Pushing the Computer Audio Boundaries :)

[jkeny]

Please participate in a listening test which is trying to determine (with some statistical significance) if ther is a difference in the sound of two files even if they are shown to be "identical" in computer terminology i.e bit-perfect. The thread is here http://jplay.eu/forum/computer-audio/ri ... ngs/#p4071

There are two files to listen to & compare to see if you hear a difference between them. I have asked the owner to download a third file which s one of the two files again (but we don't know which one) & we can increase the statistical significance above pure 50/50 guesswork.

Please participate - this is not just for "believers" - all should try this in a relaxed & genuine way. It's not a trick or test of anybody's system or ears, just a real attempt at getting some empirical data for analysis. Who knows what might come of it?

BTW, don't post your preferences here or on the JPlay forum so as not to influence others choices - this has to remain "blind testing" with no coaching. The results will be out in a week or so - then you will have some idea if this is all imaginings/placebos or there is something in it but unexplainable at the moment. Interested?

[jkeny]

I just had a thought - I don't know if upu can use the download links if you are not a member - I believe you can but I'll include them here. Here's the full post:

Ive recorded a quick acoustic guitar passage in my studio for this test, although a guitar is a little more difficult to distinguish, the differences are there and easy to noticed. Then I burned the file in a cd and then ripped.

One track is ripped with EAC in a laptop Dell xps i5 with Ac power, Firefox open.

The other one is ripped in same laptop with EAC with batteries power, fidelizer (audiophile settings) and Process lasso on High priority on EAC, no other program open.

Links (I am the author of the music so no law problems here)

http://rcpt.yousendit.com/1400034984/bf ... 62f6640aec

https://www.yousendit.com/dl?phi_action ... 0200000000

Also you can PM me your results & I will forward them to the forum member for inclusion in the results. Don't worry, it will be totally anonymous.

[Ivor]

probably like most people here I'll have to find the time and peace to do this justice. In my case it'll be tomorrow morning

[jkeny]

probably like most people here I'll have to find the time and peace to do this justice. In my case it'll be tomorrow morning

Sure, Ivor, there's no rush on this - the download link lasts for a week, I think.
I know I don't need to tell you this but others may benefit - the best way of doing this test, I find, is to just relax & listen to the files rather than trying to find the differences.

Thanks for participating & being part of helping to decide this controversial issue.

[Fran]

John - see my email. I've reported back directly to screenmusic via the jplay forum...

I'm afraid to even post a response here in case I influence people one way or the other!!!


Fran

[jkeny]

Hey guys,
These results & his comments were just post on the Jplay forum. Just thought I'd let you know:

50% said #1 is the Batt/Fid
50% said #2 is the Batt/Fid
#1 was AC
#2 was Batteries+Fidelizer+Process Lasso

What I learned from this test?
That one should zip a wav if sending it by email, the sound is more preserved, but it loose some quality. For accurate tests the file has to travel via pen drive or dvd rom data.
People can percieve a sound file brighter as a more resolution file, this test for some reason shift frequencies in the file made with Ac, making it brighter than the original. I am not saying that one file sounds better than the other only that, (becuase Ive recorded the guitar and know it by hand as well as my studio and compared the uploaded files with the original that stayed in my computer) file 2 sounds truer to the original version.
But with different material and more complex music the outcome is much more clear. The Seal, "Kiss from a rose" ripped with ac sounded like out of phase (in the original recording the phase in those choirs are not great) but it made it more evident, the batt/fid was a lot better, even the lows where more precise and everything was more focused.
I encouraged anybody to try this test and comment about the results, is a pain to ripped our whole library so it will be a good idea to do it right.

So this result is no better than guess-work would give us.
I've asked him some supplementary questions just to tease out this result & see if there is any thing significant in them:
- how many responses?
- How many described WAV 1 as brighter irrespective of whether they preferred it or not?
- how many answered the ghost question & identified it correctly?
- on your computer does the WAV1 file sound brighter?

[Claus]

After the ghost track was uploaded I could not guess which was the ghost. What does that tell me? Maybe we hear what we want to hear! ;)

[Fran]

This has bothered me over the weekend, ever since I read it.

Today I've copped on as to why that is. Unfortunately his response has strayed significantly from "do you hear a difference" to "please identify which one was using battery+fideliser".

This has kinda made a balls of what was shaping up to be a really really good test into one that is shot to sh!t.


John, I wonder could we do it again, but this time include the 3 files from the outset, and simply ask to identify if one of them is different and if so, which one? Unfortunately the horse has bolted now, so there's no point in going the route of the much simpler "is there a difference" - everyone will expect there to be a difference.

What a pity that this wasn't run slightly more tightly. All it serves to do now is shut down any investigation into the possbility that there can be differences in how the CDs are ripped etc.


Fran

[Claus]

I agree that it should have been run much more tightly from the outset. But I still think that a ghost track is necessary. I started out thinking I could hear a difference until I did some random ABX testing with the ghost track. I was bang on 50/50 when trying to identify the ghost track. This is in my belief the same "mind trick" that is used by many of the voodoo/cable/snake oil sellers at their demos: They play a track for 10 sec. then make the change, then play the same track againg for 20 sec. looking at you like "can you hear that? Or are you deaf?" Do they offer double blind testing at these demos? No! Why? Well we all know why.... :(

[jkeny]

Yea, it's all pear-shaped.
I wouldn't have a problem doing another test but beware there were 3 respondents from TirnaHifi (4 if you include me). So for a significant test we would need bigger numbers then this.
Also, the correct way of doing this is to copy the files to a USB stick & then have a tightly controlled prescribed way for these files to be copied onto their disks or maybe a better option would be to have more than one USB. What we are trying to avoid is the location of the files being of any significance. So there's a bit of work involved in setting this up & getting it right. Just saying we would need the numbers to make it worth this effort.

Of course people can do this test themselves by simply booting their PC/laptop into safe mode (minimal) & doing the "good rip" in this mode Vs the "bad rip" in ordinary full OS (possibly when running a virus scan or some other long process). This is said to make the differences more pronounced. I haven't tried it yet myself.