[DistortingJack]

Can you really?
I have here two versions of the same song, 1,000,000 by Nine Inch Nails.
The song was released in high-definition, 96k, 24 bit.
I just wanted to prove that with GOOD resampling and dithering technologies the differences are inaudible.
I used iZotope's sample rate converter, which is one of the best in the market. Want to check what it does? Go to this website and select "iZotope RX adv. (high steepness)".

SRC Comparisons

I chose this song because there are actually lots of ultra high-frequency energy in this recording, if you look at a spectrum analyzer. You decide if that information enriches the musical experience in any way.

The password is: head-fi

1,000,000, 96kHz, 24 bit
1,000,000, 44.1 kHz, 16 bit

And this is an upsampled version of the downsampled one, just to take your own software's sample rate converter out of the equation. It's the same information as the 44.1k, 16 bit file in a 96k, 24 bit shell. If your sound card does a clicking noise in between the first two, then try this file:

1,000,000, 44.1 kHz, 16 bit upsampled to 96kHz, 24 bit

Do try to create a blind test for yourself. Obviously, you need to have a soundcard that can actually PLAY the files at 96k!!! Be sure to check because many of them (many good audigy cards come to mind) do resampling on the fly which completely trumps the results.

[edit]:On a PC, use foobar2000's ABX test for about 20 times. You mustn't know your results before the end of the test. Anything below 14/20 is a guess.
On a Mac, there isn't an ABX test I know of, but I did it this way: open both files on Quicktime player, let them play at the same time (as close as possible) and then press cmd+~ to swap from one file to the other, maybe several times really fast without looking at the screen so you don't know what's playing anymore. You can manufacture your own test through this. I couldn't hear a difference anyway, so I didn't have to. For a moment I was sure I had a clear difference between the two, and I had picked the wrong one!
Take into account that this song has very large amounts of ultrasonic material. If you can't really say it sounds BETTER without them, but that it sounds slightly different maybe, that counts as a No because the question is "a difference in Quality" not a difference in sound signature. If you can distinctly hear something as "more accurate" or "more real" then vote Yes. [edit]

I thought for a moment to do a blind test for you guys and not tell you which was which, but it's too easy to cheat by watching a frequency analyser. I do think you owe it to yourself to finally have an opinion on this subject based on your own listening, which in the end is all that matters. And hey, if you can actually hear the difference, then at least you know you have better ears than theory allows.

.

 

[jojo_b2]

The files are Huge.

It's giving my connection a run for its money.

Anyway, I will try later here at the office (IBM Workstation + predator + ESW10) and also at home (Lenovo Y410 + DacMagic + Yamamoto + W1000). This seems interesting. =)

 

[DistortingJack]

Yes, the files are big because that's the way it's meant to be! No compression of any kind. FLAC is annoying on Macs, Apple Lossless annoying on PCs, and then you have different codecs, so it's better to remove as many variables as possible.
Hope it's useful!

 

[deaconblues]

lol, look at the sources listed by the only person who voted "yes" so far.
Head-Fi: Covering Headphones, Earphones and Portable Audio - View Profile: xkRoWx

 

[DistortingJack]

To each their own, I guess! Maybe he does hear up to 22kHz on his shures...
I'm glad I made the poll open though, that way you can weight the votes a bit better.

 

[Meliboeus]

Interesting, now i am downloading the files, i'll report my experience back. I have done similar comparisons myself, for example a couple of days ago i took my 24/96 vinyl rip of Nursery Cryme, downsampled it with SoX resampler to redbook standard and burned a cd audio, then abxed it against the original vinyl record ( i have a source switch before my amplifier, you have just to carefully adjust the volumes and playback position )...it was absolutely impossible to tell them apart, if there was any difference i couldn't detect, it was surely less noticeable than lossless vs high bitrate mp3...

 

[deaconblues]

Protip: Don't bother trying to tell the difference if your source cannot even play 24 bit / 96khz . Those 2 yes votes look really shady

 

[Meliboeus]

Tried on my computer system, honestly i couldn't hear the difference...my ears or equipment are not good enough...

 

[TMM]

Quote:

Originally Posted by DistortingJack /img/forum/go_quote.gif FLAC is annoying on Macs, Apple Lossless annoying on PCs

Then supply both? Lossless is lossless.

 

[DarKu]

it was disscused many times, there are no audible differeces between 44.1 and 96khz on whatever gear you want

 

[DistortingJack]

Quote:

Originally Posted by TMM /img/forum/go_quote.gif Then supply both? Lossless is lossless.

A .wav file IS lossless.

