Frequency cutoff during resampling - what is best practice?

Longtime music lover / audiophile - first-time post, here.

Having a relatively large Hi-Res audio collection, but playback chain that only allows a maximum 24-bit / 48.0kHz, I have used multiple tools over the years to Resample higher resolution files down to 24-bit / 48.0kHz for playback.  I've used many tools and each have particular features and characteristics that may prove more or less user-friendly, but ultimately I'm after the absolute "best" result.  I have searched online in various ways, but have been unsuccessful, to date, in phrasing/framing my searches in a manner that has answered the following question:

When Resampling, is it best practice to retain full frequency spectrum information within the file, up to the maximum sample rate of the resultant, resampled file?  Or, is there some reason, scientific / objective / subjective, or otherwise, to strategically filter out frequencies in the upper range for a superior finished product?

Example/Case in point:  Here is a High Resolution audio track (see uploaded image/attachment #1), originally available in 24-bit / 96.0kHz.  From the Spek spectrogram analysis of the track, you can see that there is audio information present up into the ~43 - 44kHz range.

Using two different tools, dBpoweramp and Isotope RX 10, dBpoweramp (free software) resamples the 24-bit / 96.0kHz files to 24-bit / 48.0kHz audio file, and retains audio information all the way to the maximum 24kHz "ceiling" (apologies for lack of proper term, I assume) - see uploaded image attachment #2.  Whereas, the $299, highly-touted Isotope RX 10 (I'm in a 10-day free trial period) "cuts off" / (strategically?) filters out frequencies in the upper couple of kHz, leaving "room" (empty space) in that upper ceiling of the frequency spectrum - see uploaded image attachment #3.  

I asked this question on another audiophile forum, years ago, and was shocked that no qualified opinions or viewpoints were shared.  My intuition screams that the full retention of audio information up to the maximum frequency range would be best, no?  I mean, after all, it's leaving in sound information that audiophiles seek after and cherish, right?  However, having used countless other tools, like RX 10, which "cut off" / "filter out" information in the upper frequency spectrum, I can't imagine it is simply due to inferior Resampling technology.  I assume there are reasons/purported "benefits" for this.

p.s.  I'm not interested in anyone's rhetorical, soapbox rant about how you can't hear the difference, so it doesn't matter; or, "which one sounds better to you?" responses. 😄 

Thanks in advance!

On 9/29/2023 at 5:09 PM, U2Joshua said:

Longtime music lover / audiophile - first-time post, here.

 

Having a relatively large Hi-Res audio collection, but playback chain that only allows a maximum 24-bit / 48.0kHz, I have used multiple tools over the years to Resample higher resolution files down to 24-bit / 48.0kHz for playback.  I've used many tools and each have particular features and characteristics that may prove more or less user-friendly, but ultimately I'm after the absolute "best" result.  I have searched online in various ways, but have been unsuccessful, to date, in phrasing/framing my searches in a manner that has answered the following question:

When Resampling, is it best practice to retain full frequency spectrum information within the file, up to the maximum sample rate of the resultant, resampled file?  Or, is there some reason, scientific / objective / subjective, or otherwise, to strategically filter out frequencies in the upper range for a superior finished product?

 

Example/Case in point:  Here is a High Resolution audio track (see uploaded image/attachment #1), originally available in 24-bit / 96.0kHz.  From the Spek spectrogram analysis of the track, you can see that there is audio information present up into the ~43 - 44kHz range.

 

Using two different tools, dBpoweramp and Isotope RX 10, dBpoweramp (free software) resamples the 24-bit / 96.0kHz files to 24-bit / 48.0kHz audio file, and retains audio information all the way to the maximum 24kHz "ceiling" (apologies for lack of proper term, I assume) - see uploaded image attachment #2.  Whereas, the $299, highly-touted Isotope RX 10 (I'm in a 10-day free trial period) "cuts off" / (strategically?) filters out frequencies in the upper couple of kHz, leaving "room" (empty space) in that upper ceiling of the frequency spectrum - see uploaded image attachment #3.  

 

I asked this question on another audiophile forum, years ago, and was shocked that no qualified opinions or viewpoints were shared.  My intuition screams that the full retention of audio information up to the maximum frequency range would be best, no?  I mean, after all, it's leaving in sound information that audiophiles seek after and cherish, right?  However, having used countless other tools, like RX 10, which "cut off" / "filter out" information in the upper frequency spectrum, I can't imagine it is simply due to inferior Resampling technology.  I assume there are reasons/purported "benefits" for this.

 

p.s.  I'm not interested in anyone's rhetorical, soapbox rant about how you can't hear the difference, so it doesn't matter; or, "which one sounds better to you?" responses. 😄 

Thanks in advance!

 

Have you tried just leaving the source file Bit/Sample Rate as is and just letting the Music Playback Software/Server (ie..Roon, JRiver..etc) do the leg work for you in terms of Downsampling? By this I mean, within the playback software, you Map out the various higher Sample Rate recordings to be auto Downsampled on the fly to 24/48.

If you did do this and had ill effects, then what didn't you like about letting the software manage the Downsampling efforts?

As an FYI, I do this now with Roon. Everything gets either Up/Down Sampled to 24/192 no matter what it started out as within the source file. I don't notice any ill effects that keep me up at night but if you forced my hand to the fire I might say that I think using the default Bit/Sample Rate for each song sounds a little better than Up/Down Sampling it. 

Try downsampling with trial version of HQPlayer.

CJH