Given the modern dCS are not my first, there would be so much to say. The short answer is, embrace the choice these units offer. Do not use a setting because the manual says so, because someone else said so, or because you just want to “pick the best and be done with it”. There’s no such thing, even less so now than with the dCS units from 20 years ago.
Unfortunately, it makes sense to try settings for individual recordings. I’m serious. Having said that, one will tend to get a feel for what’s going to work best, and make a decision quickly. Some impressions I got over the years:
Back when everyone said, integer PCM upsampling sounds best for redbook CD (RBCD), and later DSD upsampling, they were wrong. I later made audiophile acquaintances who preferred 24/192 upsampling with the older units just like me. And yes, there’s something inherently wrong with non-integer upsampling. Which is probably why it makes some recordings sound, well, as if they were something other than digital playback - which can be a good thing.
These days, I tend to listen to a recording without upsampling to get an idea, and decide from there. In general, DSD128 upsampling works best for RBCD, and mostly for all 16-bit material (i.e. DVD-Video rips which are often 16/48 PCM). That’s not so obvious given most even-numbered high-resolution PCM (i.e. 24-bit recordings or remasterings in the 48kS/s group, which is 48k, 96k and 192k) tends to sound best upsampled to 384k, probably because the DAC needs no aliasing filter playing back DXD sample rates. Having said that, there are recordings in the even-numbered 48k group that’ll sound better upsampled to DSD128, which in turn uses an additional filter stage. Recordings that heavily rely on timbre but not the differentiation thereof, such as some singing or string playing etc., or music that is about “flow”, i.e. may otherwise seem static rather than move forward. Which is why DSD upsampling rarely sounds “better” for instruments such as piano, but even there, it’s best to try.
One thing about DSD and DSD upsampling I’ve learnt is it doesn’t necessarily sound more correct, but more beautiful. In fact, compiling a list of pros and cons comparing DSD to PCM, what’s happened to me more than once is the pros list was longer on the PCM half of the sheet (in terms of HiFi criteria), yet the conclusion was I preferred the sound of DSD.
In short, I find I have a tendency to want to listen to CDs and such upsampled to DSD128, and high-resolution PCM that’s 48k, 96k and 192k upsampled to 384k.
Now, none of this applies to high-resolution PCM recorded or remastered in the 44k group. Without trying, there’s no telling whether a recording will sound best upsampled to 352k or DSD128.
Listen for anything sounding “phasey”, including changes in soundstage depth, instrument placement etc. Upsampling is only worth it when it’s an improvement sonically - sounds like a truism, but no intension of irony here. You’ll find there are people who don’t like upsampling at all - I’m one of them if only for perfect recordings. Of course, there’s (almost) no such thing.
When it comes to filters, I’ve noticed the choice seems to have more to do with the equipment one uses than the recording being played back. Back when I still had a ASR Emitter (a nicely warm-sounding, comparatively slow MOSFET-based amp), the Gaussian filter sounded as good as or better than minimum phase for most if not all PCM playback. With fast-settling amps such as Spectral, anything other than minimum phase won’t sound natural with PCM. Mind you, other filters may sound fuller-bodied etc., but realistic, no. You may still want to try other filters playing back amplified music (such as rock etc.), simply because the point there isn’t “natural” sound but the creation of a type of sound (which the artist may have had in mind, but still, to one’s liking).
The same goes for DSD filters. Using amplification such as tubes that will naturally (additionally) filter out of band noise, DSD Filter 1 may sound most transparent. But with wide-band megahertz amplification such as Spectral, while it may sound more transparent, it’ll make the listener turn the system off and find migraine medication. Filter 5 all the way for DSD there, a great compromise, in particular also for DSD64, which along with RBCD I’ve always thought of as the most problematic two formats (due no doubt to the early 44k and 64x44k standards).
As to upsampling DSD, whether with a dCS Upsampler or a media player, even offline converter such as HQ Player 4 Pro, listen carefully. It really only sounds better when the DAC “needs” the higher sampling rate, i.e. when, as with the early Lampizator DSD board, DSD is being filtered passively - that’s the one exception where remodulating DSD may make sense (lack of treble extension, phase issues…). With DACs that filter digitally, usually so-called DSD-wide designs, unnecessarily adding conversions, remodulations and/or filtering can’t possibly improve matters, and it doesn’t.
Now, I’m using loudspeakers with time alignment and phase-coherent filters of my own design, so I may be guilty of nerd view (professional deformation listening to filters/phase issues).
Something else: when I’m not listening attentively, such as now that I’m typing this post, I tend to like DSD128 upsampling because it appears to have positive side effects that aren’t easy to pin down, maybe dynamic compression in the higher frequency range, a smoothing-over of transients, creating a greater sense of palpability (more density to the texture, or emphasis on timbre). This becomes particularly noticeable when one turns the volume down.
Whatever you do, be sure to engage “Upsampler DSD Pass Through”. If for any reason whatsoever one needs to convert DSD to PCM, be sure to do so offline (e.g. DSDMaster Zero Phase filter). No realtime conversion is going to get that right.
As to the bit-mappers, while I agree with everyone that the new mappers (1+3) sound better, which one prefers may again have to do with the equipment one uses. A slightly higher proportion of second- or third-order distortion is much like listening to different qualities of residual digital glare. Map 3 sounds more like the harmonic distortion or bloom of a tube amp, so unless a recording sounds glary, I find it makes for a more natural or, for lack of a better term, “analogue” sound (by which I mean the sound of a microphone feed rather than vinyl playback). But I can easily see it may be too much for some equipment to handle, e.g. amps and tweeters (or rooms lacking acoustic treatment etc.).
My two cents worth…
Greetings from Switzerland, David.