This leaves only 2 logical possibilities. One, MQA cheated by doctoring the master tape signal through an artificially boosting reverb chamber before encoding it to MQA. Or two, MQA itself as a digital system commits a random incoherent phase interference error in the time domain, perhaps also with added (reverb-like) delay, which erroneously artificially creates and adds a sense of space and ambience (of pretty poor quality, since it is so vague and amorphous), to the digital recordings it tries to reconstruct and play for you.
This of course would signify that MQA in truth has very bad time domain transient response inaccuracy and distortion - the very opposite of the very best time domain transient response that MQA itself claims, that MQA adherents applaud, and that MQA's nearly ideal impulse response misleadingly seems to imply.
In sum, MQA sonically features 3 salient departures from the sound provided by PCM. We will now proceed to analyze the technical bases for these 3 sonic departures, which will help us discover whether they are due to MQA's strengths or its failings, whether they are sonic benefits or defects. Of course, even if MQA has inaccurate time domain distortions producing sonic defects, it's a free country, so each listener is free to subjectively like MQA's sonic defects. But, dear MQA and MQA apologists, please don't insult my intelligence (or your own) by claiming that MQA's unique features are objective strengths that provide sonic benefits, if and when, in objective scientific technical fact, they are fatal flaws - fatal flaws that produce worse time domain transient response reproduction (the very opposite of MQA's own claims), fatal flaws that produce sonic defects, not sonic benefits.
Technical Analysis of MQA's 3 Sonic Features
General rule #1: if a good, valid technical analysis explains what you think you heard and say you heard, then it tends to scientifically confirm and corroborate that you really did hear that, and thus that sonic effect really exists and really has the properties you heard.
General rule #2: if a good, valid technical analysis shows that the device's behavior causing what you heard is in objective fact a defect, then what you heard as the sonic result was an inaccurate error/distortion, regardless of whether it subjectively pleased you or not.
General rule #3: if a misguided, invalid technical analysis falsely claims that black is white, that a device's factually defective behavior is the very opposite (superior behavior), then that false invalid analysis has no persuasive efficacy in convincing us that the admittedly different sonic effect from that device is superior 'better-different', instead of the worse-different sonic defect that it actually is.
MQA's 3 chief sonic features, wherein it departs significantly from PCM's sonics, are all easily explained by a correct technical analysis. This correct technical analysis then confirms and corroborates what we heard coming from MQA products.
In this case, a correct technical analysis shows that all 3 of MQA's unique sonic features we heard are in point of fact unique defects, not unique strengths. A correct technical analysis shows that MQA commits gross time domain distortion of the time domain signal waveform it reconstructs, and indeed uniquely worse distortions than other competing digital systems. A correct technical analysis shows that MQA's boasts and claims about its unique technical superiority are in fact backwards, the very opposite of the truth, since these claims rely on a backwards understanding of how digital actually works. A correct technical analysis shows that MQA's unique technical 'strengths', which are in fact MQA's unique technical defects, directly cause MQA's 3 unique sonic features, which are in fact MQA's 3 unique sonic defects and time domain signal waveform distortions. A correct and good technical analysis then explains how and why all this is so.
Let's now do just a quick overview of this correct technical analysis.
MQA's first unique sonic feature is that it eviscerates and even totally erases transient attacks. A correct technical analysis shows that this severe time domain sonic defect correlates with, and thus is tied to, a severe MQA technical defect, which moreover is a severe, unique technical defect that is trumpeted as a great, unique strength of MQA. MQA boasts of its unique technical feat of achieving extremely short hence nearly 'ideal' impulse response, implying that this proves MQA will provide better time domain transient response performance for the signal waveform than any other competing digital system. But this MQA claim is wrong, and indeed is so far wrong that it is completely backwards, the very opposite of the way digital actually works (strongly implying that some digital engineers don't have a clue in comprehending how digital actually works).
