https://www.headphonesty.com/2025/12/dsd-audio-true-analog-sound-grammy-engineer/
What no-one seems to be able to explain – or at least, no-one has explained online in an understandable way, despite me looking dozens of times – is why conversion from PCM to DSD would improve the perceived audio quality. My Varese offers conversion of PCM into DSD, DSDx2, DSDx4 and DSDx8, and I can hear the sound “soften” with each step in the DSD multiplier, but why this is happening, and why it appears that I lose the sharper transients of PCM in favour of a “smoother” sound, no-one has explained. Nor is it really clear which one is closer to reality. It just seems like a trade-off between the higher-resolving-but-brittler-sounding PCM, and the "smoother-but-less-transparent” DSD. More to the point, what is the perceivecd benefit of DSD vs its increasing multipliers? What am I gaining versus giving up byh choosing plain old DSD vs DSDx8, for example? The bare fact you can increase multipliers, to a layman, suggests you are getting “more” of the thing you wanted when you chose DSD over PCM in the first place, so logically would you not want the “most” of it that’s available, i.e. DSDx8?
It would be helpful for someoone to explain the science of what is happening to the audio path, rather than just give subjective views on what it “sounds” like, so we can decide for ourselves on what trade-offs to make based on what is actually happening to the signals when we make these choices, in layman-speak.
3 Likes
The “95% conversion to DSD-like” mentioned early in the linked article seems to be alluding to merchant-silicon mass-market Delta-Sigma DAC chips (like ESS etc), but with dCS, I don’t think it’s universally accepted that transcoding PCM to DSD sounds better 
In fact, many of us (maybe most?) keep PCM in the PCM domain and DSD in the DSD domain, at least as far as our dCS DAC configurations are concerned - “Upsampling to DXD” on the R/B/L, “DSD Pass-Through” on the Vivaldi, and “PCM to DSD Conversion OFF” on the Varese.
That said, I agree there isn’t much casual coverage on the topic of PCM to DSD conversion, but it’s a topic that’s been covered extensively over the years in Audio Engineering Society Papers, especially in the early 2000s - https://aes2.org/publications/elibrary-browse/?text_search=DSD
In any case, I’d imagine if anyone can explain it well, it’d be dCS considering their deep expertise in DSD from the very beginning.
5 Likes
I can weigh in with an observation here. I work on live to two track recordings of jazz and classical sessions in Atlanta. We use AKG 414-BULS mics and split the mic feed into both PCM and DSD recording boxes. To my ears and my partner Nick’s ears, the DSD sounds closer to the live event. We have noticed this across many different sessions with varying microphones, mic cables, power sources, and musical instruments.
3 Likes
Agreed.
The insanely high sampling rates of DSD128 and DSD256 in particular enables one to approximate the source wave more closely. This is one reason why there is a small set of studios putting out great recording with the Pyramix workstation.
I also agree with @Anupc’s position and don’t do much upsampling (i.e., regular PCM to DXD) or cross-conversion (i.e., PCM to DSD), personally.
I do spend a lot of time going through the notes to determine how the file was originally recorded, and typically buy that version.
For example, Native DSD labels if the file was recorded in DXD, DSD64, 128, 256, etc., and I generally buy that version, and not the upsampled one.
As discussed on other threads, virtually no DSD files exist natively above the DSD256 format, and even the 256 format represents <1% of available music, alas, though it has a rising share of new classical music.
1 Like
Ditto.
Keeping the basic DXD upsampling setting in the BartĂłk for what concerns PCM, and now only buying digital music at the same resolution it was recorded.
And when buying DSD files, I stick to DSD128 most of the times.
Not only because of the dCS limitation (but Varèse), rather because it seems to be enough for noise rejection, while not weighing too many gigabytes on my NAS.
2 Likes
Lee. That’s very interesting.
Can you pls share more about your recording and mixing setup/approach?
What label are the recordings released on?
Thx!
2 Likes