I hope you enjoy it!
Thanks for confirming much of what I heard when I put the stack through a workout the day after the Premiere of the Varese at Paragon back in October. Like you the first thing I heard was the absence of everything but music filling the room. Actually eerie at first.
During my demo I was able to easily swap back and forth between the Varese and the Vivaldi both connected to the Relentless Pre-amp. As much as I like my Vivaldi I couldn’t choose one area where I preferred it to the Varese.
I love Paragon. Amazing dealer. Larry is the best!
It’s not that Vivaldi isn’t a reference product, it’s that Varese is breaking new ground.
To be honest, as a Rossini Apex owner, I aspire to a Vivaldi Apex.
But I think this is what we all want dCS to do. Keep innovating and create more musical engagement.
Thank you for posting this Lee. I’m excited to have the chance to listen myself in NYC this week. (Thank you @AndrewS!) : )
To me, one of the most valuable parts of your review is this statement from David Steven:
“Looking forward 2025 will see a raft of new features being brought to Mosaic (our control app and network stack), Lina, Bartok, Rossini and Vivaldi as we roll out some of what we developed for Varèse. Some of these features will be usability improvements, some will be related to streaming services, and some will be audio path updates.”
Hopefully (am I overly optimistic?) DSD256 support for V/R/B will make the list…
I think that you need to prepare yourself for a disappointment on this issue. As I understand things the reason that there is no DSD256 is that the OEM streaming board in those units does not support DSD 256 or DSD 512. I can’t envisage a control unit (Mosaic) solution which David’s statement is limited to. At first sight this would need a hardware upgrade.
If I turn out to be wrong then the guys at dCS will have worked a minor miracle.
For me the key is this part of what DS said. The others are usability improvements.
@Lee - since you wrote this after your conversation with David any chance of expounding upon it? What do you mean by the term “audio path” ?
We will see what is his answer to your question, but everything done “touching” the audio path, can deteriorate or improve the sound quality…I guess dCS is only focussed on improvement…
unless Lee knows further information ( which he may well), I understand that future iterations of Mosaic will draw from experiences learned during the creation of Mosaic Actus. This implies that “Audio Path” is a similar feature to what Roon call “Signal Path”, that is a map showing what the signal source is , what units it passes through and ( as appropriate) if there has been any DSP applied.
As we are discussing this matter, when connecting Roon I have two options: Either going via the Upsampler or going via the Vivaldi Dac.
I always choose the Vivaldi Dac because Roon seems to "know it better ". But it is unintuitive because the logical front end should be the Upsampler…Anyone is doing the same or am I wrong saying Roon to connect to the dac ?
I’m not at liberty to say. It’s really best for the company to announce it’s future plans or announce when the upgrades launch.
I go via the upsampler and use the UPS to do all signal processing.
Roon for me is merely a library controller - all upsampling, volume levelling, EQ or anything else that impacts the signal path, is switched off.
YMMV.
Roon is 100% network-based and can only address clients over a network interface. So the only way to use Roon with a Vivaldi DAC is with an Upsampler or other network-capable device in between.
Hi Omni,
I agree with this with one minor variation:
You can set Roon’s DSP to downsample DSD512 and DSD256 to DSD128, thereby enabling V/R/B to play those files.
This further enables you to buy the highest resolution of the original recording today for your library (usually DSD256, as essentially no original DSD512 content exists), and when/if we have DSD256, you can just change the setting in Roon back to pass the file natively, as you originally suggest.
AFAIK, if Roon has recognised the Vivaldi UPS it automatically downsamples to the maximum rate the UPS can process. No need to set anything up.
I didn’t realize that, thx!
I will try with Rossini and see if it also does
Just a point of semantics, but I believe it is an important one. DSD512 is not “higher resolution” than DSD256 per se. There is no more audio information there whatsoever - even though the bitrate is doubled.
Similar to PCM, DSD sampling/bitrate increases are all about moving and changing the gradient of the low pass filter, not about “resolution” in the sense of, say a video signal, where a higher bitrate contains more pixels. To quote NativeDSD, speaking about a new DSD512 release:
I’d like to get smarter about this and welcome the help Andrew.
I just logged into NativeDSD and downloaded the same song in four different formats, yielding these file sizes:
DSD64: 61MB
DSD128: 121MB
DSD256: 242MB
DSD512: 484MB
(a) If there is no additional audio information why are the file sizes so different?
And/or
(b) Are you saying: essentially, the same information is repeated at, for example, 8 times in DSD512 v DSD64 and this enables an easier/more natural filtering process?
And/or
(c) If (b) is true then couldn’t the codex just instruct the DAC to create that bit string, since, as you indicate, there is no additional audio information?
And/or
(d) Other?
Thank you!
Haha. It’s a great question Richard. I have a fairly good understanding of PCM upsampling, but when it comes to DSD remodulation I understand the high level principles, not the practice.
I suggest you ask Jussi at HQplayer, whose software is used to perform the actual remodulation. He hangs over at the AudiophileStyle forum as ’Miska’. He can tell you exactly what is going on.
If (big if) it is conceptually similar to PCM upsampling the answer is essentially ”air” (i.e. nothing). A bit takes up the same amount of storage whether it is a 1 or a 0, whether it is part of the recorded signal or added padding. However the more air (null samples in between the actual audio samples) the higher up (further from the audio band) and gentler the low pass filter can be. While ”padding” has no sound, changes in filter profiles definitely do.
That’s about as much as I can tell you unfortunately!
Ok. I just want to make sure we are talking about the same thing:
A 192kHz 24 bit PCM recording, for example, is capturing considerably more information than a 44.1kHz 16 bit PCM recording. This is why those files would have different sizes.
Similarly, a 1-bit DSD64 sample is capturing 1 bit at 2.8 MHz, and a DSD256 recording is capturing 1-bit at 11.3 MHz, also considerably more information, hence the larger file size.
This is different from post-recording processing, i.e., upsampling and re-modulation.
I’m taking about the former, the amount of information at the time of recording. I’m not suggesting that the upsampling process, i.e., the latter, adds more information. That latter point has been discussed and clarified previously by dCS on this forum.
(Let’s pls drop DSD512 from the conversation as really no one records at that sample rate; though, per above, a DSD256 recording could be “Upsampled” to DSD512, while providing no additional information. In fact, many NativeDSD files are available at DSD512, but they were not recorded at that rate, which is clearly marked in the album notes. I personally avoid the DSD512 files.)
R