IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
MP3 listening test done at MaximumPC, What ...
sizetwo
post Apr 21 2007, 00:47
Post #1





Group: Members
Posts: 143
Joined: 22-April 03
From: Kristiansand
Member No.: 6114



Just came by this test done at MaximumPC.com entitled "Do Higher MP3 Bit Rates Pay Off?". I have to say I am quite ... intruiged ?

Their choices of bit-rates were 160, 320Kb/s Mp3's (encoded in iTunes, VBR) and WAV. blink.gif

Anyone read it and feel its quite a strange choice ? Also the issues with it, such as not doing true ABX'ing?

I guess what I found to be the most strange is that the people doing the tests were all able to identify some of the 320Kb/s track, which I would assume
should be transparent for most people. [ducks]
Sounds like they didnt normalize the volume between the tracks or otherwise screwed up the test somehow.

edit: typos.

This post has been edited by sizetwo: Apr 21 2007, 02:03
Go to the top of the page
+Quote Post
saratoga
post Apr 21 2007, 01:51
Post #2





Group: Members
Posts: 5168
Joined: 2-September 02
Member No.: 3264



Probably just crappy methodology.

Though ABXing 320k iTunes MP3 isn't exactly difficult if you pick the right tracks.
Go to the top of the page
+Quote Post
Squeller
post Apr 23 2007, 22:22
Post #3





Group: Members
Posts: 2351
Joined: 28-August 02
Member No.: 3218



QUOTE (sizetwo @ Apr 20 2007, 15:47) *
I guess what I found to be the most strange is that the people doing the tests were all able to identify some of the 320Kb/s track, which I would assume
should be transparent for most people.
People come up regularly. Maybe it's just we don't know how we have to listen to a 320 kbps cbr vs. original, in order to distinguish it :|
Go to the top of the page
+Quote Post
LANjackal
post Apr 23 2007, 22:50
Post #4





Group: Members
Posts: 731
Joined: 26-October 05
From: Various networks
Member No.: 25371



It's not an official ABX test. Unfortunately everyone else out there will take it as the gospel truth... *sigh*


--------------------
EAC>1)fb2k>LAME3.99 -V 0 --vbr-new>WMP12 2)MAC-Extra High
Go to the top of the page
+Quote Post
chelgrian
post Apr 23 2007, 23:00
Post #5





Group: Members
Posts: 316
Joined: 27-April 03
Member No.: 6228



QUOTE (LANjackal @ Apr 23 2007, 22:50) *
It's not an official ABX test. Unfortunately everyone else out there will take it as the gospel truth... *sigh*


Plus it was linked to by digg *doublesigh*. I wonder if they bothered to turn the X-Fi's Crystaliser pile of doodoo off.
Go to the top of the page
+Quote Post
LANjackal
post Apr 24 2007, 03:10
Post #6





Group: Members
Posts: 731
Joined: 26-October 05
From: Various networks
Member No.: 25371



Eh, it's what happens (quite often) when people who consider themselves computer experts make the sad mistake of thinking that also makes them audio experts. I used to be one of them, lol... until (thankfully) I came here. Unfortunately most of them don't and so we have misinformation and awful methodology like this plastered all over the internet.


--------------------
EAC>1)fb2k>LAME3.99 -V 0 --vbr-new>WMP12 2)MAC-Extra High
Go to the top of the page
+Quote Post
sizetwo
post Apr 24 2007, 06:15
Post #7





Group: Members
Posts: 143
Joined: 22-April 03
From: Kristiansand
Member No.: 6114



Well yeah I share you guy's sentiment on this. Its quite obvious that mixing 'computer knowledge' with specific audio compression knowledge is not necessarily compatible. Once again another test that people will believe, go tell their friends about and end up being another fallacy being considered 'the truth' since its done by 'professionals'.
Go to the top of the page
+Quote Post
halb27
post Apr 24 2007, 10:35
Post #8





Group: Members
Posts: 2446
Joined: 9-October 05
From: Dormagen, Germany
Member No.: 25015



Hmm, the author's kind of speech is a bit strange, but as far as I can see it was a blind listening test with no evidence that something was done in a way to bring up placebo thinking and feeling.

Is the result really astonishing?
To me the only remarkable thing is that there is a certain bias that with Pluralist's and Hipster's tracks the lossless version could be differentiated from the compressed version in a way that can be iterpreted more than by chance, and this was possible especially for those persons who knew these tracks very well.
Any other result can be interpreted as been succesful/not successfull by mere chance.

This just tells me that with very intensive listening (not to be mixed up with common listening situations) and especially with well-known musical stuff there are mp3 encodings that can be differentiated from the lossy version, even when using high bitrate.

I don't consider this a strange result.

This post has been edited by halb27: Apr 24 2007, 10:35


--------------------
lame3100m -V1 --insane-factor 0.75
Go to the top of the page
+Quote Post
greynol
post Apr 24 2007, 18:10
Post #9





Group: Super Moderator
Posts: 10341
Joined: 1-April 04
From: San Francisco
Member No.: 13167



As far as I can tell they were only asked to match up the 12 samples just one time. If this is in fact true and after looking at the scores, you cannot dismiss the possibility that they were guessing.

This post has been edited by greynol: Apr 24 2007, 18:22


--------------------
Your eyes cannot hear.
Go to the top of the page
+Quote Post
halb27
post Apr 24 2007, 21:08
Post #10





Group: Members
Posts: 2446
Joined: 9-October 05
From: Dormagen, Germany
Member No.: 25015



Yes, I see it like this too, and looking at it globally the results could have been achieved by mere guessing. Looking at the author's details there seems to have been quite a lot of guessing in fact.

Anyway though there probably was only one trial for each person for each track it looks like the testers had done a lot of effort for getting at their results. So I wouldn't call the result totally worthless (but also not extremely surprising) concerning the two samples which were guessed correctly as encodings especially by those people who knew the tracks very well.

I agree however that the result's significance is very restricted due to the very few testers, samples and guesses.

This post has been edited by halb27: Apr 24 2007, 21:09


--------------------
lame3100m -V1 --insane-factor 0.75
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 28th December 2014 - 01:26