Hi All,
I need some advice on how to make Vivaldi remember all my input sampling rate to output upsampling combinations. Here an example:
I use the album “So” from Peter Gabriel:
On Tidal, I select the CD version with 44.1kHz and upsample it to 176.4kHz.
Then I switch to Tidal’s MQA version. This is 96kHz and it is kept at 96kHz.
When switching back to the CD version, all is fine and it will upsample from 44.1kHz to 176.4kHz.
Now I switch to the Qubus HIRES version of the album. This is at 96kHz and I upsample it to 192kHz.
When I then switch back to the CD version on Tidal. The 44.1kHz are upsampled to 192kHz. When I correct this to 176.4kHz and switch back to the Qubus version, this is then upsampled from 96kHz to 176.4 kHz.
What do I need to do, that 96kHz are always upsampled to 192kHz and 44.1kHz are always upsampled to 176.4kHz.
The Upsampler does not remember input/output combination settings . Further MQA cannot be upsampled anyway. If MQA is selected upsampling is disabled (see User Manual p.30). Note that MQA is decoded only by the upsampler and the rendering is carried out in the DAC.
However, if you set the DXD option to lock it will always output 44.1kS/s base inputs ( 44.1, 88,2, 176.4) as 352.8 KS/s and 48 kS/s ones ( 48, 96,192) as 384KS/s. That may be an acceptable compromise for you.
Pete, am I right in thinking that none of this is a concern for you since you prefer to listen without upsampling? Or am I conflating two different lots of settings?
Ben, you are right I don’t use any upsampling. This was simply in answer to the OP’s question whether or not Vivaldi Upsampler could be set to remember that 44.1 input would always be upsampled to 176.4 and 48 input always to 192.
I have posted several times about this in the past but briefly I am sensitive ( unusually so it seems) to spatial aberrations that upsampling can introduce. I suspect that this would not be of significance or even noticeable for most listeners as it manifests itself particularity with classical music recordings which , unlike other genres, are based around a recording of a real space within which a performance is taking place and where the approximate position of the performers, if not exactly known, is at least understood from experience of live concerts.
Joachim, in my experience, configuring the Upsampler to full DXD (locked) as suggested by Pete, and DSD as passthrough, yields the best sound quality in a full Vivaldi stack. Personally I found no sonic benefit to fractional rates below DXD for PCM sources.
I sometimes hear differences between the upsamling rates settings. Most often they are subtle. Sometimes I can identify them easily. I rely on my auditory intuition or gut feeling, which is smarter than my thinking mind in this regard.
To each his own of course (by the way, I can recommend a good Pinot Nior which will likely yield even better sound quality results if drunk 20min before your listening session ).
Objectively speaking though, in my view the Vivaldi Upsampler has more than enough horsepower to upsample redbook to DXD without detriment, yielding better sonic results than just a fractional rate below that, all else being equal of course; filter, mapping etc.
OK, let’s elaborate You choose an aspect of the upsampler technology to make a point - the processing power of the hardware built in. There is at least one assumption in the background - namely that higher output bit rate must equal better result. And as the hardware is capable, highest bit rate equals best. The omitted component in this logic is the listener. I return to that soon.
First I’m coming back to my question above. The DAC itself oversamples with enough processing power. If we would reduce the affair to the property of processing power a separate upsampler would be inaudible. dCS stated something to that end here:
I have seen your post below and am not versed enough in the details of digital audio processing. So I can’t argue at a granular level. My general point holds though - at least I believe…
A hifi system is complicated. It can be deconstructed in and understood through its constituent parts. That’s what we engineers have learned and are practicing. The human body and mind both are complex systems. They work differently. Here the dynamics between the system constituents are important. Prediction is hard. There are unknown knowns. When you add a human listener to a hifi system the resulting system becomes complex. To deal with complex systems it is useful to employ these tactics, wich we actually implicitly follow here in the community:
building relationships + working with patterns of interaction
sense making by collective interpretation
working in communities of practice
act, learn and plan at the same time
building on what works
Here’s an article approaching the system classification from a different angle. For the ones who enjoy gymnastics in logic:
My resulting hypothesis is, that reductionistic arguments are not sufficient to deal with hifi. The complex nature of the hifi + listener system requires the approach of collective sense making we practice here in the community. And to be useful that must include the ‘messy’ part - the listener.
Naturally, as there’s not much point in debating a subjective opinion
Objectively though, the higher the upsample rate the better the transient response for any given reconstruction filter. Integer upsample rates of 88.2k and above all have the same set of filters available on the Vivaldi stack, so, assuming there are no other detrimental effects with increased rates such as increased jitter (see footnote 1), it tends to reason that we should let the Upsampler do its job to its fullest extent possible.
Unfortunately, there’re no objective data published that I could find which shows whether or not the Vivaldi stack measurably performs better when a redbook track is upsampled to 352.8k (DXD) versus, say, to just 88.2k or 176.4k.
So, we’ll just have to leave it at that
Footnote 1:
The only objective measures I’ve seen online of a dCS Upsampler is Stereophile’s review of the dCS Purcell in 2001 - the only focus for measurement was whether there was increased jitter with increased upsampling rates - there wasn’t.
There was a more recent Chord Hugo M Scaler review (Chord’s Upsampler) in 2020 where Stereophile was able to show not just a transient response difference, but also a clear frequency response and noise floor improvements from Upsampling.
Both are worth a read if you’re keen on exploring the more objective side of things
).
).
).
”.