MQA's 'impulse response' is indeed extremely short, and indeed uniquely so, just as MQA boasts (even rivaling air itself). In fact, MQA's intrinsic impulse response is uniquely so very short (as shown in the Stuart/Craven AES paper) that it extends only 5us to at most 8us (at its -10 dB point) into each sample interval, thus does not even span the 20us sample interval (48 kHz sample rate). But this technical feature, instead of truly being MQA's greatest technical asset, is actually the complete opposite, its worst technical liability, which causes its worst sonic defect. MQA's very short 'impulse response' means that its coefficient function, which is the guiding template for the playback reconstruction filter, cannot gather the information required for reconstruction, not even over the entire span of one sampling period. Thus, MQA's reconstructed time domain signal waveform cannot effectively even span one sampling period. This might not be a severe problem when connecting the dots for a low frequency repeating sine wave. But it is deadly when trying to reconstruct a single transient attack that does not repeat (that's what makes it transient) and that contains high frequencies (that's what makes it an attack).
Consider a transient attack's sharp and narrow single peak, from a Bob Dylan guitar pick, just as we heard it in MQA's own MQA vs. PCM A-B. Suppose the signal waveform of this sharp, narrow peak happens to be initially sampled unluckily (just as most high frequency peaks are likewise unluckily sampled). The sample dots will be way down on the flanks of this transient attack peak, and the high peak itself will have occurred somewhere near the center of the sampling interval, so its high peak amplitude will not be represented directly by any sample dots at all.
How does MQA, with its very short, nearly 'ideal' impulse response, reconstruct this unsampled transient attack peak to play it back correctly, in the time domain that MQA boasts of such superiority in? It doesn't. Instead, MQA, with its extremely short 'impulse response' that's even shorter than a sample interval, can only see the sample dot information very near each dot, thus only at and from each one dot itself. Thus, MQA can reconstruct, for its time domain playback, only two isolated low amplitude narrow spikes, representing the two sample dots that sampled the low skirts of this transient attack peak, but unluckily entirely missed the high amplitude sharp peak itself. What then does MQA do with this high amplitude sharp transient attack peak that was there in the original signal of the guitar pick? MQA can never even see this peak, so it never reconstructs it, thereby utterly erasing it (or at the very least eviscerating the true hard peak strength of this guitar pick's transient attack).
In sharp and utter contrast, a good PCM digital system, with a correctly designed digital reconstruction filter, can gather information from a very wide field, spanning many, many sample dots. This required reconstruction filter ability, having a wide field of data collection, technically happens to look like a very long 'impulse response'. The data required to correctly reconstruct that poor unsampled, high sharp transient peak, is actually implicitly contained within the many other sample dots beyond the one sampling interval where the peak itself occurred. But a correct reconstruction filter design must have a very wide field of view (hence a very long 'impulse response') in order to even see all these other sample dots and gather their information. And also a correct reconstruction filter design must employ the correct reconstruction algorithm, to perform the complex curve-fitting calculations required, to correctly interpret this information from other far away sample dots, and to correctly use this interpreted information to reconstruct the correct curved path between sample dots, in this case correctly reconstructing the full true height and correct sharp shape of the unluckily unsampled peak transient attack.
Thus, what is in fact necessary for reconstructing a digitized signal waveform correctly, and accurately in the time domain, with excellent reconstruction of all time domain transient response events (including unluckily sampled transient events), is actually a filter with very long 'impulse response', the very opposite of what MQA erroneously achieves and boasts of.
This fully explains why, in MQA's own A-B comparison, the PCM version had the correct transient attack strength and hardness for all guitar picks, whereas the MQA version eviscerated and even erased entirely these sonically important guitar pick transient attacks. From the PCM version, the guitar picking sounded real. From MQA, the guitar picks sounded like weak phantoms or were even totally absent, and the later sustain tone of the resonating guitar string sounded as though it had emerged for no physical reason, like an effect without a cause, like a bell ringing without ever having been struck - in sum, a very unrealistic portrayal of a true guitar being played.
