Vivaldi upsampler's future

A question while listening music through my Vivaldi dac and NBridge…When Qobuz or any streaming service shall stream 24 bit 192 khz or 384khz data…is an upsampler still useful ?
Or shall dCS upgrade to a version processing 64 bit 768khz or even more ?

Not sure to catch your rationale : what is your threat or concern ?

You do have a good point. If your source material is already at DXD rates or above, then yeah, the Upsampler doesn’t need to do any actual up-sampling in it’s current form.

However, when your source material is at 192kS/s rate or lower, then you’re going to benefit from the Upsampler doing its job.

So ask yourself this question; what percentage of current or future content is at DXD rates? I’d guess far far less than 1%. It’s taken the music industry nearly 40-years to get to this point where less than 1% of digital source material is available at native DXD rates or higher. How long do you think before its significantly more than 1%? 20 years? And so, do you think dCS should spend their R&D $$ focused on that right now? :grin:

I would just add that everyone should remember that upsampling does not add any actual musical information. As far as I can see it basically allows the digital filter to operate further out of the audible band. There has to be a point where any remaining artefacts from filtering become totally inaudible.

Is there any evidence that sampling rates need to be increased to 768kHz to achieve this? Surely as the sample rates become higher and higher, whilst arguably solving one aspect, others such as clock accuracy and associated jitter become more problematic than at lower frequencies?

1 Like

If upsampling higher than the current rate is not necessary, than it possibly answers my question: is an upsampler still necessary when streaming service shall deliver data to 24/768 ? …one day in a near future.
But someone might have a different view on this…your point was already very interesting to me.

You mean in the year 2049? I guess dCS will have an appropriate Upsampler to 24bit / 2.822MS/s rates by then :laughing:

Correct, a very small percentage because generally from a record label point of view there was/is no need for DXD as a carrier of commercially available recordings.

AFAIK DXD’s development was not as a music carrier for the public but to enable the editing in PCM of DSD masters with minimal artefacts caused by the process of transcoding.

The entire question of whether or not resolution above 24/96 is “necessary” is in itself the subject of conflicting views. My personal position is that this is just another version of the “numbers game” which infects the audio world from time to time.

An interesting read on the subject from today:

1 Like

Thanks for that link. Good to see additional data points being provided for this assessment. To paraphrase one of the commenters on that article, even my old ears can hear the difference.

In 2049…at least 256 bit…otherwise I would be disappointed…If I am still alive :sweat_smile:

Yes, I read the article although not being an AES member the full paper is not available to me. However I think it worth reiterating that this thread is about upsampling to very high resolution. That is very different to native recordings made at those rates. When you upsample, say, 16/44.1 to 24/192 that is not the equivalent of a 24/192 recording. The article is therefore only of tangential interest in regard to any future developments of a dCS upsampler.

Actually, the full paper is available to non AES members as well;

http://www.aes.org/e-lib/browse.cfm?elib=20455

(Just click on download it for free)

1 Like

Actually, the full Melchior paper is open access and can be downloaded by anyone. With respect to this thread, there is an intersectionality, and that is the utility of upsampling for filtering with minimal audible impact.

Thanks Greg and Anupc.

I have now read the Melchior paper which is a great overview of the history of sampling and filtering during the digital audio era. However it was not quite what I expected from Jim Austin’s introduction in Stereophile referring to a paper in which the authors used test tones to prove that listeners can discriminate between hi-res and redbook audio. That was obviously another paper altogether.

I would certainly not dispute the ability to hear hi-res audio. Even my aged ears can do that. However the question remains open as to the limits of bandwidth and , to some extent, bit depth required in relation to human perception. I think it reasonable to judge that bit depth exceeding the current standard of 24 bit is excessive. Indeed Melchior points out that its 144dB dynamic range cannot be used in practice. Yet I do occasionally read of audiophiles calling for 32 bit resolution which, for the recording and reproduction of music, is IMO absurd. So that leaves open the question of sampling rate and the point at which its frequency and relationship to digital filtering ( or absence thereof e.g. Marantz’ top line player) ceases to have relevance for human perception.

Same here. I thought the Sfereophile piece was (probably unintentionally) misleading about which paper was which. I tend to agree with your other comments. I suppose we could ask EMM; doesn’t their DV2 upsample 16x? :wink: