Head-Fi.org › Forums › Equipment Forums › Sound Science › BLIND TEST: Lossless vs. MP3 320
New Posts  All Forums:Forum Nav:

BLIND TEST: Lossless vs. MP3 320 - Page 3

Poll Results: BLIND TEST: Lossless vs. MP3 320 kbps

 
  • 26% (38)
    A is Lossless
  • 42% (61)
    B is Lossless
  • 30% (44)
    No Difference
143 Total Votes  
post #31 of 107
Quote:
Originally Posted by xnor View Post

Quote:
Originally Posted by bowei006 View Post

THanks to brooko. Now what do these numbers even mean?

You mean those in the abx logs? Like this: 00:45:00 : 08/10 5.5%

It's the current time, number of correct choices (8), number of total trials (10) and chance that you're just guessing (5.5%) which should be below 5%.

The 5.5% represents the probability that the results could have been 8/10, 9/10, or 10/10 if you could not actually perceive any audible difference in the files. It is called the p-value. Depending on what alpha value was chosen prior to testing, 8/10 may or may not be statistically significant.
post #32 of 107

i get it now. thanks. i had to look at a log from page 2 first :D tahnk you :D will try out with my own files. I'll do a lossless and 320kbps mp3 first and then lossless vs 160kbps mp3 tmmrw :D 

post #33 of 107

The dead giveaway was the richer bass on the lossless track. It took about five seconds of each track to figure which was which.

 

But the excerpt was very good.

post #34 of 107

Nice test! I chose B after listening for 20 seconds each song. IMO B is more detailed and clear. Hope this is the right one.

post #35 of 107
Quote:
Originally Posted by poikkeus View Post

The dead giveaway was the richer bass on the lossless track. It took about five seconds of each track to figure which was which.

 

But the excerpt was very good.



Frankly, your comments aren't worth a hill of beans in this thread unless you post logs. rolleyes.gif

post #36 of 107

A.PNG

B.PNG

post #37 of 107

how is that helpful to this thread?  :/  you're creating an expectation bias.  especially given that the rolloff your graphs show only BEGINS at ~20khz, where none of us can hear anything anyways
 

Quote:
Originally Posted by larry View Post

*spectral analyses*

 



 


Edited by El_Doug - 2/16/12 at 9:43am
post #38 of 107
Quote:
Originally Posted by El_Doug View Post

how is that helpful to this thread?  :/  you're creating an expectation bias. 

Exactly that's the reason why I wouldn't add polls to such topics anymore. Instead just let people post abx logs.. everything else doesn't matter.
post #39 of 107
Quote:
Originally Posted by xnor View Post


Exactly that's the reason why I wouldn't add polls to such topics anymore. Instead just let people post abx logs.. everything else doesn't matter.


Agree - anyone else getting thoroughly amused at the posters who chime in with a one or two liner usually consisting of a "well I listened for X secs, and X is more noticeable in A than B" response?  Personally I don't believe anyone until I see an ABX with at least 15-20 results wink.gif.

 

And to the prospective testers (especially the 'newbies') - it's not a crime to fail the test.  mp3-320 / aac256 is supposed to be pretty transparent - that's the way it's designed.  If you can't reliably tell the difference - don't despair - you're really not supposed to anyway. 

 

post #40 of 107

About to ABX. I chose full instal but i didn't get ABX installed with it :/ don't know why. whatever. i just downloaded it off the site and moved the .dll to the components folder and voila.