We have before us a fascinating technical riddle: when is the 'best possible' performance actually the worst possible performance? Answer: when the designer's technical comprehension is so wrong that it is entirely backwards, the very opposite of the technical truth. In this case, MQA attempts and achieves the best possible time domain 'impulse response', that is as close as possible to their engineering 'ideal' - but, because their comprehension of how digital actually works is so wrong it is backwards, their well-intentioned but naively wrong-headed achievement of nearly ideal time domain 'impulse response' performance actually produces the worst possible time domain distortion in MQA's attempted reconstruction of the original pre-sampled signal waveform. Indeed, MQA's actual time domain reconstruction is so bad that it is virtually equivalent to not having any time domain signal reconstruction at all, to not even having any reconstruction filter at all.
MQA, by effectively not performing virtually any reconstruction whatsoever, is effectively unable to even connect the sample dots to one another, to even bridge the gap between adjacent sample dots at all, with any type of bridge at all (let alone the correct bridge that correctly replicates the original signal waveform's curved path between each pair of sample dots). MQA commits the worst possible time domain distortion of the signal waveform by letting the reconstructed signal waveform completely collapse, and drop all the way down to nearly zero amplitude absence, between the isolated spikes (at each sample dot) that constitute MQA's maximally distorted profile, of the original signal waveform path between sample dots. That curved path of the original time domain signal waveform, between sample dots, is erased by MQA from its full correct reconstructed amplitude (and curved profile shape), all the way down to nearly zero amplitude (with no profile shape). That is indubitably the worst possible distortion, in the time domain, of the original time domain signal waveform.
MQA's second unique sonic feature is its airy, open quality on high frequencies. This is an attractive sonic quality that could seduce many listeners, especially in contrast to conventional PCM, which does sound too closed in. But as objective scientists we know that more is not always better (just as Goldilocks taught us all) - it is accuracy that is always best. And our suspicions were aroused because MQA's high frequencies also had other qualities that indicate problems or defects, e.g. sounding soft, defocused, diffused, phasey, and fuzzy smeared (not articulate, individuated, or coherently focused). This suggested that something was wrong with these high frequencies (perhaps the same factor producing their airier, more open quality).
A correct technical analysis explains everything we heard here. MQA actually generates several kinds of high frequency spurious garbage, and this high frequency spurious garbage added to the music makes the high frequencies sound more extended, airy, and open. This spurious garbage is of course very wrong for accuracy, but it does not sound all that bad, because it is modulated by the music signal, so it effectively sings along with the music, changing the sonic qualities of the music away from accuracy and toward distortion (albeit a distortion that some listeners might be seduced by).
The first technical cause of MQA's spurious high frequency garbage is that MQA's version of signal waveform 'reconstruction' consists of isolated sharp spikes modulated by the music signal waveform, not the accurate music signal waveform itself. These sharp spike shapes in MQA's waveform contain tons of spurious extra added spectral high frequency garbage. This high frequency garbage is modulated by the music signal, so its sonic effect is to distortedly change the character and qualities of the music signal, rather than sticking out as a separate sound. Note that this MQA sonic defect would occur even without any of the image leakage defects noted below.
The second technical cause of MQA's spurious high frequency garbage is that MQA's reconstruction filter, with its extremely short 'impulse response' approaching the 'ideal' of an impulse, allows in all ultrasonic frequencies, just as a brief impulse itself contains all frequencies including very high frequencies. MQA thus allows the spurious ultrasonic images of the baseband, previously caused by the sampling process, to remain in the signal waveform it plays for you. These spurious ultrasonic images in turn can cause multiple sonic problems and defects. They can reach down into the audible audio band, where they become directly audible, and would sound weird (unlike the music's genuine treble energy), because they are negative frequencies, that also spectrally progress backwards (with higher frequency image garbage actually appearing at lower audible frequencies). They could beat with the passband's genuine music signal waveform, thus causing interference patterns that would sound soft, defocused, diffused, phasey, and fuzzy smeared - exactly the sonic qualities we heard.
