Why need to upsample?

I use DCS Bartok and I am very happy how it reproduces music. Why not try to improve however? I was wondering why upsampling feature cannot be switched off (unless it can while I can’t figure out how :slight_smile: ). We can chose in our DCS dacs to upsample to either DXD or DSD, we can chose MAP (1, 2, 3), we can change filters. Great. I can’t understand why we have to upsample. Does it have to do with RING dac construction principles? I read a review of a particular dac and it was reported to sound great in NOS (no oversampling) mode. It sounded even better with signal oversampled by external software – HQ Player in this case. Oversampling software built into this dac did not sound particularly good in comparison. In view of this I would love to read from DCS guys if ‘Oversampling OFF’ mode is being considered by any chance. If not – why?

1 Like

I can’t answer the why question, Piotr, but can add to the intrigue by noting that people with Vivaldi Upsamplers (rather awkwardly, given the name!) can choose not to have upsampling.

I’m not sure if it also applies to Vivaldi One users. I’d love to try it without on my Rossini, but just not possible. You’ve not missed a setting anywhere :slight_smile:

Thanks Ben. Glad to learn I’m not the only one who would like to try the dac in NOS mode, should it be available. I’m really curious what made DCS folks not to make it available for us.

Probably the fact that if you didn’t like the oversampling you wouldn’t purchase the Bartok or Rossini in the first place.

On the Vivaldi, it’s done by a separate extra cost unit you may choose to purchase or not depending on whether you think it improves the sound, so you get the same choice.

You think these are commercial reasons? Perhaps you are right, I have not thought about it.
But still would like to find out from DCS if it is technically possible to switch oversampling off.

Hi Piotr,

To answer this, I should note that Oversampling (and subsequently the Upsampling option on your Bartók) is directly tied to the filtering process - I appreciate that sometimes isn’t clear.

The short answer is that if you don’t Oversample (and perform the subsequent digital filtering to remove Nyquist images), you either:

  • Have loads of Nyquist images in the audio signal which are not filtered up to wherever your analogue filter is (100kHz in a dCS DAC) which would ruin the sound quality.
  • Or you run a very steep analogue filter at 20kHz-22.05kHz to remove these images, which would have real problems of balancing left and right channels and would be non-phase linear.
    This filter is then the same for all sample rates, so using such an analogue filter defeats the point of higher sample rate audio.

Using digital filters from 20kHz-100kHzby performing Oversampling to remove Nyquist images there, then having a gentle analogue filter removing anything above say 100kHz, is a very good way to address the problem of Nyquist images. Not Oversampling means you are subject to loads of images, or a brutal low analogue filter, both of which will harm sound quality.

For more info on filtering, take a read of the posts here. The whole series of posts explains why we do things the way we do in our DACs.

4 Likes

Hello James.
I really apreciate your prompt and detailed reply. I understand now that not upsampling would have a negative impact on sound quality and that is why you decided it has to be always ON. Makes sense.
If this matter has been already explained before I apologise for not searching for it long enough. Need some time now to ‘digest’ your comprehensive clarification :slightly_smiling_face:
Many thanks.