Multichannel MQA again, from April 2018 (Vol.41 No.4):
"Not MQA again!" I'm sure some of you will say, regardless of where you stand on it. The loud debate is often less about how MQA actually sounds and works than about opinions and, to an alarming degree, charges of dishonesty and bias. While Stereophile's John Atkinson and Jim Austin continue to strive to rise above all that, all I want to know is what value MQA might have for my own enjoyment of music.
This is not my first stab at this: In the May 2017 Stereophile I both recounted my long-ago first exposure to MQA, which predated its naming, and wrote of listening to stereo and multichannel MQA recordings, using three Mytek Brooklyn DACs. In the latter, I concluded that MQA "made a real and consistent improvement" but that "the differences weren't blatant; I couldn't hear them without paying close attention."
In his "Measurements" section accompanying Jason Victor Serinus's review, in the January 2018 Stereophile, of Aurender's A10 caching network music player/server, John Atkinson revealed "the A10's misapplication of the MQA reconstruction filter to non-MQA files stored on its internal drive."
In his comments accompanying Jim Austin's review of the Mytek HiFi Brooklyn DAC in November 2016, John stated that "for MQA, the filter is set to a fixed, minimum-phase, slow-rolloff type," and that to select any other, possibly more suitable filter, "MQA playback must be disabled." That information is printed in the Brooklyn's user's manual, but I missed it. I had presumed that the Brooklyn's detection of the MQA flag in the digital data would switch in the MQA filter automatically, and that in the flag's absence the DAC would revert to the standard PCM/DSD reconstruction filter, much as an FM radio's detection of a subcarrier signal in a station's/broadcaster's signal would turn on the stereo decoder. (Yes, that's how old I am.)
Because the Mytek Brooklyn, by design, dismissed my filter selections, I wasn't making the exact comparisons I thought I was making. Armed with this realization, I wanted to revisit those comparisons to hear if my conclusions would stand.
Playing high-resolution multichannel and CD-quality (16-bit/44.1kHz) stereo tracks, I began by again comparing the Brooklyn's three available reconstruction filters: Fast Rolloff (FR), Slow Rolloff (SR), and Minimum Phase (MPH), the last filter the default when MQA is enabled. With most tracks, the differences were apparent, but which filter I preferred was not consistent from track to track. The most telling differences were in the tonality and presence of solo voices and the perceived spatial relationship of the voice, its accompaniment, and the recording venue. But I've done this kind of thing before, and it's never completely satisfying: all the time and effort spent just reinforces my suspicion that there is no perfect choice. Still, my non-statistical preference was most often for the MPH filter, followed by SR and FR. What this means is that, even if I'd done my homework, I would have chosen MPH. This suggests my previous comparisons are still valid.
I decided to reassess my comparison of MQA (FLAC) with hi-rez (DXD) files by adding a third element: DSD files of the same nine multichannel recordings, all from Norwegian label 2L. As I found before, the MQA files sounded consistently smoother in the treble than the DXD. However, the corresponding DSD files had a smoothness that approached what I heard from MQA. On the other hand, the DSD files had the same spectral balance as the DXDs, while MQA sounded just a bit more forward, particularly in the upper strings. I repeated this three-way comparison many times and MQA generally sounded more open, with greater contrast between the players and the ambience of the recording venue, less noise within the ambience, and more detailed and silky trebles.
MQA was delightful, but its sound differed from the sounds of the DXD and DSD file formats—both of which also came direct from 2L, a proponent of MQA—in quite another way. The more I listened, the more I found the overall balance of DXD and DSD more even, more continuous, and more convincing in multichannel through my system with these tracks, all from the same label. The magnitudes of these differences were very similar to those among the reconstruction filters of the Mytek Brooklyn DAC, or among the mind-spinning seven filters of the Oppo UDP-205 universal Blu-ray player. I wonder if the subjective enhancements I heard from MQA might be obtained from the other two simply by careful selection of the filter from a library of same.
However, if I switched in room/speaker equalization with the DXD and DSD files (now downsampled to 24-bit/192kHz PCM for compatibility with the DSP), all niggling about subtle differences went out the window. I can't say that MQA files played through my system would not equally benefit from room EQ—that's not possible at this time—but whatever felicities MQA offers are dwarfed by the benefits of room EQ. The results from DXD and DSD were now superior in all parameters, including detail, smoothness, and balance, especially through the midbass and lower bass.
Today, I am less enthusiastic about MQA than I was. High-resolution streaming and downloads are now readily available, and the compactness of MQA files does speed downloads and conserve storage space. Storage space is cheap, however, and I stream music only for casual listening. MQA requires the purchase of compatible equipment, and holds the potential to eventually control all signal processing, such as room EQ. I don't see a need for it, therefore, and I hope it doesn't force the elimination of high-resolution, non-MQA downloads (footnote 1).—Kalman Rubinson
Footnote 1: See John Atkinson's "As We See It" in February 2018.
Footnote 1: See John Atkinson's "As We See It" in February 2018.
