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.
1 2 3 4
6 7 8 9 10 11 12
  1. elwappo99
    Great thread! 
    I REALLY like the work you put into this. Keeping them the same file size, same sample, and volume equalized must have taken some work. Well done! 
    I only went through the first one. Based on my quick listening, I did get the answers correct, but I always love listening to these things, because they show how subtle of a difference all the Flac vs mp3 is, especially when you consider the 320 bit rate. I saw some people's comments about listening to brightness or presentation to pick out the differences. I believe somewhere a study was done and people with ibuds liked 128 the best. I think this leaves out some details in a song, and makes it sound more forward. 
    Anyway, the way I can easily pick out the differences are when I hear any kind of moraca or a drum brush (like at the 0:10 mark).  Also something to listen for is the subtle decay of guitars (like the decay in the 0:04 - 0:06). The subtle decay from a brush or the individual beads hitting are lost in 128, and 128 can easily be spotted doing this. 320 and Flac is a little more difficult. 
  2. catspaw
    Group 1 i noticed a diffrence from 128 to the other 2 but that was it. Had no idea if was better or not.
    Group 2 i noticed there was more clarity on the first song on the high notes, but didnt know if thats supposed to be good or bad ;D the other two did seem same to me.
    Group 3 Got it right very fast (was actualy the only group that was clear, others seemed very similar in general).
  3. Sleinzel
    Didn't do this Test, but i got some tracks where I can tell the difference over 80% between flac and 128 and around 65% between 320 and flac.
    - Pink Floyd - The Dark Side of The Moon - Time
    - Netsky - 2 - Love has gone
    - Hans Zimmer - The Dark Knight Rises - Imagine The Fire (actually a 24/196, because the cd has clipping :p)
  4. Jeremy1234
    I was quite surprise that my colleague passed me 2 songs which are so detail and nice but they are only 128Kbps..............................[​IMG]
    And I thought 320Kbps should be better.....................................[​IMG]
  5. jake120

    Yes I agree with this.
    I tried an 128kbps vs 320kbps vs flac song.
    at some times i could agree there were differences
    the 128kbps was not as loud
    the 320kbps was louder and instruments were clear
    the flac was as loud as the 320kbps but some instruments felt like they followed through until the end more.
    and then some songs the differences i should have noticed did not register.
    i wonder if its not just equipment but its taste as well that matters...
  6. KamijoIsMyHero
    this thread is interesting since I still hear a difference with 128 vs 320 from illegal DLs, but if the music is being passed from a competent legal source (or one that actually has a ripped copy)  then I don't hear a difference at all
  7. chewy4
    Makes sense, you don't know what that file has been through.
    It could have been burnt to a CD and re-ripped as mp3 several times, converted between lossy formats, anything really.
  8. soundstige
    All tests were done just now, with about 5 minutes spent on each zip package. They were done with a HiFiMan HE-400 headphone through a PCM2704 DAC and CMOY amp, and played through a regular playlist in Foobar2000. Obviously, they were done before reading the results/statistics. Also before reading the results, I think my Etymotic earphones might have had more chance at dissecting every possible artifact, but the HE-400 are certainly more "real-world" listening devices. 
    Group 1:
    Track 1_C is obviously the 128 sample. There are at least a dozen audible examples of pre-echo or warbling artifacts. I would say I'm 95% confident on this. I believe I hear a very small amount of artifacts in the cymbal crash around 21s that are similarly present in sample C, in sample B. If I had to guess, B would be 256, C is 128, and A is Lossless.
    So, 1_A Lossless, 1_B 256, 1_C 128!
    Group 2:

    I'm very unfamiliar with this type of music. If I must guess, and it is a guess, I'd say 1_B is the 128 sample, only because I THINK some complex parts sounded softer and less pronounced than others. As for the other samples, I can't give any meaningful input, even without knowing the results. I just, basically, don't know what to listen for. The music is nice though!
    So, 2_A ???, 2_B 128?, 2_C ???
    Group 3:

    The high E string plucking on 3_C around 22s sounded like it got a little mangled towards the end of its resonation. A slightly warble. Other than that, I'm not confident of any difference I heard between the tracks. If a gun were to my head, I'd say C is 128, A is 256, and B is Lossless. Not much real confidence about the guess or track as a whole.
    So, 3_A 256?, 3_B Lossless?, 3_C 128

    Cheers and thanks for the test!
    Edit: after reading the results, my impressions on my impressions:
    The only one I was truly confident about, Group 1, I was right! (only on the 128, but that's all I'd consider to count lol). The rest, all guessing. Though I was kind of right on 2! looool fun test :)
  9. ozarkcdn
    yo chewy... think you could rig a test to record outputs of a sample run through different DACs? I think results would be very interesting...
  10. stv014
    It has been done already [​IMG] Check this, this, or this thread.
  11. JUKILOP02
    A is the best, B sounds like 128 and C 320 Group 1
  12. OmarSy
    Great thread, I'll participate when I get home!
  13. Prowler

    Even though the test was concluded (results were posted in a spoiler, really I didn't peek, didn't read any other posts, didn't even notice just how old the first post was, really!), here are my results. I made a notepad++ document and I'll paste it into a spoiler.
    Before doing this test I was previously doing my own LAME V2, V0, lossless ABX testing, so I was already a bit familiar with mp3 artifacts (all thanks to recently buying a headphone amp for my Titanium HD, which I should have done when I built the PC, but hey, I had no idea it'd help that much).
    I was using: Foobar2000 -> Sound Blaster Titanium HD -> FiiiO E09k -> Sennheiser PC 350 (modded).
    Group 1
    A - 320 kbps mp3
    B - lossless
    C - 128 kbps mp3

    Group 2
    basically random guessing, couldn't hear any differences
    A - 128 kbps mp3
    B - lossless
    C - 320 kbps mp3

    Group 3
    A - 128 kbps mp3
    B - 320 kbps mp3
    C - lossless
    above is what I write in notepad++ after I concluded testing, below are comments added at the time of this post after seeing the spoiler.
    I spent about an hour on this, way more than I thought till I looked at the clock.
    My testing methodology was to first identify the 128 kbps, relatively easy for groups 1 and 3 but I pretty quickly skipped over group 2 as I struggled with it.
    I had an easier time identifying group 1 than group 3. Group 3 took a bit more time but eventually I had an ABX test < 10%, enough for my conclusion in this test.
    Finally I re-visited Group 2 but somehow my mind was so fixated on the fact there must have been a 128kbps encoding present then I committed to the above. After I saw the results for group 2 I just laughed.
    Yeah, the group 2 results are actually what I expected (320 kbps vs lossless),so  I feel fine using LAME V0, especially for a portable player. The differences between 320 kbps and lossless are so minute that without the foobar ABX tool it'd be virtually impossible to identify.
    I think I'm dome with lossy vs lossless testing... unless someone sets up another one of these tests and I find out about it before it ends :)
    Thanks for the creating the test chewy4.
  14. MrLazyAnt
    Hah! Got it right! I only listened to the Shpongle one, but I'm well stoked that I got it right! I have a pretty decent stereo, albeit very dated (speakers form the 70's and an amp from the late 80's), but my soundcard is an onboard laptop cheapie. So yay me! =P
  15. sonitus mirus
    You really need to ABX any 2 of the samples at least 15 times before getting excited. If you just are just trying to guess which file is which, you might be right, but can you repeat that or was it just a guess?

    Someone might correctly choose how many fingers I am holding up behind my back, but that doesn't mean they are psychic.

    You got it right on one song. The next step is to prove to yourself that you can pick the right files every time.
1 2 3 4
6 7 8 9 10 11 12

Share This Page