2. digital
  3. 175065-151049

48kHz vs 96kHz: audible?

As a so-called audiophile, it is easy to lose one’s balance within many discussions and end up doubting, or at least questioning, whether that subtlety which one hears is real or imaginary.
 
Today, while engaged in a pastime, I was playing Holst’s "The Planets" in the background, but not at a low volume. I thought that it didn’t sound right. The strings in particular sounded a little abrasive. I noticed this on "Mars," the first composition, so it didn't take me long to perk up. On closer examination, I noticed that the DAC front panel was reporting 48kHz sample rate. I knew that this version of The Planets is 96kHz. Sure enough, JRiver Media Center (MC) was converting all PCM data (whether higher or lower) to 48kHz upon playback. I fixed the MC settings back so that all PCM rates play back at their native rates (up to the capability of my DAC), and all is well now.
 
Sometime in the recent past, whether due to an application or OS upgrade (of which there was one a few days ago), the MC sample rate conversion table got corrupted or reverted to a default configuration.
 
It would seem that I am able to hear the difference between 48kHz and 96kHz, at least under these circumstances. The difference was enough that I noticed it while passively listening (I was focused on drawing; the music was “background”) before I suspected a technical issue.

I wonder whether I could have heard this difference in a formal ABX test session? From my past experience with ABX testing, when the differences between the test objects are subtle, observations could easily have been obfuscated due to mental noise consisting of test anxiety, listening fatigue (to same passage over and over) and tedium. Whereas, in my case above, I noticed the difference when I was relaxed and focusing on something else entirely.

I am interested in thoughtful replies.

128x128mcdonalk
erik_squires
16,350 posts
 

Hi there,

So here are my thoughts:

1 - New DAC’s do a lot better with Redbook data than dac’s more than 10 years old.

2- While they do perform much better, there is still a measurable difference in performance of the anti-aliasing filters in the top octave so I’m not surprised it’s audible, but what it’s not is significantly better any more.

In the past, 96/24 music had a significant delta to 44/16 data. That’s mostly vanished, and any changes I hear can be replicated by a decent upsampling algorithm. This tells me that the mathematicians were partly right, that 44.1 kHz was good enough to encode music, but they were partly wrong in that there was a bigger delta due to how DAC’s handled different data density.

It’s time to stop buying DACs because hi rez sounds so much better, it doesn’t. Your DAC just sounds bad with Redbook. 🤣

TBC: I upsample to 96/24 when necessary, and otherwise am happy to stream 96/24 from Quboz.  I still prefer the DAC performance at this sample rate, but I don't think it's the data so much as the filter performance.

sandstone
358 posts
 

Isn't it possible that you are simply hearing  the artifacts of JRiver's extra step to downsample the file due to your software setting?  Maybe you're just comparing the effects of "native" to improperly refiltered/resampled streams.

Hard to actually draw any actual conclusions from this.

ja_kub_sz's avatar
ja_kub_sz
273 posts
 

I haven't gone that far with your supposition (48 vs 96), but I did blind A/B listening with a friend testing my CD 16/44.1 files vs 24/48 and 24/96 HD files and was able to discern between the two 90% of the time.

So is it possible, absolutely. Could there be another explanation, absolutely also.

But your experience is interesting.

curiousjim's avatar
curiousjim
4,875 posts
 

Last time I used JRiver, Hi-Res didn’t exist. 😀

I guess I’ll have to see what all it does these days. Oh drat, there isn’t an iOS app for it yet. Guess I’ll have to fire up the laptop.

 

fuzztone's avatar
fuzztone
5,635 posts
 

sandstone best me to the real point.

You cannot be sure of hearing things that do not exist. We know that the downsampling DOES exist by your description.

More to discover