post #41 of 107

 

foo_abx 1.3.4 report
foobar2000 v1.1.11
2012/02/16 14:51:46
 
File A: C:\Users\Bowei\Downloads\A.wav
File B: C:\Users\Bowei\Downloads\B.wav
 
14:51:46 : Test started.
14:55:07 : 01/01  50.0%
14:56:21 : 02/02  25.0%
14:57:17 : 02/03  50.0%
14:57:40 : 03/04  31.3%
14:58:02 : 03/05  50.0%
14:59:34 : 04/06  34.4%
15:00:31 : 05/07  22.7%
15:01:49 : 05/08  36.3%
15:01:58 : 06/09  25.4%
15:03:19 : 07/10  17.2%
15:05:39 : 07/11  27.4%
15:06:18 : 07/12  38.7%
15:07:38 : 07/13  50.0%
15:08:10 : 08/14  39.5%
15:08:36 : 09/15  30.4%
15:08:51 : 10/16  22.7%
15:09:16 : 10/17  31.5%
15:09:25 : 10/18  40.7%
15:09:44 : 10/19  50.0%
15:09:51 : 10/20  58.8%
15:09:57 : Trial reset.
15:10:10 : 00/01  100.0%
15:10:12 : Trial reset.
15:10:14 : 01/01  50.0%
15:10:23 : 01/02  75.0%
15:10:30 : 01/03  87.5%
15:10:31 : Trial reset.
15:10:37 : 01/01  50.0%
15:10:46 : 02/02  25.0%
15:10:55 : 02/03  50.0%
15:11:15 : 02/04  68.8%
15:11:26 : 03/05  50.0%
15:11:35 : 04/06  34.4%
15:11:55 : 04/07  50.0%
15:12:03 : 04/08  63.7%
15:12:27 : 05/09  50.0%
15:12:34 : 06/10  37.7%
15:12:42 : 07/11  27.4%
15:13:07 : 08/12  19.4%
15:13:21 : 09/13  13.3%
15:13:41 : 09/14  21.2%
15:14:00 : 09/15  30.4%
15:14:09 : 09/16  40.2%
15:14:18 : 10/17  31.5%
15:14:27 : 11/18  24.0%
15:14:34 : 11/19  32.4%
15:14:39 : 11/20  41.2%
15:14:46 : Test finished.
 
 ---------- 
Total: 22/44 (56.0%)
 
 
post #42 of 107


Bowie

 

Well done for actually doing the testing.  Now compare this with your original quote - and you'll understand why ABX testing is a 'must'

Quote:
Originally Posted by bowei006 View Post

The beginning was hard to tell apart. but i finally got it

 

Warning: Spoiler! (Click to show)

i realized...this song actually had vocals after 2 mintues...silly me. so i set both to start at :30 seconds...and teh difference was pretty easy to tell. In A, the vocals don't flow. immidatley at :30 to :31 with the "across the floor" you can tell there is some staticy and the vocals jsut can't reach as easy.  i didn't read any of this before i tried. i did this with Shure SRH 440's and not my HFI 580's...so thus i wouldn't have heard the better bass at the beginning of B.

 



According to your results from the abx - you were completely guessing the difference - ie you couldn't tell.  Nothing to be worried about - the majority of us won't be able to either.

 

The best thing about this is that by knowing it, for your portable listening you can now use MP3 320 and save space on your DAP with complete confidence.  For idevices, I recommend aac256 as it seems to be technically superior to mp3, and I cannot tell the difference (on well transcoded tracks) between lossless, 320mp3 and 256aac.

 

YMMV.  To be sure - do your own transcodes of music you know well - and retest yourself at different bitrates.  It's pretty enlightening.

post #43 of 107
Quote:
Originally Posted by Brooko View Post


Bowie

 

Well done for actually doing the testing.  Now compare this with your original quote - and you'll understand why ABX testing is a 'must'



According to your results from the abx - you were completely guessing the difference - ie you couldn't tell.  Nothing to be worried about - the majority of us won't be able to either.

 

The best thing about this is that by knowing it, for your portable listening you can now use MP3 320 and save space on your DAP with complete confidence.  For idevices, I recommend aac256 as it seems to be technically superior to mp3, and I cannot tell the difference (on well transcoded tracks) between lossless, 320mp3 and 256aac.