Quote:

Originally Posted by DarKu /img/forum/go_quote.gif it was disscused many times, there are no audible differeces between 44.1 and 96khz on whatever gear you want

Well, that's not what I've been reading on these forums as of lately. Mainly, people not believing in the theoretical basis explaining why the difference between 16 and 24 bits is nonexistent in a mastered track, and another set of individuals believing that ultrasonic material somehow influences musical perception. I would love to say the myths have already been dispelled, but some VERY respectable sound engineers and designers (Rupert Neve, for example) have been fooled on this. So when I heard this song, i thought to myself "this track has more potentially musical ultrasonic information and detail than anything else I've heard (and seen) before, the recording is very bright, textured and crisp, and it's been perfectly engineered all the way up to those frequencies. If I take care of the downsampling and the dithering myself and can't hear the difference, then for ME the myth is dispelled."
Another interesting thing to consider is that lopass filters going below 18kHz, let's say 16k, can be very noticeable on an A/B comparison, but they don't necessarily sound bad. I remember a music video I had downloaded and the music sounded perfect, so I tried to rip the music out into a wav, and I saw there that all frequencies above 13k had been completely removed. It lacked a bit of air, but it still sounded great. Also, moving around a concert hall can seriously roll off frequencies over 10k, but you don't hear people complaining it doesn't sound real enough...
We should realize we have reached the thresholds of the human hearing, whether we want it or not, and that an increase in quality will depend in what we do inside that threshold, and not anywhere else.

 

[JaZZ]

That's one of the worst recordings I've ever heard. Why this has been released in hi-rez is beyond me.

The stated «shedloads of ultra high-frequency energy» is a consequence of constant clipping and heavy distortion. The ultrasonic content is relatively modest, though, and -- for whoever might be sensitive to it -- certainly completely masked by the massive harmonic distortion.

This thread is something like another confirmation of my bias that people who think that there are no sonic subtleties in audio other than on sound-transducer level are insensitive to sonic subtleties. Instead of this recording you could have chosen construction noise or even uncorrelated pink noise.

Consequentially I can't hear a difference between the hi-rez sample and the downsampled version. I could in the case of another example (in a parallel, earlier thread) with well-recorded music instead of distortion. (I think I don't have to explain why I renounced voting.)


Quote:

Originally Posted by DistortingJack /img/forum/go_quote.gif ...and another set of individuals believing that ultrasonic material somehow influences musical perception. I would love to say the myths have already been dispelled, but some VERY respectable sound engineers and designers (Rupert Neve, for example) have been fooled on this.

And I thought this is the science forum.

.

 

[DistortingJack]

A very significant percentage of Nine Inch Nails' output is ambiental, quiet, and beautiful. They just released an 4-CD album with nothing but experimental, instrumental tracks, called Ghosts, some of which are softer, more textured, more creative and more subtle than most other "Progressive/alternative/independent (creative) rock/pop". This one happens to be a big hyper-aggressive electro rock song. I'll put some links at the bottom of the post if anyone's interested in the other side of this artist (of course millions of people know who he is but hey, apparently not all head-fiers do).

I didn't pick the song because it sounded great (actually I don't really like it very much), but it has some complex textures, and that's objective. Complexity is not related to niceness. You can't call that amount of high frequency content "modest" when it's as present as everything else in the spectrogram. The thumbnails on the first post are calibrated in Mels. If you switch to linear view, you can see that there is probably more ultrasonic sound energy than in any other frequency band.

Harmonic distortion is the most powerful generator of high frequencies available to a sound designer. Most acoustic instruments, be it a violin or an flute, will produce ultrasonic content, but those frequencies fade off very quickly, and are even more masked by the audible part of the sounds, since the sound is smooth, amplitude and phase coherent. The roughness of distorted sounds allows to separate the different frequencies more easily, which is exactly what we're going for in this sound sample.

We're not comparing musical tastes. We're testing the effect, if any, of ultrasonic overtone interaction with the audible part of the sound. It doesn't matter if those overtones are natural (like in acoustic instruments) or artificial (like in a distorted synth); all of those harmonics are still mathematically related to the fundamental, and should interact with one another in a similar way. Disliking distortion is not the same thing as not being able to analyze it. The most revealing test I can think of for frequency content in equipment testing is to listen to white noise, and it's not like I love the sound of it. Yes, construction noise could have done the trick as well, actually, had the recording been good enough and the ultrasonic content present enough. The musicality of ultrasonic material is not relevant to our perception of it.

Your post actually reads as "this music is too ugly and aggressive and annoying for me to be able to hear a difference between the two versions, and to me they sound the same, in my über-nice electrostats". I think that qualifies as a vote.

YouTube - La Mer, The Great Below - NIN
YouTube - Nine Inch Nails - 28 - Ghosts IV
YouTube - Nine Inch Nails - Ghosts IV - 34

 

[Ham Sandwich]

I scored an easy 5/5 in a Foobar ABX test with the 96 kHz sample and 44.1 kHz sample. Yet I can't hear a difference.

My source (M-Audio FW 410) would make an obvious change when switching sample rates quickly in an ABX style test.

Switched to using the file that was upsampled to 96 kHz and could not hear a difference.

 

[JaZZ]

Quote:

Originally Posted by DistortingJack /img/forum/go_quote.gif We're not comparing musical tastes. We're testing the effect, if any, of ultrasonic overtone interaction with the audible part of the sound. It doesn't matter if those overtones are natural (like in acoustic instruments) or artificial (like in a distorted synth)...

Like I said: You could just as well have used contruction noise -- according to your premise. I'm not talking about bad music, but bad recording quality. What you don't get is that for identifying sonic subtleties there have to be sonic subtleties in the first place which aren't masked by heavy distortion.

Edit: The recording is not quite as bad as I initially thought: Unfortunately I've loaded a Patchmix session with a too high level on my E-NU 1212M, which caused additional clipping. After all it's bad enough to keep my objection valid.
.