1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

Lossless vs 128kbps mp3 vs 320kbps mp3 blind test

Discussion in 'Sound Science' started by chewy4, Jan 15, 2013.
First
 
Back
1 2 3
5 6 7 8 9 10 11 12
Next
 
Last
  1. chewy4
    Well I did mess up on group 2, I swear it wasn't intentional(as tempting as doing something like that was).
     
    I edited the key with the change.
     
     
    What I did find interesting is that multiple people said that they guessed on which was 320 and lossless for this group, but said that they weren't guessing on the 128kbps. Which didn't actually exist. 
     
  2. Strangelove424
    Well, disregarding the error in group 2's electronic stuff, I thought the worst sounding files in Group 1 + 3 were C. Group 1's C was indeed 128kbps, but as it turns out Group C's 3 was lossless, and I believe that mistaking 128 for lossless by itself invalidates all of my observations. Those were acoustic samples too, which should be easiest to tell apart.      
     
  3. Luchico638
    Group A
    Oddly enough, I don't notice a difference betwen the 3.
    I think C may have been the lossless one.
     
  4. liuk
    Turns out my guesses where indeed no more than random guesses, and my brain did all the "justification" for me (and many of ours did too, I see).
     
    Out of curiosity yesterday I did a second blind test with some of my music with my girlfriend.
     
    Only piece I could get right to be compressed 128 vs Lossless was the 4th movement of Dvorjak's 9th symphony. And I had to take the 128 compression down to "default" quality, because with "high" quality I really could not spot any difference.
     
    Lesson learned: compress everything to 256kbps and you'll never have to complain about anything.
     
  5. musical-kage
    Mighty River was a lot more complex in sound. 
    I'm interested in maybe creating one myself with a classical track and maybe one other at different parts, with complex and more analogue. 
     
  6. chewy4
    Just remember that you do need to volume match for 128kbps. It wasn't neccessary for 320kbps for the files and encoder that I chose, but 128kbps always came out a little quieter(around a half dB, could vary even more for songs with a high dynamic range).
     
    How I did it was using dbpoweramp's replaygain apply. I converted the files, compared their replaygain levels in foobar, and then converted again(from the source) using the adjustments needed in the replaygain apply DSP. I'm not sure if there is a way to get that DSP to make the adjustments automatically, if there is I didn't figure it out.
     
  7. ozarkcdn
    Very nice - and love you did this for all the 'I don't listen to anything less than 320kps' crowd.  Maybe in a secret bunker in a deep underground layer... [​IMG]
     
    You did some good work here - what would also be fun is comparing AAC vs MP3 etc, which they do over here
     
    That's where I got the idea to compress my mobile library to 96k AAC files. IIRC, the 96k AAC files sounded almost a little better than 128k mp3... and takes about 33% less room-ish.
     
     
    Anyhow - I'd like to see some others report their report from ABX plugin on foobar, here's mine!
     
    Nice selection of tracks!
     
    My best guesses were on:
     
    #1 - B
    #2 - A
    #3 - C
     
    Hahahahahah.... guess I messed that up!
     
     
     
    Test data:
     
     
    Track #1:
    B vs souce was the most reliable
    foo_abx 1.3.4 report
    foobar2000 v1.2.1
    2013/01/21 15:23:53
    File A: C:\Users\Geoff\Downloads\Group 1\1_A.wav
    File B: C:\Users\Geoff\Downloads\Group 1\1_Source.wav
    15:23:53 : Test started.
    15:25:09 : 00/01  100.0%
    15:25:56 : 00/02  100.0%
    15:26:56 : 00/03  100.0%
    15:27:02 : Test finished.
     ----------
    Total: 0/3 (100.0%)
    --------------------------------
    foo_abx 1.3.4 report
    foobar2000 v1.2.1
    2013/01/21 15:27:45
    File A: C:\Users\Geoff\Downloads\Group 1\1_B.wav
    File B: C:\Users\Geoff\Downloads\Group 1\1_Source.wav
    15:27:45 : Test started.
    15:29:39 : 01/01  50.0%
    15:30:43 : 02/02  25.0%
    15:31:18 : 03/03  12.5%
    15:32:53 : 03/04  31.3%
    15:33:06 : 03/05  50.0%
    15:33:42 : 04/06  34.4%
    15:34:36 : 05/07  22.7%
    15:35:30 : 06/08  14.5%
    15:35:35 : Test finished.
     ----------
    Total: 6/8 (14.5%)
    ----
    foo_abx 1.3.4 report
    foobar2000 v1.2.1
    2013/01/21 15:35:51
    File A: C:\Users\Geoff\Downloads\Group 1\1_C.wav
    File B: C:\Users\Geoff\Downloads\Group 1\1_Source.wav
    15:35:51 : Test started.
    15:36:18 : 00/01  100.0%
    15:36:44 : 00/02  100.0%
    15:37:14 : 00/03  100.0%
    15:37:38 : 00/04  100.0%
    15:37:41 : Test finished.
     ----------
    Total: 0/4 (100.0%)
     
