Hi-rez to Redbook and back again ...

[fas42]

Just curious and perhaps it's already been asked or mention made of it ...

Has there been any substantial testing of a "difficult" track recorded in good quality Hi-rez, which is next transferred to Redbook, and then upsampled from that back to the same Hi-rez format to create effectively a second generation digital copy of the original, and then comparisons made?

I'm talking here of doing "proper" AB testing of the original and digital copy from a music server through the best DAC to see if people can pick an audible difference ...

Frank

[Bruce B]

I thought this was the whole premise of the MM paper.
I'd certainly be willing to put up some samples of 2 hi-rez files, with one being the down/upsampled one for everyone to listen to.
So what "difficult" track would one propose? I'd probably start off with a quiet solo piano piece.

[Vincent Kars]

The Meyer Moran paper: http://drewdaniels.com/audible.pdf

[amirm]

I think it will be a fun exercise no matter if it has been tried. I suggest something with notes that decay to nothing. If the piano track has that, then it is good. Otherwise some quiet string guitar would be nice.

[fas42]

Had a quick look at that paper, and it is not doing what I'm talking about. In that test the comparison is made while an extra layer of complex electronics is in operation, and that is exactly what has to be taken out of the equation. In other words, when doing the AB test the only electronics functioning at that moment is a music server with two tracks to be chosen between, a high quality DAC and high quality headphones. Absolutely nothing else is involved when doing the actual comparison!

Frank

[garylkoh]

What Frank has proposed should be even more transparent than the "Meyer Moran" test. Some of the best upsampling/downsampling algorithms are two-pass, and hence not executable in real time. If Bruce does the downsampling/upsampling, I'm sure that he will do his best job, and I'll bet a that no one will be able to hear a difference - or to reliably ABX identify the tracks.

Publish your ABX logs!!

[Phelonious Ponk]

I've decided not to trust myself with these kinds of tests. For one thing, I don't expect to hear a difference, so I probably won't. But more importantly, I just don't have the patience for it. What am I listening for? Did I hear a pre-echo (what exactly is a pre-echo?)? Was that reverb tail longer than the last one? Did it break up just before it faded to black? Wow, that guitar passage is gorgeous. What was I listening for again?

Maybe if the music really sucked I could care; anybody got any hi-res polka?. This much my listening has proven, to me, anyway: These differences, if they exist, are so small, that unless my system is otherwise perfect, and it's certainly not, I have much more important things to worry about.

Tim

[Bruce B]

Here they are. Solo piano at 176.4
There are 3 tracks. Tell me which one/s has been down/upsampled! No cheating!!

Sample Tracks

[audioguy]

These differences, if they exist, are so small, that unless my system is otherwise perfect, and it's certainly not, I have much more important things to worry about.

Tim

+1

A friend and I call that the "squint factor". If you have to close your eyes (or almost close your eyes) and listen REALLY, REALLY, REALLY hard and you are STILL not 100% sure you heard a difference, invest your time and energy elsewhere. (like room treatment or digital room correction, or better yet, new music!!).

I have saved a ton of money by following that advice (Except when I don't follow it)

[Bruce B]

I take it by the lack of responses that it's impossible to tell which file has undergone sample rate conversion?