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 5
7 8 9 10 11 12
  1. MrLazyAnt
    Pardon my ignorance, but what's ABX?
  2. Brooko Contributor
    It's a double blind comparison test - basically comparing two different file containers (same song), removing the effect of placebo (blind) so you can see what you actually hear.
    Quick link (http://www.head-fi.org/t/655879/setting-up-an-abx-test-simple-guide-to-ripping-tagging-transcoding) to a simple guide I wrote that covers ripping, tagging, transcoding AND setting up an abx in Foobar.
    sonitus mirus likes this.
  3. MrLazyAnt
    Isn't that what Chewy was doing? I mean, I get that I may have "cheated" by only doing one song, but isn't the whole point of this thread to do a double blind audio test?
  4. sonitus mirus
    Please don't take my comment as a denigration.  That was not my intent.
    What I was attempting to show was that this test is about a collection of results from a group, and not about any individual achievements.
    It is like having 50 people guess a number between 1 and 10.  If 5 out of 50 people guess correctly, it doesn't mean those 5 people are extraordinary unless those 5 people can repeat the test with similar results multiple times. 
  5. Brooko Contributor
    But to compare the tracks properly - you need to use something like Foobar 2000s abx tool.  Setting it up is in the 5th post of that link I sent you.
    How did you perform your double blind?  How many iterations?  Or did you just close your eyes, do it once, and called it good?
  6. sonitus mirus
    Sure, but if I guess heads on a coin flip and get it right, should I be excited about my coin-flip-guessing ability?  This test is about many people guessing the result of a coin flip, and seeing what the overall results might be.  
  7. MrLazyAnt
    I listened to the three samples 6 times on shuffle-repeat wrote down thusly: 1 reference 2 reference 3 reference 4 lossless 5 320 6 128 7 128 etc..... when I was done I checked the playing history using the foobar console, and wrote down next to each selection A B or C, then verified with chewy's answers, and the last 3 sets (7-18) were consistently correct.
    EDIT: But, to be fair, I have the original track.
  8. Brooko Contributor
    OK - again - suggest you go to my post, set up an actual blind abx, compare just 2 of the files - run 15 iterations, and post the logs.  Then do the next two files.  Etc.  Takes some time but worth it in the end.
  9. MrLazyAnt
    Maybe some other time...... Right now I'm just too damn lazy to download and install the necessary software. Not to mention my PC is a 7 year old hunk-o'-junk that I bought on a budget, it can't even run Cubase through ASIO smoothly any more and I don't know how many more programs/plug-ins I can install before it crashes once and for all. It's already dying the poor thing. But maybe one day
    I did succeed in the "Mighty River" one, and didn't bother with the "Telefon Tel-Aviv" one having read that 2 of them were 320kps, and being a lazy **** who just wants to listen to his music.
  10. HE1RO
    I honestly dont think its the bit rate that makes songs sound crappy, but how an what it was compressed with. I rip CD's all the time and the program i use has tons of options to choose from. I could run a lossless file through it and not alter its bit rate, but change a few settings and you would think it was an AM radio transmission lol
    Imo, its the overall file size in comparison to its format and compression and what options were chosen upon digitizing it that makes a difference. If its a 320kbps MP3 song thats 3minutes~ long and its 5.0mb+ in size, its probably going to sound just as good as if it was straight from the CD itself.
  11. yesman94
    thankyou chewy this is a great way to test and learn to differ quality songs. I can tell a different between the lossless or not on the accoustic one, the second song, since i dont like techno or whatever is that, i cannot enjoy and tell a different. It is easy to tell that lossless or lossy but its hard to tell between 128kbps and 320kpbs.
    here's my answer, from greatest quality to most poor :
    1. B>A>C
    2. C>B>A
    3. C>A>B
    A little missed from the answer what Iam happy that i can tell between lossless or not.
    Thanks again for chewy :D  :D
    Sorry for my poor english.
  12. chewy4
    Well a 3 minute 320kbps file is always going to be the same size(7.2MB) regardless of settings. Bitrate is just a ratio of time and file size.
    But I did indeed use different than default settings for these files. There is a quality slider for LAME compression that isn't by default set to the highest quality. Not sure how big of a difference it makes, but it's simple enough to change it.
  13. Sound Quest
    Sorry for digging this thread back up again. I see the OP said that he/she used the LAME encoder.
    Out of interest, which version of the LAME encoder did you use? And what speed did you have the encoder set to?
    Also, did you use joint stereo or full stereo mode?
  14. chewy4
    I used the most recent release of LAME. 3.99 I believe, as it hasn't been updated since October 2011. I can double check to make sure when I get home but I didn't install the software until about a month or so before the test so I'm 99% sure here.
    It was set to slow speed, high quality. Full stereo. Basically I didn't change anything from default other than the quality slider.
  15. Sound Quest
    Would it not be better to do an ABX comparason using Joint Stereo mode instead?
    The "Joint Stereo" mode in LAME is mathematically lossless at 128kbps and above. It will allow the encoder to save bits where both channels are 100% identical. It can then make use of those extra bits in the bit reservoir.
    In other words, you'll get better sound quality, making your ABX test even more challenging.
1 2 3 4 5
7 8 9 10 11 12

Share This Page