I remember the first time clearly. Mastering engineer and one of the founders of DSD, Gus Skinas, sat me down and played me two recordings: one in PCM and the other in DSD. I sat there with my mouth open: the instruments had body. The room had air. Notes decayed into silence without the faint mechanical graininess I'd come to accept as the price of digital and the detail. The inner detail was stunning.
The reason is architectural. PCM samples the waveform at discrete intervals and reconstructs it using steep digital filters. Those filters are remarkable engineering, but they introduce phase shift and time-domain artifacts that accumulate in ways your ears detect even when your brain can't name them.*
*Over the intervening years PCM sample rates have gone way up to the point (352,.8kHz 32 bit to be exact) where the differences have become minimal, but that's for another post.
DSD takes a completely different approach than PCM. A single bit toggling at 2.8 million times per second for DSD64 — 11.2 million for DSD256. The reconstruction filter is a simple, gentle analog affair that avoids the phase anomalies of brick-wall digital filters entirely. The result, to my ears, is a more organic presentation — particularly in the high frequencies and in the spatial cues that define depth.
At Octave Records we record and master natively in DSD256. Not because it's fashionable, but because after more than two decades of listening, I still believe it's the format that makes me forget it's digital or, for that matter, even a recording.
It just sounds live.
0 comments