The third technical cause of MQA's spurious high frequency garbage is effectively a combination of the above two. Because MQA allows in the spurious ultrasonic garbage of sampling images, this spurious ultrasonic energy creates spurious squiggles in the music signal waveform that should not be there. These spurious squiggles represent extra added spurious high frequency garbage energy, and thus create extra added brightness distortion.
With all this added high frequency garbage, why doesn't MQA sound excessively bright? A correct technical analysis also explains this. MQA's inability to reconstruct sharp (i.e. high frequency) transient peaks between sample dots, discussed above, decreases MQA's brightness, since bright high frequency transients are never reconstructed, and instead are erased by MQA. The added high frequency garbage discussed here actually does indeed add brightness, thereby bringing MQA closer to tonal balance neutrality (by erasing bright portions of music and substituting bright garbage). But, it seems, even the addition of this bright garbage is not quite enough to bring MQA's tonal balance into equilibrium. It still sounded too dull, thanks to MQA's erasure of all those quick transient peaks from music's true signal waveform.
So MQA engineers cooked up a way to add more upper treble energy to the signal, and sure enough it sounded better. They erroneously thought that this better sound meant that adding extended ultrasonic energy to their 'transient time domain waveform perfect MQA system' was somehow sonically important to human hearing. But in reality the sonic improvement they heard was due to MQA's defect of natively being too dull, because (as shown above), MQA's native baseline 'reconstruction' of the music signal waveform was anything but perfect, actually being fatally defective by erasing high frequency transient information.
A correct technical analysis also explains further all the sonic qualities we heard associated with MQA's high frequencies: soft, defocused, diffused, phasey, and fuzzy smeared trebles. As shown above, MQA natively sounds too soft, even before any spurious high frequency garbage is added, and even before any interference patterns develop with this spurious added garbage. MQA sounds too soft in playing all high frequency attack transients that are naturally and correctly hard (e.g. Bob Dylan's guitar picking), because MQA totally erases or at least eviscerates this strong hard transient attack.
You see here a fascinating case study (and educational object lesson) showing how engineers who blithely and carelessly make an assumption, that happens to be so wrong it is backwards, can then lead themselves to go awry (indeed run amok) in misinterpreting subsequent empirical tests, and thereby be spurred on to design 'fixes' for this mis-interpreted 'problem', said 'fixes' actually making their system even worse instead of better.
In this case, MQA's design engineers see the virtually 'perfect, ideal impulse response' they have created in MQA, and blithely assume that this means MQA produces virtually perfect time domain reconstruction of the original music signal waveform (they never suspect that this 'impulse response' actually means the complete opposite, namely that MQA in fact produces the worst possible distortion of the 'reconstructed' signal waveform). They, assuming that MQA's native reconstruction and signal accuracy is perfect, then conduct an experiment wherein they deliberately add an extra dose of very high frequencies, and they find that this makes purportedly 'perfect' MQA sound even better. They conclude from this experiment that very high, ultrasonic frequency music information is indeed audible, and also beneficial. So they then devise a further twist to MQA's design, whereby it deliberately discards bits 17-24 of bit depth signal resolution, and uses those bits to instead encode ultrasonic music information, which they deem to be sonically more beneficial than the fine musical detail offered by bits 17-24 of bit depth signal resolution.
But this is actually yet another huge blunder by MQA. MQA's designers don't have a clue that all their thinking here is backwards, about their assumption, about their experiment, about their mis-interpretation of their experiment (and their inference and conclusion therefrom), and finally about their further twist to MQA's design, of adding yet more ultrasonic energy, devised in response to their backwards mis-interpretation of everything.