 

YMMV.  To be sure - do your own transcodes of music you know well - and retest yourself at different bitrates.  It's pretty enlightening.



yes i understood this for a while now if you check my other posts on sound quality and stuff on other threads. L3000.gif I do know that a properly encoded 320kbps mp3 sounds very very similar if not exact to the ALAC. i 100% found this out about a week ago when i was curious and did some testing. before that i knew but wasn't sure exactly what the difference is. Yes, I can tell that the two files are different. But they both sound "good" without any glaring audio deficienes. so it's very hard to tell which one is CD ripped into compressed lossless format and which one was an lossy format.

 

yeah i use itunes to encode them. like this thing, i have yet to really find any difference from using itunes to rip and encode as it does the job and has error checking. I use to just download music..320kbps..haha now those 320kbpsthat were downloaded from regular CD ripped ALAC  converted to 320kbps myself..haha you can definatley tell a diff then...which is weird....as you then wonder how the heck the person that encoded did encode it? :/ it's easy to tell bad mastering apart from the audio codec usually. you should hear some 320kbps...it's hilarious. i would have believed if someone told me that used to be an 60kbps mp3

 

yes i do do my own encodes and i have done testing with stuff like this. I will probably be transfering the majority of my files to AAC 256kbps or 320kbps VBR MP3. But I will leave some files that you can tell apart form MP3(i have a couple) as ALAC and still retian the original files on my Macbook Pro..

so:

256AAC/320kbps MP3 on ipod touch-->amp-->lod-->headphones

 

computer--alac---optical out--optical cable--DAC--pre amp-- driving amp--headphones

 

 

and yeah the difference from 320kbps sounding bad is not the file itself. it's how the person encoded it. this is just a statement for any lurkers reading. bad mastering is also an issue and for vinyls..you get way way more issues. mainly with ripping process

 


Edited by bowei006 - 2/16/12 at 3:04pm
post #44 of 107
Code:
foo_abx 1.3.4 report
foobar2000 v1.1.18
2012/02/16 14:48:42
 
File A: C:\Users\Username\Desktop\A.wav
File B: C:\Users\Username\Desktop\B.wav
 
14:48:42 : Test started.
14:49:08 : 01/01  50.0%
14:50:01 : 02/02  25.0%
14:50:39 : 03/03  12.5%
14:51:36 : 03/04  31.3%
14:52:10 : 04/05  18.8%
14:53:09 : 05/06  10.9%
14:54:12 : 06/07  6.3%
14:55:16 : 07/08  3.5%
14:56:16 : 08/09  2.0%
14:57:33 : 09/10  1.1%
14:58:13 : 10/11  0.6%
14:59:14 : 11/12  0.3%
15:00:27 : 12/13  0.2%
15:01:34 : 13/14  0.1%
15:02:31 : 14/15  0.0%
15:02:58 : 15/16  0.0%
15:03:51 : 16/17  0.0%
15:04:53 : 17/18  0.0%
15:05:49 : 18/19  0.0%
15:06:11 : 19/20  0.0%
15:06:56 : 20/21  0.0%
15:08:18 : 21/22  0.0%
15:09:05 : 22/23  0.0%
15:09:40 : 23/24  0.0%
15:10:19 : 23/25  0.0%
15:10:43 : Test finished.

 ---------- 
Total: 23/25 (0.0%)

That was way too easy!













Actually, I don't even have Foobar installed. I didn't even download the test files. It was faster to fake the ABX logs than it would have been to take the test. I just wanted to demonstrate that posting ABX logs doesn't lend any extra credibility to claims of being able to accurately differentiate between codecs, bit rates, etc. Those who want to lie about it can still do so with ease. Those who choose to be honest will still be honest.
post #45 of 107

You're not going to be able to stop anyone that wants to be a poser.  Want an easier way - take one of the files, and manipulate it - then abx the 2 files.  You don't have to fake the logs that way.

 

We have to rely on honesty - otherwise what is the point in posting.

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › BLIND TEST: Lossless vs. MP3 320