Head-Fi.org › Forums › Equipment Forums › Sound Science › Lossless vs 128kbps mp3 vs 320kbps mp3 blind test
New Posts  All Forums:Forum Nav:

Lossless vs 128kbps mp3 vs 320kbps mp3 blind test - Page 3

post #31 of 118

Just bumping this because I think it's very interesting and useful.

 

I've done 128kbps vs 320kbps tests in the past before and save for some very minor things (more distortion on clear high notes with lower kbps files for example), I can't reliably tell a difference. As I have always felt, recording quality is what it's all about, not bitrate. Granted, I am a music listener and not really an audiophile.

I am sure certain listeners have their ways of finding out structural differences between bitrates though.

post #32 of 118
Ops, I take back my comments about music choice as I was playing the files direct from my macs download folder and had not realised that it was playing the rubbish I had downloaded last night after each of your tracks.
No wunder they all sounded like dubstep.biggrin.gifWarning: Spoiler! (Click to show)
ive just had another go with the 2nd one and I'm not sure i can tell any differance, this is the hardest one. I'll go with a as lossless and b and c are 128 then 320 but guesses
post #33 of 118

Very nice.

Warning: Spoiler! (Click to show)

 

Well, I listened to these for quite a while, and i'm not sure I can tell the two better files apart on my laptop, so I'm only going to make an attempt at the 128K files.  Group 1, file C somehow seems annoying...Group 2, file A seems gritty and B is not much better....I'm not sure I have an opinion on this material after listening for a while.  I like the music....i may be buying this one.  :)  Group C I had to convince myself I heard differences...all versions lacked air and depth needed to tell them apart....it may have been over produced from the start.

 

post #34 of 118
Warning: Spoiler! (Click to show)

1 - B>A>C

2 - C>A>B

3 - C>B>A
post #35 of 118

Why don't you just post the answers and type them in spoilers? We can check them straightaway, rather than waiting a week and forgetting about this thread

post #36 of 118
Thread Starter 
Quote:
Originally Posted by thepredestrian View Post

Why don't you just post the answers and type them in spoilers? We can check them straightaway, rather than waiting a week and forgetting about this thread

Because people cheat. While most wouldn't there probably would be a few people that would.

 

I'll leave the files and everything once I do post the answers so people can still ABX and the like, but I'd like to get some statistics beforehand. It'll probably be closer to Monday or Tuesday, as the views have really skyrocketed this weekend(there were just over 300 on Thursday I believe, now it's in the thousands). There have been over 200 downloads of the first group too...

post #37 of 118
Warning: Spoiler! (Click to show)

It is still possible to cheat for those who want to, by simply looking at the spectrum of the files (even the foobar2000 "Spectrogram" visualization is sufficient for this purpose, no audio editor is needed).

By the way, Group 2 does not seem to include a 128 kbps file (although it could have been intentional to see who "hears" a difference between identical files) ?


Edited by stv014 - 1/19/13 at 9:00am
post #38 of 118
Quote:
Originally Posted by chewy4 View Post

Because people cheat. While most wouldn't there probably would be a few people that would.

 

I'll leave the files and everything once I do post the answers so people can still ABX and the like, but I'd like to get some statistics beforehand. It'll probably be closer to Monday or Tuesday, as the views have really skyrocketed this weekend(there were just over 300 on Thursday I believe, now it's in the thousands). There have been over 200 downloads of the first group too...


it would be great if, alongside with the answers, you publish some statistics per group and file type.

 

Table 1: will tell us the percentage of users that got the answers correctly either by group and by filetype 

GROUP 128kbps 320 kps lossless tot participants
1 n k j n+k+j p1
2 o p y o+p+y p2
3 e d f e+d+f p3
tot n+o+e k+p+d j+y+f    
participants p128 p320 ploss    

 

 

with n,k,j etc being the number of people that got the answer right, and p1,p2, p3 the total of participants for the groups and p128,p320, ploss the number of retrieved answers per filetype (right and wrong)

 

 

I also assume that 320/LL will present the highest margin of error among users, so another interesting table could answer the question "can users reliably distinguish 128kbps mp3s from higher quality files?

 

Here we need Table 2, which can actually be derived from Table 1 but for the sake of handiness..

GROUP right/wrong 128kbps 320/lossless
1 Right n k+j
1 Wrong w1 q1
2 Right o p+y
2 Wrong w2 q2
3 Right e d+f
3 Wrong w3 q3

 

With a little bit of statistics we can do the math. I can do that, if you provide the table - I'll deliver. Only problem is that we probably have not many data points per group, especially group 2. Actially, Group1 and Group3 are not that diverse as music genres, so if we lack of data (we need at least 10-15 participants per group, not necessarily unique) we can aggregate that.

 

This way we can add some actual science to the result of the test and conclude (or not conclude) with statistical significance (or no statistical evidence) that people on head-fi are good/bad at detecting sound quality of music files.

post #39 of 118

This was pretty awesome. 

