Undecoded MQA is demonstrably lower fidelity than redbook

Whilst nobody can argue against your subjective preference, undecoded MQA is demonstrably lower fidelity than redbook. Not that I have a problem with someone enjoying that. People like cassette tape, after all.

But please don’t decry mp3 for the very same reasons.

I’ve followed your posts, respect your opinion, and know you have a sound basis for this opinion. So please educate me on why you think this is so.

For example let’s narrow the field to recent new recordings that used MQA:

Wouldn’t someone listening to undecoded MQA experience the de-blurring done by the MQA ADC?

Then for older digital recordings with a known ADC, wouldn’t MQA’s remastering algorithms provide some de-blurring benefit compared to the original redbook?

I’m not a recording engineering nor an employee of any audio enterprise. Just an old guy with bad hearing who likes to listen to music instead of tinnitus.

This has been covered further up the topic, which I will admit has been long and meandering to say the least!

The compatible portion of the MQA signal is equivalent to about 13 to 15 bits at a sample rate of 44.1 kHz or 48 kHz. The loss of resolution is due to down sampling, dither noise, and pseudo-random noise from the high-frequency compression channel which occupies the lower 8 to 11 bits.

Again, I have no issue with someone listening and preferring the output. (I’m a SET amp owner, and am under no illusions that euphonic colouration plays a large part in my preference for it.)

On undecoded MQA? I don’t see how it could.

1 Like

Can you really hear that?
I read that point a lot, but in my comparisions I never could sense a noticable blur in a redbook format file.
So for me the whole de-blurring argument is only a marketing bubble (like so many others I hear in audio gadget promotions in the last 20 years).

1 Like

Clearly Bob Stuart’s statement is incorrect then?

@scolley Steve, First of all, let me say that I understand and admire your efforts to keep this thread on topic…as well as the Topic itself

But I feel it’s important to respond to Mark’s linked article above within the context of this thread…as I feel it gives the impression to other present and future readers here that essentially, “you can’t possibly be hearing what you’re hearing”

If you wish, a Mod can move these two posts out into their own thread…leaving your thread to its own very useful purpose

@anon55914447 Can I ask why you have chosen not to post this same article link in the multitude of DSD related threads [both Native DSD…and PCM converted to DSD] that appear here on Roon??..as the article is EQUALLY dismissive of DSD as a format or as a sound quality improvement tool??

This is demonstrably incorrect, in that MQA has been shown by the same people he references in his article that MQA contains information included in tracks sampled at 192khz and 352khz??..and that approx 70% of the Tracks available from Tidal’s MQA library unfold to 192khz…while there are several hundred tracks that unfold to 8-by rates of 352khz

What I also find ironic, is that the disputed claims of various 1-person “Blogs” are referenced as the “gold standard”…while the Peer-Reviewed claims in several AES Papers are treated as mere “marketing fluff”

As I say Steve, feel free to request a Mod to move these two posts to another thread…but as long as the article quoted above remains, then I feel it important to say that people can still have their own [positive or negative] listening impressions…in spite of an article essentially claiming that you can’t be hearing the ‘fidelity’ that you think you’re hearing


Why do you take critics on MQA so personally?

No, you misunderstand Benchmark’s statement, which is referencing the MQA first unfold. True lossless resolution wise, that is limited to approximately 17 bits at 96 kHz. Benchmark is not addressing any subsequent >96 kHz unfolds/renders, which are not lossless, nor does MQA claim them to be.


Again in deference to Steve’s thread, I will limit my response

@WiWavelength I would appreciate if you would refrain from telling me what I “misunderstand”…I have read and understand all of Miska’s and Archimago’s various postings…my response is in relation to the article link posted, not to their blogs

In, my my direct quote from the article above, the statement made is crystal clear, particularly the "at best" part

Likewise, please point at exactly where in THAT article, he refers to a first, second or 3rd stage Unfold or Decode

In fact, the only reference in there to decoding, refers to “fully decoded” as follows

The context of the relevant section of the Benchmark blog is MQA with a 96 kHz sample rate input. Look at the block diagram of the encoder.

Next, the resolution refers to bit depth, and the true bit depth of MQA is limited to 17 bits at the 96 kHz sample rate. That is a factual statement.

Now, not the context of the Benchmark blog, but if there are any subsequent unfolds/renders, the bit depth resolution at higher sample rates, such as 192 kHz, is even more limited – because MQA exploits the triangular narrowing of the actual information space at ultrasonic frequencies.


John Siau of Benchmark admits to taking only part of the MQA Patents, Papers and submissions but yet still arrives at the conflated conclusions below

You say the context is within “a 96khz sample rate input”…but a conflation of his statements will simply be read by most here and elsewhere [an objective I’m sure is not by accident]

And to put it mildly, I believe that to be misleading to most readers

Lastly, I will say again that all of these posts are IRRELEVANT to the topic of this thread…but more importantly to enjoyment of the music…and the focus of this and other MQA threads should be about “does it sound better to your ears”…and not about the alleged technical superiority or inferiority of the MQA format


Because I haven’t read them.

MQA work more or less like this:

1 - Get preferentially a 192/24 recording;
2 - Reduce this to 96/24 using of a very short/fast low pass filter;
3 - Take the 96/24 recording, and get the lower 24kHz band, at 24 bits, and noise shape it into a 48k/16b format;
4 - Put the result in a 48/24 container, but you still have 8 bit unused;
5 - Go back to the original 96/24 original. Get the upper band and re-sample it at 48kHz;

Now you have 192/24 packed into 48/24, and to play do the inverse order.

This is lossless because mathematics say so.
Very clever but completely useless, but to reduce network bandwidth while selling lossless streaming.
Basically (sound-wise) mp3 on steroids.

@Le_Baron_d_Merde MQA works nothing like that.


Yes, imagine if the master copy, say 24/192 is encoded down to 24/48, this is as good listening to a down sampled version of the master copy. This type of playback is meant for compatibility. To properly reproduce the master copy one needs to properly decode back to 24/192.