Sorry to hear about your condition - I sincerely hope you feel better soon and can fully recover from it.
2 Likes
I’m happy Theon is happy. 
IMO, this thread is totally pointless. Its going nowhere very quickly.
The discussion is the political equivalent of trying to convince a Republican or Democrat to switch political views. Never going to happen. What’s the end goal?
3 Likes
This is not about “changing” positions - but rather about seeking clarity on MQA and its (dis)advantages, since there is close to zero transparency from “Bob” or the company itself when it comes to how the so-called origami thing works, or whether its SQ really is above traditional lossless formats (and no, subjective views and “psychoacoustics” allegations from them are not enough).
And again, I say this as someone who subscribes to Tidal and has a MQA-compliant amp/DAC.
I essentially agree with this, if you’re posting hoping to change other people’s minds it’s likely a fruitless exercise. I’m also aware that I’m currently the 10th most prolific poster on the thread, and this post isn’t really helping my case 
1 Like
asking for transparency is like asking Microsoft for the source code of Windows 10 OS.
1 Like
If you relied on Windows 10 for core business and were nervous about Microsoft’s viability this is actually a problem. Escrow arrangements aren’t unheard of in these circumstances. If I owned MQA content, I’d be a little concerned about the long term viability of the second unfold in the event of the company going under. An offer to put the second unfold secrets into some kind of escrow so that they were published openly in the event of MQA’s demise would be a positive move in my book.
That is definitely not the same thing; not to mention that, even if we were talking about source code (which is copyrighted, thus NOT the same as a trade secret), everyone and their dog may see it, yet would know that it is MQA’s intellectual property and cannot simply be copied/used elsewhere without authorization - likewise for the patents that they have.
The closest to an objective analysis of the full “unfolding” experience is in spectrograms of the same source files and where the lossy aspect is, so that one can know which format is more faithful to the origin material. Again: subjective statements about “psychoacoustics” are not helpful here unless there is a clear explanation of why/how MQA needs to consider those in-air sound wave gains/losses when “optimizing” DACs.
Hi, I still would like to know why a 16/44 mqa decodes to 24/88.2 when the original recording was only 16/44. 
I found this ::
I noticed my decoder is outputting a 24 bit file from a 16 bit MQA encoded file (from one of those albums which Tidal replaced with 16/44.1 MQA).
You have the same issue as me. Even Roon support won’t comment on this. Why do they avoid the issue ?
What is even more surprising is Lumin firmware guy (he is here to) clame decoding is correct. It can’t be. It’s an error. Actually an display error. Which Bob says doesn’t matter if I understand his foggy explanations.
There is no doubt. Transport rate is 16/44.1. Original encoded file is 16 bit. Have never been upsampled. It’s a MQA Authentication at 44.1.
Roon has done a good job to show us what we get. Under the cover they always tells us what the transport rate and container is. After they display the original encoded file, which is verified by the authentication flag in SW. (Not so with tidal).
But the Core Decoder has an error. I assume we must blame MQA Ltd. Not Roon.
I’m quite sure the real file of cause is 16/44.1 (minus MQA bits or whatever has been done by the decoding machine.
The error also applies to all 48kHz/24 that display MQA 48 kHz.
Ow, and here is the source 
I pray you are soon cancer free 
3 Likes
It’s hard to believe that anyone can have avoided learning the answer to this elementary question that has been asked over and over, and answered the same number of times. 
All DACs (except Non oversampling) do upsampling to whatever data rate their modulators work at - typically 352.8 kHz. Your input 44.1 kHz signal is always going to be upsampled, either in signal processing or on a dedicated chip.
The MQA decoder is written to output at 2 Fs (88.2 kHz or 96 kHz, depending on whether the input is 44.1k or 48k).
If the MQA input signal is a packed hi res signal, the unfolding will produce a 2 Fs signal.
If the MQA signal is 44.1/16 original with no unfolding because it didn’t start as a high res file, the decoder (presumably) does the first stage of upsampling to produce a 2 Fs signal. (presumably because I haven’t seen it written, but this is elementary stuff).
