This review page is supported in part by the sponsors whose ad banners are displayed below
Given the market failure of physical SACD and DVD-A, I'd not blame anyone for wondering whether the current DSD-over-PCM aka DoP craze shouldn't have the DSD acronym stand for double strength dupe. If you've followed the loud cheerleading for the standard by Rob Robinson of PFO and Michael Lavorgna of AudioStream, you could fear that you're missing out on something truly major if you presently weren't streaming DSD. Yet anyone who thinks over why such files should or might be superior to high-resolution 24-bit PCM at 96/176.4kHz and bound to become the next big thing faces a few hurdles. On a technical plus there's the simple schematic borrowed from Ivo Linnenberg's site below.


It clearly shows how a DSD-enabled DAC processes such input signal more directly than even I²S and far more so than standard coax or Toslink. Bypassing a few stages would strike most of us as advantageous. Where our budding enthusiasm clouds over is contemplating how vastly higher out-of-band noise could possibly be better. As Igor Levin of Antelope Audio put it so plainly in the introduction of his Zodiac Platinum review, "traditional 1-bit DSD DACs suffer a sharp build-up of out-of-band noise as a result of using 1-bit Sigma Delta chips and being intrinsic to the DSD process. This noise is so strong that it easily overloads sensitive analog amplifiers into distortion. Hence DSD's HF noise is typically filtered with sharp analog filters set to between 30-50kHz. Unfortunately these filters introduce phase nonlinearities in the audible band and slow down transients to defeat most of DSD's benefits. Fortunately there is a way to reduce this build-up of ultrasonic noise and the damage done by the noise filters. The rate of noise build-up relates to the order of modulators used which relates to the oversampling ratio. The use of 64 x oversampling in traditional DSD requires 5th-order modulators to generate rapid noise build-up. The Platinum upsamples to 256 x and uses much gentler 3rd-order modulators. As a result we can use gentler noise-reduction filters to realize the true potential of DSD." Whilst high upsampling of PCM files enabling shallower reconstruction filters with more benign phase behavior has matured for decades now, the same isn't yet the case for DSD.


Next on the potential debit ledger is the provenance of DSD files. How were they derived, how much conversion from PCM was involved? Here we remember that the format's origins weren't mastering but archival uses. I'm told that even something as basic as gain adjustments on a DSD master require conversion to PCM. Then comes the vital question of whether—and how soon—this format will get applied to music which you, a/ want to listen to and, b/ don't already own. How many more versions of "Hotel California" do we need? Whilst for most of us in-depth familiarity with DSD files is still in its infancy, those in the reviewing community are dipping their feet. Here is what Polish contributor Wojciech Pacuła had to say in his review of the Mytek DAC: "When auditioning the Lumin I had no doubts that properly produced DSD files sounded more relaxed than even 24/192 PCM files... yet I still couldn't really name particular differences between DSD and PCM. The Mytek Stereo192-DSD allowed me a closer look at this without intermediaries. And now I found a quite serious difference. I could make simple comparisons mostly thanks to the Japanese Net Audio magazine which includes DVD-Rs in their issues with different file types. As I already have quite a few of these I was able to conduct a proper test. There is one more issue with DSD. It is still under debate whether regular DSD sampling—2.8224MHz aka DSD64—is sufficient or whether twice that rate should be used as studio equipment already does (5.6448MHz or DSD128). Some even promote the idea of quad DSD. In any case for now the first two are in use and both were playable on the Mytek. I used an Opus3 DSD Showcase sampler to compare DSD64 and DSD128 files..."

photo from Wojciech's review

"Comparing 24/192 PCM files to DSD was quite interesting. DSD better showed dynamic changes at the small end of the scale. PCM always delivered a head-on attack of the music. The difference between 192kHz and 96kHz was significant but still comparing apples. DSD was more of a pear. It allowed for a more sophisticated way of introducing the listener to a particular piece of music with more minute changes in dynamics... DSD also seemed to deliver more natural midrange and bass. I really loved the softness from its proper sense of attack which has phases not just a punch. So DSD sounded softer and more delicate... I think I finally understand why Japanese audiophiles love SACDs so much. This sound in many ways is the sound of a vinyl record whilst adding less coloration. The midrange is most important and carries a lot of energy to be very rich but not as euphonic as LP. Vinyl still wins on relaying sudden changes in macrodynamic. But there was another element which this test allowed me to study up close: treble and spacing. From the first time I listened to DSD files and then SACDs it was clear that this medium presents the treble with a little smear, without properly sharp definition... With the Mytek I could easily tell that cymbals on 24/192 PCM sounded more like real cymbals played live or played back from an analogue master tape. On the other hand DSD made them sound more like what I know from vinyl playback. DSD offers a more enjoyable pleasant sound but on the other hand a less realistic one. PCM tries to capture the true nature of the sound. None of these solutions are perfect and if I were to choose, I'd choose analogue tape."


Repeating the early days mantra, I find myself agreeing with Wojciech on DSD's treble aspect. To me the DSD files I've compared to their PCM equivalents sounded sweeter but also slightly fuzzy or soft. I even thought that Meitner's DAC with its FPGA programmed to operate as a true 1-bit chip bestowed some of that same aroma on its playback of regular Redbook files. But the upshot for the purposes of today's little ditty is this. If you're currently enjoying 24bit/88.2kHz or higher PCM files and your system is resolved enough to parlay their superiority on good productions, you're not missing a beat if your DAC ain't got DSD. Yes good DSD would seem to exhibit a different flavor but for now I'd not call it better, just different. It's important to say this if one views the current interest in DSD as mostly a marketing ploy to reinvent the high end once again. Should Igor Levin's Platinum DAC seriously undermine my opinion by going places with its DSD256 upsampling I've never gone before, I'll of course be the first to say so in its forthcoming review...