Upsampling synced to sample rate, DSD independence from upsampling

Daphile music server has a feature that allows upsampling to be done based on the original sample rate, so if the maximum sample rate is set to 192K, a 48K or 96K file will upsample to 192K, but a 44.1K or 88.2K file will upsample to 176.4K. This keeping upsampling synchronized with the original rate.

Also, DSD should be kept at native rates and not affected by upsampling settings unless set to DSD over PCM.

Thanks for listening.


are you aware that upsampling brings no benefit at all?

1 Like

I think like Jnorris,
upsampling in the mathematically straight range - 48 to 192 and 44.1 to 176.4 is less error prone and should therefore sound better. I think a wish that the audiophile users of Volumio, a Primo for me, have. I use upsampling because it brings an audible gain in fine detail and spatial staggering.

from mathematical point of view, upsampling brings no benefits at all

you can’t improve sound quality just adding samples that were not there in the first place

Shannon-Nyquist sampling theorem is a mathematical theorem, it does not leave room for “subjectiveness”

It’s true it cannot be made better with upsampling. However upsampling does not exist for that purpose. Upsampling needed because we are working at the end with analog signals. When the digital stream enters the digital/analog converter chip there are several methods to convert it to analog signals. A (the better) group of the converters does this by simply setting the output voltage at the chip at every sample. So if sampling rate is 44.1kHz the this output voltage setting happens 44100 times in a second. These are voltage steps, so output voltage really is stepped. It can be seen with an oscilloscope. In order to not to amplify and listen to this 44.1kHz stepping noise there are analog filter stages after the converter chip. But these filter stages work on analog domain therefore they cannot “stop” noise just attenuate is. Unfortunately this 44.1kHz noise is not much far above the desired bandwidth (22.1kHz) so we cannot filter it with analog stages efficiently and it ruins the sound quality sometimes very heavily.
But… and this happens back from the first CD players, if we insert interpolated samples to the stream then this stepping noise will elevate to a much higher frequency domain and then we can effeciently filter it out with analog stages.
So upsampling is essential for NOS (No OverSampling) hardware. Quantity of these type of converters is rising because they sound better than other type of converters. I’m using one also so I would suggest this feature too. I would like it very very very very very much. :slight_smile:

Maybe if I could reach this portion of code which sets the upsampling rate and there I could query the source data samping rate then I could make a modification to the code on git repository. The only problem I can see now that actually upsampling setting change halts the actual playing track so it should be done before a new track begins. Therefore maybe this should be written into the initialization of playing the tracks.

1 Like