 

Warning: Spoiler! (Click to show)

Alright. Lossless to 128.

 
 
Group B
B-A-C
 
B and A were really close but I felt B had a little bit better separation so I went with B. 
 
Group C
 
I really liked this one
 
C-B-A 
Finding the Flac seemed easier in this test where B and A stood no chance. 
Though I did find the A was more enjoyable to listens to then B. In B you could hear more instruments just popping out and the bass notes were more prominent but in return felt muddy. I almost pick 128 for this but i knew it was just one of those moments when a song cant do it all. A was more enjoyable because there wasnt as much bass so there was less to muddy it up. Still not as clear as the highs in B but Sounds nicer. B was just trying to hard to be lossless. 

 

post #40 of 118
Quote:
Originally Posted by chewy4 View Post

Because people cheat. While most wouldn't there probably would be a few people that would.

 

I'll leave the files and everything once I do post the answers so people can still ABX and the like, but I'd like to get some statistics beforehand. It'll probably be closer to Monday or Tuesday, as the views have really skyrocketed this weekend(there were just over 300 on Thursday I believe, now it's in the thousands). There have been over 200 downloads of the first group too...

 

skyrocketed urghh? nice, it´s propably becouse i gave it on reddit :-) http://www.reddit.com/r/headphones/comments/16ti95/lossless_vs_128kbps_mp3_vs_320kbps_mp3_blind_test/
 

post #41 of 118

I'm not the most active Head-Fi member, but saw this on reddit, hehe. 

Group 1 (Click to show)

A: 320kbps

B: FLAC

C: 128kbps
 
C sounds crappy, A and B required me to switch headphones a couple of times (from Beyer BT770 to Alessandro MS-1 and back), but it's definitely noticeable.

 

 

Spoiler 2 (Click to show)

A: FLAC (guess)

B: 128kbps (guess)
C: 320kbps (guess)
 
Can't really tell. B and C sound too similar to be 128kbps and 320kbps, but if I had to pick I'd go for C. A seems to sound a little tad closer to source than B and C. The source definitely sounds like it already has some compression artifacts too, could be a mistake by OP but might as well be the recording itself and the fact that it's glitchy music and I'm quite high atm.

 

Group 3 (Click to show)

A: 128kbps

B: 320kbps
C: FLAC
 
I know those tunes really good since I'm a huge Shpongle fan. The plucks just sound broken in A and B and have plenty of artifacts.

 

post #42 of 118
Thread Starter 
Quote:
Originally Posted by Maker View Post

 

skyrocketed urghh? nice, it´s propably becouse i gave it on reddit :-) http://www.reddit.com/r/headphones/comments/16ti95/lossless_vs_128kbps_mp3_vs_320kbps_mp3_blind_test/
 

Well that would explain it then, thanks for sharing it.

 

I'm just going to post the key now, I figure there is enough to form some statistics. I'm too hungover to make accurate statistics right now so I'll do them Monday night, and I'll try to make them as in-depth as possible. Make sure to not edit your post after this point, as I will consider your answers invalid.

 

Key:
 

 

Warning: Spoiler! (Click to show)

 

1: Mighty River - Railroad Earth
B - lossless
A - 320
C - 128
 
2: Fahrenheit Fair Enough - Telefon Tel Aviv
A - Lossless
B - 320
C - 320
 
3: Falling Awake - Shpongle
A - 128
B - 320
C - lossless
 
 
stv014 mentioned that there didn't appear to be a 128kbps file in group 2. If this is the case, it wasn't intentional.I thought I was pretty careful with it... so can anyone elaborate on that? I'm no pro at looking for this kind of stuff.
 
EDIT: Just re-converted, I did fudge up on group 2. The one I had as 128kbps is indeed just another 320kbps file... Didn't intend for this, so sorry guys. I will post the real 128kbps if people want to ABX it.

 


Edited by chewy4 - 1/20/13 at 7:37pm
post #43 of 118
Ooh 2/3, I'll give the tracks another listen tonight, first one was really hard

now just 1/3 -.-
Edited by KamijoIsMyHero - 1/20/13 at 8:44pm
post #44 of 118

With the Mighty River track, I simply got the 320kbps and Lossless files mixed up but I was right with the 128kbps.

 

I only got the 320kbps right on Fahrenheit Fair Enough. I got the 128kbps somehow mixed up with the lossless... How I don't know.

I just opened that up in Audacity though, and its clipped on the Lossless version in that case... Look at the levels... They are clipped compared with the other two. Whether that had something to do with it.

 

Didn't get any right on Falling Awake :p I thought the 320kbps file was 128kbps... haha


Edited by musical-kage - 1/20/13 at 6:25pm
post #45 of 118

My answers were no more statistically meaningful than guesses. I don't see any rhyme or reason in them compared to the answer key. Based on these results, I think I can listen to mp3s and MOG now without second guessing the quality.           

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › Lossless vs 128kbps mp3 vs 320kbps mp3 blind test