In essence the whole thread comes down to a simple sentence: The MQA decoder for streamed applications appears written to expect a 2Fs output from the first decode stage, and if it doesn’t get there by unfolding a packed signal, it does so by upsampling.
Is it obfuscated and do consumers need to know? I don’t know. However there are individuals and websites dedicated wholeheartedly to shredding MQA, and doing that without much knowledge of ordinary signal processing or of the meaning of a simple Fletcher Munson (hearing) curve. So I’m not altogether surprised by the company’s limits on public information.
This is the decoder code we’re talking about. Signal has already been authenticated (so no outside upsampling took place). Internally MQA can do anything it likes. Upsampling has to be done anyway, in all DACs, to raise the sampling rate to whatever is required by the modulator.
Madonna
Whole different story. An original 192 kHz (4Fs) track will be a packed signal in MQA, and is subject to a couple of unfolds in decoding The upsampling discussed in this thread is just what the decoder [presumably] does for a 1Fs 44.1k or 48k signal.
The zeros are being added to the MQA delivery file and are not impacting the encoded information…Bob is explaining there is no compromise of the MQA process. That said, this issue Bob talks about is irrelevant to the sampling rate coming out of the MQA decoder. Whether it is a 16b MQA or 24b MQA the decoder will always output at 2x the MQA file (delivery) rate. This is NOT an error in Roon and you can see the same behavior in the Tidal desktop application.
My head is spinning as fast as my red book CD in the player. Simple question from a simple mind:
I stream a Tidal MQA via Roon and disable all MQA functionality be it at core or renderer level. The signal path says 44.1/16 which on the surface looks great. Is this red book 44.1/16 or something else as a result of ‘MQAziation’ messing with the master?
I’m with @joel - it’s not an error. If one uses the MQA Core decoder from Lumin (instead of Roon), it results in the same 24/88.2kHz result. I expect all MQA Core decoder for all certified MQA transports to output the same thing.
“Deblur” is a name applied to specific processing techniques. The techniques are used. Their effects are audible, both to studio engineers and apparently to commenters on consumer forums as well. In the latter case, “deblur” is probably heard in combination with the effect of MQA filters, and so might not be separable outside a studio environment.
Both “deblur” and “authentic” are MQA marketing speak masquerading as important technical advancement or even downright benevolent provenance. As I have posted before, this is selling audiophiles on philosophies, getting them to buy in on specious ideas in advance.
Authentic means what it says. Now if you think Warner Music do not have the authentic versions of the albums they own to release with MQA, then take it up with them. They say they do have such versions, many at 16/44 and have released MQA versions of these.
‘Deblurring’ of the source to remove audible artefacts introduced by analogue-digital converters, mixing and mastering.
In the early years of digital, recording and production equipment was much less sophisticated than today. At one level this might be an advantage – less ‘messing with the recording’ between studio and release keeps it cleaner – but also early digital technology introduced systematic defects that we are able to detect and correct. (Some of this is described in our AES paper[1])
I guess that I am never clear why people insist that something cannot have been done when the individual who actually did it makes a statement saying that he did?
1 Like
Bill_Janssen
(Wigwam wool socks now on asymmetrical isolation feet!)
225
Because a lot of people are liars? I mean, really, that’s why we have the word “claims”, as in “he claims to have done it”.
There are about 35 years of published papers from the people involved in MQA, papers that develop the techniques and ideas of it. You’re accusing the people who invented burial of data in pseudo-random noise, who developed the first lossless audio compression algorithm, who had significant roles in developing DVD and Blu-ray specs, and who have extensive design experience in noise-shaping, dither, apodizing filters and optimization of DAC chains of LYING about MQA?
I love the way you continue to use this as a claim to authority, as if studio engineers haven’t been responsible for some of the most abominable, cringeworthy recordings in history.
I believe MQA was vetted for years with specific studios and later with labels and others. It wasn’t a casual process. It would be better had listening tests been carried out to the point of publication, but formal listening tests take too long and rarely get done. If you have a better idea on how to vet codecs or other algorithms, let us know. Listening tests are what’s usually done.