    Group 2
    ---
    foo_abx 1.3.4 report
    foobar2000 v1.2.1
    2013/01/21 15:41:18
    File A: C:\Users\Geoff\Downloads\Group 2\2_A.wav
    File B: C:\Users\Geoff\Downloads\Group 2\2_Source.wav
    15:41:18 : Test started.
    15:43:05 : 01/01  50.0%
    15:43:16 : 01/02  75.0%
    15:43:43 : 02/03  50.0%
    15:44:27 : 03/04  31.3%
    15:45:11 : 04/05  18.8%
    15:45:18 : Test finished.
     ----------
    Total: 4/5 (18.8%)
     
    --
    foo_abx 1.3.4 report
    foobar2000 v1.2.1
    2013/01/21 15:45:40
    File A: C:\Users\Geoff\Downloads\Group 2\2_B.wav
    File B: C:\Users\Geoff\Downloads\Group 2\2_Source.wav
    15:45:40 : Test started.
    15:46:50 : 01/01  50.0%
    15:47:23 : 01/02  75.0%
    15:47:49 : 02/03  50.0%
    15:48:22 : 03/04  31.3%
    15:48:45 : 03/05  50.0%
    15:49:12 : 04/06  34.4%
    15:49:58 : 04/07  50.0%
    15:50:02 : Test finished.
     ----------
    Total: 4/7 (50.0%)
    ---
    foo_abx 1.3.4 report
    foobar2000 v1.2.1
    2013/01/21 15:50:20
    File A: C:\Users\Geoff\Downloads\Group 2\2_C.wav
    File B: C:\Users\Geoff\Downloads\Group 2\2_Source.wav
    15:50:20 : Test started.
    15:51:28 : 00/01  100.0%
    15:52:09 : 00/02  100.0%
    15:52:28 : 00/03  100.0%
    15:52:30 : Test finished.
     ----------
    Total: 0/3 (100.0%)
     
    ------
     
    Group 3
    ---
     
    foo_abx 1.3.4 report
    foobar2000 v1.2.1
    2013/01/21 15:54:45
    File A: C:\Users\Geoff\Downloads\Group 3\3_A.wav
    File B: C:\Users\Geoff\Downloads\Group 3\3_Source.wav
    15:54:45 : Test started.
    15:55:04 : 01/01  50.0%
    15:55:25 : 02/02  25.0%
    15:56:12 : 02/03  50.0%
    15:56:42 : 03/04  31.3%
    15:57:00 : 04/05  18.8%
    15:57:05 : Test finished.
     ----------
    Total: 4/5 (18.8%)
     
    --
    foo_abx 1.3.4 report
    foobar2000 v1.2.1
    2013/01/21 15:57:21
    File A: C:\Users\Geoff\Downloads\Group 3\3_B.wav
    File B: C:\Users\Geoff\Downloads\Group 3\3_Source.wav
    15:57:21 : Test started.
    15:57:58 : 01/01  50.0%
    15:58:09 : 02/02  25.0%
    15:58:31 : 02/03  50.0%
    15:58:41 : 02/04  68.8%
    15:59:10 : 02/05  81.3%
    15:59:24 : 02/06  89.1%
    15:59:26 : Test finished.
     ----------
    Total: 2/6 (89.1%)
     
    --
    foo_abx 1.3.4 report
    foobar2000 v1.2.1
    2013/01/21 15:59:43
    File A: C:\Users\Geoff\Downloads\Group 3\3_C.wav
    File B: C:\Users\Geoff\Downloads\Group 3\3_Source.wav
    15:59:43 : Test started.
    16:00:04 : 00/01  100.0%
    16:00:26 : 01/02  75.0%
    16:00:38 : 02/03  50.0%
    16:01:09 : 03/04  31.3%
    16:01:40 : 04/05  18.8%
    16:01:45 : Test finished.
     ----------
    Total: 4/5 (18.8%)
     
     
  8. chewy4
    I posted some statistics in the original post. I stopped at the point where I posted the key, so sorry if you just missed it.

