Here are Mike Moffat's comments on DSD from a Headfi article. Mike is the cofounder of Schiit and developed the first ever standalone DAC back in the 1980s.
"I have never published my DSD opinions. Here they are. I say opinions
because the design of audio gear should adhere to hard science. The
user's response however, is totally in that user's psyche. When I worked
in Peru, there were tribes in the Amazon region who spoke in
vocabularies limited to grunts and delighted in eating insects they
found under logs. Then there are people like myself who prefer meat,
coffee, dairy, sometimes things green or fruity, starches, and lots of
salt.
In the early days of digital audio, multibit reigned. It
was suitable, but expensive, derived as it was from weapons guidance and
medical science. Note the use of the word science. Analog numbers were
converted to digital, and the reverse yielded the same number. Nothing
was averaged, no noise was added, no economic engineering geniuses were
allowed to make anything cheaper with smoke and mirrors.
The
earliest DACs were pretty marginal, but natural selection led to the
Burr-Brown PCM-63, an amazing multibit DAC, still pretty good today.
About that time, Burr Brown was sold to Texas Instruments. There began
to appear delta-sigma dacs, which is a fancy name for reduced bit width
DACs which used the above alluded to tricks of averaging and noise
shaping to make up for the data they threw away. Soon we had TI,
Wolfson, Crystal Semiconductor, Phillips, and many more manufacturers of
these (now marketed as audio - read dogschiit) DACs. Why stoop to make
them?? Simple - they're cheaper! Never mind they can't be used in
medical imaging or defense applications because of their inherent data
loss/hallucination. Too late, the audio customer had far cheaper gear.
The chip makers sold lots of parts.
Enter DSD, the ultimate
extension of this idea. More noise and less bitwidth. You get for free
with the bargain, the elimination of the nasty anti-alias filter effects
used in the recordings. Cool, huh. This idea works well just as soon as
every recording studio on the planet switches over. When that happens
(right), what about the old recordings like all of those from SACD days
of yore!! Oops, they are already recorded with the filter in place...
Unfortunately, they are the bulk of the current DSD catalog available.
Can you get DSD from iTunes?? Download DSD from Amazon?? Oh...
What
about 1, 2, 4, or 87.6x native DSD recordings. Yeah there's a few - I
really loved the Folsom Prison Castrati Singers doing Handel soprano
motets. My all time fave is the Orkney Island shepherd's Poems and Cries
of Ecstacy with the sheep. The plaintive cries and bleats of all
involved were immaculately suspended in perfect panoramic image. Even
the subtle sounds of the shepherds gently placing the sheep's rear legs
in their boots were clearly audible.
Nobody ever explained to me
how to design a multi-rate 1x, 2x, etc DSD DAC without a real expensive
adaptive filter. Do you optimize it for 1X? 2X? 5.76X? Trouble is, then
all of the other rates are compromised. Maybe the over $10K DACs do
that. I haven't figured out how to make an over $10K DAC yet, maybe
someone will teach me.
In conclusion - this is opinion, mine
with respect to DSD: How can I express just how underwhelmed I am.
Adjectives such as stillborn, faith-based, and ludicrous come to mind.
But
wait - I actually built the Loki DSD DAC! How can I be such a
hypocrite! The answer is that I will try almost anything once. If I
don't like it, I won't do it again. But I could be wrong - if servers
ever get big/cheap enough that iTunes and Amazon offer DSD downloads AND
major label music providers begin to provide native DSD recordings in
substantial numbers - then I will cook and eat a crow at RMAF.
Meanwhile, all you DSDers - enjoy the grubs!! Buy a Loki!"