It’s not an error, the Core Decoder in Roon is showing the decoder output rate. From there, the output will continue to be upsampled, either by an MQA-capable DAC or by your regular PCM DAC. If MQA-DAC, it will also have DAC correction applied.
2 Likes
You mean, all DACs that do upsampling do upsampling?
If only that were true! But we’re over 3500 posts in!
And, frankly, that makes me happy. I dread the morning when I check out this thread and find out that no one found MQA disappointing overnight!
4 Likes
and then …
I know mqa does upsampling x2 for 44 and 48kHz, but why is that?
To make up for the quality loss when they inserted the unnecessary mqa signalling?
(unnecessary as a 16/44 mqa is actually bigger in size than the original redbook)
1 Like
Arguing for arguing’s sake?
There’s gotta be a reason in here somewhere why MQA is wrong, just gotta keep tryin’…
4 Likes
I’m just asking why mqa is upsampling a 16/44.1 mqa to 24/88.2
when after that the dac is upsampling it anyway to the maximum it is capable of (let’s assume 352.8 here)
I learnt it’s best to do a one step upsampling by the dac.
So (pcm)
44.1 =>x8=> 352.8
and not (mqa)
44.1 =>x2=> 88.2
88.2 =>x4=> 352.8
The only reason I can think of is because they want to make the 44.1 pcm sound better by upsampling it 16=>24 and 44.1=>88.2 (called “mqa core decoding”) because they made it sound worse when inserting the mqa signalling
You aren’t a software designer, I can see that.
The decoder is written to output 2 Fs. When you write software, you have to be able to tell whatever module follows it what data rate to expect, so that the next guy can write his software.
MQA obviously was designed to be used within an MQA environment. The break after the decoder permits other digital processes (room correction, HQ Player upsampling) to be substituted instead if the user wants it but that’s an after-the-fact change from the original MQA design. It works out just fine. What you get is MQA’s filter with its short no-ringing impulse response for the first upsample, and then whatever else you prefer after.
There is no loss because of MQA signalling, the bits are (I’m assuming) buried in pseudo-random dither.
What the 44.1/16 signal definitely will have is the A/D correction applied to the input file, as well as the MQA upsampling filters and DAC correction in an MQA DAC. Those make a lot of difference in older CD files.
Again, if you don’t want MQA, just don’t use it!!
3 Likes
String s = “Yes I am”;
Application->MessageBox(s.c_str(), “Am I a software engineer?” ,MB_OK);
Run
There’s already 88.2 and 96 coming out of the core decoder, it’s not just one sample rate.
It might show that mqa wasn’t created with 44.1 and 48 in mind… (looks quite obvious,
as you won’t save space by mqa-ing those) but they mqa-ed those anyway. For what?
It seems you don’t want to discuss the format, just shout.
(looks quite obvious,
as you won’t save space by mqa-ing those)
From my above post:
What the 44.1/16 signal definitely will have is the A/D correction applied to the input file, as well as the MQA upsampling filters and DAC correction in an MQA DAC. Those make a lot of difference in older CD files.
That’s why so many people complain about 16/44 mqa and prefer the original redbooks.
Besides… what AD correction?
They batch converted digital PCM files… there’s no A only D.
And don’t bring on they can detect the chain of AD converters that were used and correct for all of them. Look at all those remasters available in mqa… how many A/D convertions happened and more importantly, how would they have to do correct for all those?
It’s already proven impossible to detect if a file was upsampled or not and that’s just one step.
I’m pretty relaxed about all this and I love learning stuff but I read all your posts now and the only thing I’m thinking of is that you have used a graph as a fact without researching it and sold it as proving something is wrong. I guess I’m finding it difficult to find the credibility now, I’ll keep trying though as I like balance in life.
What are you talking about?
I’m not talking about the graph here. I’m talking about a 16/44 mqa that core decodes to 24/88 for no reason I can think of, but to make up for some of the quality loss they created themselves by inserting an unneccessary mqa signal in the 16/44pcm in the first place.