    Thank you to everyone who participated!
     
    Also,
     
     
    I'll post a real 128kbps file for group 2 later tonight for those who want to ABX. 
     
  9. deadlylover
    I recognised Shpongle, so that's the one I tested for.

    I could only ABX the 128kbs sample, maybe I'll try the 320kbps again later tonight when the house isn't as noisy, but I think I'll fail again as I believe the 320kbps sample is transparent for this track. :p

    foo_abx 1.3.4 report
    foobar2000 v1.1.18
    2013/01/22 14:39:47

    File A: C:\Users\Alice\Desktop\Group 3\3_A.wav
    File B: C:\Users\Alice\Desktop\Group 3\3_Source.wav

    14:39:47 : Test started.
    14:44:13 : 00/01 100.0%
    14:45:02 : 01/02 75.0%
    14:45:15 : 02/03 50.0%
    14:45:46 : 03/04 31.3%
    14:46:07 : 04/05 18.8%
    14:46:49 : 05/06 10.9%
    14:47:58 : 06/07 6.3%
    14:48:04 : 07/08 3.5%
    14:48:20 : 07/09 9.0%
    14:48:38 : 08/10 5.5%
    14:48:42 : 09/11 3.3%
    14:48:55 : 10/12 1.9%
    14:51:43 : 11/13 1.1%
    14:51:55 : 11/14 2.9%
    14:52:01 : 12/15 1.8%
    14:52:44 : 13/16 1.1%
    14:52:56 : 14/17 0.6%
    14:53:22 : 14/18 1.5%
    14:54:04 : 15/19 1.0%
    14:54:13 : 16/20 0.6%
    14:54:46 : 17/21 0.4%
    14:55:08 : 18/22 0.2%
    14:55:32 : 19/23 0.1%
    14:56:01 : 20/24 0.1%
    14:56:23 : 21/25 0.0%
    14:56:30 : Test finished.

    ----------
    Total: 21/25 (0.0%)
     
  10. cer
    chewy4, what mp3 encoder was used?
     
  11. chewy4
    Quote:
    LAME via dbpoweramp.
     
  12. sonitus mirus
    I listen almost exclusively to 320 kbps mp3 files using MOG or from ripping a CD with LAME.  In every group in this test, I selected the 320 kbps file as the lossless version.  I think that is what I am used to hearing, so it sounded best to my ears.  But with such a small test sample, I was probably only guessing and the results were most likely a chance happening and nothing more.
     
  13. musical-kage
    ...  Whoops
     
  14. pathway
    Thanks for doing this - really interesting to blind test myself.
     
    (Kit used - Laptop, Topping D1 Mark II, AKG550)
     
    I got Group 1 100% correct and I was pretty confident about it - there was no guesswork but it did take quite a number of back to back listens. I back to backed 2 tracks and ranked them - then did the same again for another pair and the order fell out quite easily. That said, I would never have been able to tell if it was lossless or heavily compressed (128) without having the chance to listen multiple times back to back.
     
    More interestingly I got group 3 flipped (as it seems a few others did) - ranking the lossless file as 128 and vice versa. I found that group far more difficult to differentiate in general and guesswork crept into it. the 128 track 'felt' more open to me and that is why I ranked it lossless! - perhaps as there was just so much more 'going on' in the track it was harder to pick out details but it is great example of putting too much onus on the numbers
     
    Oh - I didn't try group 2 as read something during the thread about it by accident and didn't feel I would be totally 'blind' to the test.
     
    REALLY interesting experiment - many thanks for providing.
     
    O.
     
  15. MpX5
    Got new headphones, didn't warm up them properly, so here's my guess:
     
    Group 1: sounds pretty same: 
    B sounds sharper for me so let it be flac
    A is 128 cuz I didn't like it first time :)
    C - 320
     
    Group 2:
    from flac to 128

    B
    A
    difference - C sounds wider, like there is more sounds
     
    Group 3:
    no drums here(
    A is 128 because sounds like there is lack of bass
    B & C has not much difference for me
    C is flac
     
     
First
 
Back
1 2 3
5 6 7 8 9 10 11 12
Next
 
Last

Share This Page