first unfold cannot survive application of DSP. But after the first unfold, MQA encoding for informed decoding of any subsequent unfolds can survive intermediate DSP. One of those things that makes you go hmm…
AJ
This article (long and semi-technical) is well worth a read. It does some measurements of software unfolding vs hardware and the results may help you get excited about software decoding, might even slow you down in that decision to buy an MQA enabled DAC.
[quote=“WiWavelength, post:87, topic:20611”]
…after the first unfold, MQA encoding for informed decoding of any subsequent unfolds can survive intermediate DSP. One of those things that makes you go hmm…
[/quote]Makes me think of room EQ.
Good to point out, so you see there’s so much confusion and unanswered questions flying around, this is really NOT good! The ‘manufacturer’ itself should answer all these questions and not us! Besides, they are the one who creates this in the first place!
I wish there’s a forum from them, then they will take charge to all the queries and should not be afraid to face difficult questions. As I raised up earlier on, the confusion and doubts will keep on continue, Dam! It is almost 2 years! Enough is enough!
@WiWavelength @mitr @MusicEar
They have shown a block diagram where such processing is inserted between the initial and the final processing. So they have confirmed that it works.
Do they owe us an explanation of how it works?
What that implies, as you know, is there’s nothing past the first unfold.
Yes, of course it does work… but only after the first decode. Doing any form of DSP, even a simple volume attenuation before the first decode will simply destroy the code and makes it undecodable. This is fundamental limitation of an encoding and decoding process.
I, for one, would like Roon to take all the time they need to do it right, first time.
And with that I mean, taking advantage of the awesome, unique, platform they have build.
They are able to control the chain, which is exactly what mqa is looking for. They know the dac and can do a full software decode (all foldings) for it. Even if it has hardware support for mqa decoding. Would you want this? Well, maybe the purists that have a perfectly treated room don’t, but most of us will benefit more from dsp than from the final unfolding steps on mqa…
Roon can bring the best of both worlds!
As long as they (using the color scheme already in place on the ’ lights’) show whether it is a true ’ clean’ MQA stream, or an MQA with dsp stream, they are true to the MQA philosophy I would think.
If it takes a little more time to convince team MQA to see the light (enter: awesome demo for them), then I would really prefer that. Coming out with anything less is would be a bloody shame.
Would much prefer another month or so, to do it right first time, then going half baked and loosing the discussion opportunity to do it right.
I’d be surprised if team Roon doesn’t feel the same way, as far as I can see, they have architected Roon for a reason…
So, I am going to lean back, enjoy 1.3 with this awesome convolving stuff, and hope it will take a while for Roon to come with MQA, doing it right.
I’m imagine Roon would welcome being able to provide a whole MQA ecosystem. But, the question is not technical, it is business and maybe even legal depending on existing contracts. I do not think MQA will make any exception to their “final decode happens only at the licensed DAC hardware level”; at least not for many years if at all.
Before you make such statements you should look at what MQA claims. This is from Stuart’s answers to reader questions in Computer Audiophile, answer to question 65.
This is exact what I meant, you cannot do any DSP before the first decode (bitstream input) this has to be conveyed losslessly to the first decode (decoder 1), then you can do a optional DSP (optional immediate processing) thereafter as indicated in the diagram.
The word ‘Must be conveyed losslessly exactly’ to the MQA decoder. So you can’t do any form of DSP without disrupting the code at this point! The diagram is self explanatory! Dude, do you understand?
Anders, this really confirms what a lot of people are saying: i.e., that the first unfold is the one that contains actual musical data. That’s why the input must be lossless / exact - to reconstruct any real content.
After that, it’s all just upsampling + filters; i.e., you can insert additional processing, etc., prior to the second (hardware) “decode” . . . because the hardware “decoding” is actually only upsampling with a custom filter, no real musical content. Which is pretty obvious given there is usually only silence and or random noise above 48k, in any case, but interesting to see this coming out.
@jhwalker That is incorrect. Of course there is up-sampling - there has to be as there is a change in sample rate. Of course there are filters - it’s public knowledge that (at least) part of these are generic or DAC-specific (in full MQA decoding solutions). But MQA’s audio quality facet is all about preserving timing. AFAIK, with my limited knowledge of advanced DSP, you can’t “create” better timing by up-sampling.
I suggest reading the (free to download) A Hierarchical Approach to Archiving and Distribution paper. The encoding and decoding implications of the revolutionary (in audio terms) ideas are pretty clear to me. All the clues are there.
It means that if Roon does MQA step 1 decoding, it can add DSP, before sending it to the DAC.
We have always known that you can’t do DSP entirely before MQA. But this split processing, which wasn’t widely known before the CA explanation, allows that. And it will fit in with Roon’s processing (we guess). Very cool.
I know people are saying that. Doesn’t make it so.
See MQA for civilians in Computer Audiophile. This diagram of the final step illustrates how information is pulled out from its hiding place below the noise floor and applied in the upper range:
