Bob, Peter and Amir have worked in this field for decades and we have people with very little knowledge calling them charlatans.
Occam’s Razor applies.
3 Likes
There has never been any independent organization that has in any way shown MQA to be better than FLAC, WAV, or other lossless format (and vice versa, I might add). And, as long as MQA is a closed format, there never will be. The only thing we have about the technical side of it is what it’s inventor tells us. If you want to believe in his totally biased, unsupported spiel, go ahead. If you say your ears tell you it’s better, THAT I will somewhat buy into, as long as you don’t tell people that prefer truly lossless formats they are wrong to prefer them. All I know is that for ME, I do not in any way enjoy listening to MQA. But that applies only to me. There is no right side to belong on, nor a wong side to belong on. There are only individual preferences, and biases. The number one thing we need to rethink about MQA is why the heck are we letting it polarize us so much. Don’t we have the same goal, ultimately, to enjoy music to the fullest level we can, no matter how?? I want us all to have the biggest, baddest musical smiles on our faces possible.
The only problem being, and why I laugh at the PCM acolytes, is that the PCM is treated as if it’s bit perfect or lossless golden representation of reality when in fact it isn’t. We know it factually contains time/phase distortions caused by the way it was made.
So we have the MQA file which is different because it’s addressing these issues not by magic but by math and is storing it in the inaudible portion. So you can’t hear what you lost but you can hear what you gained.
2 Likes
Yeah, I’ve seen that before and the AES papers, interestingly the writer and I agree on something
I’ve listed some relevant AES papers and Meridian patents in the References box, but be warned: they aren’t easy reads, and appear to be deliberately incomplete or strategically vague in places.
That’s my issue. While I’m not a DSP expert, I’ve got previous experience assessing the performance of image compression routines. The term “visually lossless” bemused me then in the same way that “audibly lossless” does now. So not a complete layman either. It’s the deliberate obfuscation, the creation of new terms, missing information and proprietary secrets that stop a fair assessment been made. MQA’s strong desire to stop you seeing fully unfolded/filtered data is another issue.
That’s a stretch, in radio astronomy the earth’s rotation causes time smear issues, but that’s far from an audio issue.
And if MQA offered a licensed filter to “correct” PCM on the fly, perfectly feasible, for those who prefer it, then I’d be a lot more relaxed about it. It’s the secrets, lock-ins and license fees while sacrificing my own DSP freedoms that I have a problem with. Give me the filter of my choice and the chance to correct for my room/headphones any time over a closed system with restricted choice.
@Magnus_Back an attempt to explain the “mathematical magic” you seem to understand might help? Repeating MQA marketing claims isn’t likely to win this layman over.
2 Likes
Audibly lossless? You are again fixated on data when content within the data is what counts… Musical content that humans and perceive. This is all preserved in MQA streams.
1 Like
Some of MQA is straight forward filter math, like apodization. It is described well in that Sound on Sound article I linked earlier. It’s actually a pretty good article and describes the filter/math parts well.
The folding of data into the inaudible spectrum we don’t know how it’s done, that’s proprietary.
We only know the end result is intended to end up below what’s audible. See graphs in the SOS article.
Time domain corrections to adjust for the chain of equipment/methods used is obviously different for every instance. The method used is “just” counting backwards to get back to the original as close as possible. How they create these reverse filters and how good they are we can’t know.
Oversampling pushes the time/phase distortions well above what humans can hear. And, DSD recording pushes it much, much higher that we hear.
I can think of two pages of examples without trying, but this is OT and boring to most. Consider anything that influences time coherency. There are many physical processes as well as mathematical ones that disrupt the tight coherence of a signal. They can spread an individual signal from sharp to blurred, can affect the sharpness of a result when multiple time samples are processed together but differ in phase or time of arrival, or mix unrelated samples (for example time aliasing in circular convolution).
As @Chrislayeruk says, audibly and visually lossless refer to (testable) limits posed by psychoacoustic limitations. There haven’t been published listening tests for MQA, that’s definitely true.
1 Like
From the SOS article “Sony’s Direct Stream Digital (DSD) system used in the SACD format takes a long stroll in this direction, but to make the data rate practical, it operates with a one-bit PCM format which is difficult to process and requires very heavy noise-shaping to achieve a workable signal-noise ratio in the audio pass-band. So it’s still not an ideal solution.”
To get rid of the noise created analog filters are used after the DAC. Not a perfect method. Better than PCM but not perfect.
And this is just one of the MQA “tools”.
2 Likes
I wouldn’t put any faith in an article that refers to DSD as PCM. ALL DSD is PDM, not PCM, and needs extremely little filtering.
Deary, deary me! Audio time smearing? The most common cause of this is a result of bad speaker placement and/or bad acoustics in the listening room. Or maybe unfortunate combinations of amp and speakers.
AFAIK the time smear problems with DSP have been solved long before the advent of MQA.
2 Likes
It didn’t refer to DSD as PCM. Reading comprehension much?
Reflections from walls etc are several hundred milliseconds apart from the direct signal. It has been studied what the human hearing can separate or not. See some examples below. We have no problem sorting echos/reflections out from the straight signal but are much more sensitive to short phase distortion in the millisecond range. This is what MQA tries to correct for.
DSD improved on it some, MQA tries to take some more steps.
Bad combination of amp and speaker, I have no idea what you’re referring to there or how that would introduce time distortion.
As for DSPs, again I think you need to elaborate. Maybe you’re referring to Apodizing which Meridian has implemented in a number of products? (Thanks Bob, lol).
Getting tired of the nonsense some folks spread with no real knowledge.
1 Like
I wish I still had my 2000 page syllabus from my college acoustics course. There was some fun math in there. Might have helped the topic along more than marketing blurb.
2 Likes
Umm…read the text below figure 3 of your linked article. Neil’s comprehension on this point is fine.
3 Likes
?
“Figure 3: This graph offers an alternative view of the linear-phase (red) and minimum-phase (green) filters shown above, but this time as the energy or power distribution with a logarithmic amplitude scale. Note that the time offset between the main impulses is due to the nature of the FIR hardware used to implement the two filters and is typical of real-world executions.”
Keep reading below the explanation for fig. 3
Direct from YOUR post
From the SOS article “Sony’s Direct Stream Digital (DSD) system used in the SACD format takes a long stroll in this direction, but to make the data rate practical, it operates with a one-bit PCM format which is difficult to process and requires very heavy noise-shaping to achieve a workable signal-noise ratio in the audio pass-band. So it’s still not an ideal solution.”
Yes, it DOES indeed refer to DSD as PCM. Undeniably!
1 Like
I see it, it’s wrong. Maybe he meant PDM?
That’s all good in theory, but MQA can only address original issues with the A → D conversion and thus PCM distortions if it knows what they are and how they were created in the first place. This requires knowing what devices/tools were used on each recording/track. But clearly 99.9% of tracks are being batch processed, thousands at a time it seems, with no knowledge of what it’s actually trying to fix or address. They are simply slapping some preset/proprietary DSP across thousands of tracks at a time. There is no “Master Quality Authenticated” process going on here. If they really wanted to show off their process, they surely could have attached some metadata to MQA tracks specifying the provenance and other factors determining how they “Authenticated” and fixed the original master. But they don’t. Because it doesn’t exist.
1 Like