MQA disappointing


It’s not clear who “they” are and definitions do vary depending on context, obvs.


One is an insult and one is a comment. That would be “even” in the school yard, but not for grown ups.


Hmm, I take it you’re related to Trump then.


I’ve just checked in volume one of my Shorter English (dictionary) and yes, words do have different meanings.

No relation to Trump, I do like the way he negotiates with the Chinese though.


in different contexts. Didn’t have to reference a dictionary to understand that.


Part of the problem with the debate in this thread is people keep “talking” past each other with respect to the definition of lossy/lossless.

In information theory, the amount of information in a signal is a measure of its entropy. When sampling any continuous analog signal there is always a loss of information if for no other reason that the measurement is band-limited. Therefore, every sampled sound is lossy to some degree (this is independent of whether the loss is perceptible by a human or not).

We can apply filters, band-limits, and correct sampling to mitigate the loss but there is always loss.

In the digital domain we encounter compression schemes. These compression schemes can be “lossy” or “lossless” but often these terms are not being used in reference to the original analog signal but to what is happening to the original bits compared to the reconstructed bits resulting from the compression/decompression operation. So, a digitally lossless encoding process, while mathematically accurate, doesn’t say anything about the initial digitization of the signal.

MQA claims that if it knows enough about the original ADC, it can compensate for some of the limitations of the ADC and do a better job at reproducing the analog signal at the input of the ADC. This is certainly technically possible but it seems impractical as multi-track recordings typically go through multiple channels which aren’t always preserved.

And herein lies the issue that should be the core of the debate. MQA uses a digitally lossy encoding scheme (this is a fact) to apply corrections to the high-resolution digital master to obtain a higher fidelity analog reproduction (and this now can only really be tested subjectively or, at best, through well-controlled psychophysical experiments).

In my opinion, MQA’s claim is tenuous when applied globally to all the MQA releases out there. Unless there was extra care taken to fully characterize the ADC’s used to create the high-resolution digital masters, or alternatively, the original analog tapes were digitally remastered in a way MQA specifies then I think there is little benefit to the MQA encoding. My personal experience listening to comparable high-resolution digital masters to the MQA release is there is no audible difference.

Finally, I will reiterate my opinion that to the extent there are improvements to the mastering process when using MQA these are negligible (at least in my audio system and my ears) to the differences I detect between different masters.

(Jeremy) #712

The audible difference is quite obvious. What is your setup?

Yes this is a fact. It seems incredible that anybody would even bother to challenge this.

Lossy encoding is detrimental to data integrity. In the case of MQA if you took the original audio and passed it through the MQA end to end processing several times (take MQA output and then re-process that output with MQA again) then it would probably completely ruin the signal in less than three or four cycles (yes it is that bad!). Most modern high end A to D and D to A equipment can go through up to 10 or even 20 cycles (D to A and back to A to D repeatedly) before there even begins to be ANY audible loss in quality.


Apparently not.

Why would you do that? What happens if you jump up and down on it?

(Jeremy) #714

As a test or proof of how lossy a type Processing or encoding is. If Processing A can be repeated 2 times before audible degradation but Processing B can be repeated 15 times before audible degradation then Processing B is much less lossy.

For example, take an old analog VHS casette - if you copy this and then take the copy and copy it again and so on - it won’t be long before the copy of the copy is completely unacceptable and unusable. Alternatively, copy a DVD or a Blu-ray Disc with appropriate bit perfect software on your PC and repeat that 1000 times and you still have a perfect rendition.

In the case of MQA it is all digitally processed however around 8 bits of dynamic range are simply thrown away to provide space for their Origami ( used to encapsulate high frequency noise)


OK, I think it’s been established that by one meaning of the word, it is lossy. That would be relevant if it affected the what you hear. Presumably MQA gets processed once, so processing it several times doesn’t give any indication of how it sounds in real life. It’s about what you hear, surely?

(Chris ) #716

I have just had a weekend of fine live music, I know what it sounds like and MQA does a fine job of reproduction for me.


They think 8 bit is more than enough to encapsulate the high frequencies and give you near CD quality (16 bit) at audio range. MQA is actually a 16+8 bit system (after it gets decoded) to me it is hardly called a true Hi-Res 24 bit system.

The unfolding mechanism has its drawbacks; it causes aliasing due to down-sampling after re-construction of high frequencies. The encoding is limited to 88.2/96k sampling, anything above is just up-sampling (rendering), no new information is retrieved, only compensate for timing. The minimum phase filters used are weak which allows ultra-sonic noise to slip through the audio range. The filters itself also introduces its own coloration and distortion not like the standard filters. People find these coloration pleasing but this is not the actual recording sound from the studios and definitely not what the recording engineer want you to hear.

All and all, if you are taking Hi-Res 24 bit 192kHz, MQA doesn’t even come close to that.

(Jeremy) #718

Thank you for this information. It is extremely enlightening.

I suppose MQA are UNIQUE in the industry to think that 24 bit audio is a waste of bit space and we can just throw away 8 bits. And all those manufacturers with their latest DACs that achieve an amazing 21 bit actual analog resolution at the output must be wasting all their design efforts. If only they talked to MQA and learnt the truth!

(Espen) #719

Actually, most of the industry thinks 24 bit is overkill at the listening end.


I guess they need to revise the current standard. Allocate 21 bit to match the current DAC resolution and just put 3 bit to encapsulate the ultrasonic noise. That will be nice :joy:


Roon PC Core -> iPad endpoint -> Woo Audio WA7 Fireflies DAC/Amp -> Massdrop HD58X Jubilee headphones

I’ve seen your comments about distortion and I’ve tried hearing it (intently) to no avail. When I strain sometimes I can imagine I hear it…but I don’t do this normally when listening to music…for pleasure.

I can tell the difference, in my setup, between my HD58x and HD6xx headphones which is supposed to be quite subtle but I can’t tell the difference between an MQA decoded and digital high-res master. I must not have classically trained ears…


Yep, just checking back in on this thread, Its like a full time job for some in this thread, lol. Well back to listening to music :+1:



(Music and Shawarma Lover) #724

Some people can type and listen to music at the same time :musical_note:

(ein) #725

It’s not truth.