Maybe "MQA" have come up with an algorithm that has a go at fixing a track that has already been mixed.
Or maybe they can only do tracks where the original multi-track masters still exist?
Here's a thought. Perhaps what MQA does in the decoding is to pulse the DAC in such a way that it produces an additional sine wave simultaneously which supplements and/or cancels the flaws in the "master wave"?
Maybe a little bizarre but it would explain is why it has to be on the playback side, too, and why it has to be hardware based.
Both 'unfolding' and 'bitstacking' require a decoder. That much is clear from the info made public in the MQA patents and papers.
The problem with making assumptions about how the originals were recorded and 'guess the ADC' pretty much ensures, I suspect, most times they'd be wrong. Too many variables.
Trying to do so by analysis of the results falls into what signals engineers know as the 'deconvolution problem'. You can sort out trivial cases, but not complex one like real music.
If the ADC matters, so does the choice of microphones, their placement, the amps and cables, etc, etc. They could all affect the timing alignment. And analogue recorders vary. On that topic old-time recording engineers could probably tell some tales! Plus in pop/rock areas the people making and balancing may well have fiddled with everything in sight to get what they decided 'sounded good'. Perhaps with no idea what some of the sliders or boxes did. Possibly not noting down all the details.
The 'additive pulse' idea isn't really different from describing the process in other ways. Mathematically, you can define these things in various ways.
Personally, I suspect the 'orgami' process does, indeed, simply add a burst of HF which produces a result that isn't the same as what went in to the encoder, but 'adds salt' to make it sound similar. But this is simply a suspicion I have.
However so far as I can tell, the 'unfolding' (MQA decoding) will still leave the spurious anharmonics it used to 'hint at' the folded down into.
In itself, the above seems fine if people prefer it *and* they are being told what is going on. However, beyond the obvious question of what they are preferring it *to* not being the actual sorce version, it brings me back to the more basic question:
Since you can make similar reductions in the required stream rate or file size by 'open and free' methods that *preserve* the input HF timing alignment, etc, *without* any need for a 'decoder' or 'folding down the HF, what's the point MQA? Why have something that mimics the source when you could have the source?
I can see why the *music companies* might prefer MQA here. They can sell people the same music yet again whilst clutching the source to their hearts until they start offerring something 'better than MQA' in the next decade. But why would end-users prefer this merry-go-around *given any say in the matter*?