Head-Fi.org › Forums › Equipment Forums › Sound Science › Blind Test - Take Now!
New Posts  All Forums:Forum Nav:

Blind Test - Take Now!

post #1 of 88
Thread Starter 

A blind test for you guys.  This is a lossless vs 128kbps mp3 test.  I encoded both the original wav and mp3 file into lossless flac format at random flac encoding level to ensure their identity is not revealed.  So ignore the slightly different file size.

 

If you think this test is rigged, biased, or faulty, then just do not post.  This test is for my curiosity only.

 

Attached are two files A and B.  The track is Tamacun by Rodrigo Y Gabriela.  Just leave a post saying which you think is the lossless and which is the mp3.  If you can not tell, then say that.  Thanks to all who try.

 

http://www.fileden.com/files/2008/2/16/1763003/1_A.flac

 

http://www.fileden.com/files/2008/2/16/1763003/1_B.flac

post #2 of 88

I'll bite, I can't tell the difference. Although I would say it's difficult to tell with very percussive music, for me at least bit rate makes a difference for synth, strings or cymbals where the aliasing is more noticable probably because there's fewer bits available to capture the wide range in frequency.

post #3 of 88

i say b (edit, forgot to say what b was ha) is lossless, but the song choice makes it interesting i think. pm me crazycarl if i got it right!!!  i wont tell i swear

 

M


Edited by mmayer167 - 10/19/10 at 8:00am
post #4 of 88

They really sound similar, but if I would have to guess, I would say that B is mp3. I listened to the first 2 seconds over and over, and focused on the overtones that were produced by Rodrigo's lower notes. A's overtones sound more present. 

 

Listening to the whole clip made it near impossible to tell which was which, but if I set a list of 50% A and 50% B and shuffled randomly between the list, I found if I only listened to the first two seconds, I could accurately tell B and A apart. 

post #5 of 88

 

Quote:
Originally Posted by Rooford View Post

I'll bite, I can't tell the difference. Although I would say it's difficult to tell with very percussive music, for me at least bit rate makes a difference for synth, strings or cymbals where the aliasing is more noticable probably because there's fewer bits available to capture the wide range in frequency.


 

Ha!  That's why I like listening tests that don't have cymbals.  Too many times listening tests for a song get whittled down to just the milliseconds around a cymbal hit and using just that to judge which one is lossy and which is lossless.  I'd rather have listening tests that force people to listen to the entire song and the overall sound rather than zeroing in on specific pre-echo artifacts in specific milliseconds around cymbal hits.

 

We did a recent 128 mp3 vs. lossless test with a Mozart piece.  No cymbals.  That was fun.  A difference can be heard, but it's not in pre-echo type artifacts.

 

For the Rodrigo y Gabriela I'd suggest listening for the space that is inside and just around the guitar(s).  The picups are inside the guitar.  They're acoustic pickups.  You can catch the space that is the body of the guitar with a good headphone and good gear.  Differences when picking the strings closer to the neck vs. closer to the body of the guitar.  Differences especially when drumming around on the body of the guitar.  With a good headphone and good gear you can hear the drumming swim around your head as they drum around on different parts of the body of the guitar (look at the YouTube videos to see what they're doing, and then listen for that).  You can hear clear differences between just fingers vs. an open palm hitting the body of the guitar.  Things like that.  The MP3 version should be missing some of those clues or have those spacial clues more blurred.

 

I like to use the Rodrigo y Gabriela CD as listening samples at meets.  I've heard Tamacun and Diablo Rojo on a variety of good headphones.  Some very good headphones don't get those spacial cues as well as other headphones.  The best I've heard for those spacial cues have been the Grado GS1000 and PS1000.  The big salad bowl Grados are amazing in that respect.  Soundly upstaging even headphones like the LCD-2 (but the LCD-2 gets the tone and timbre right, so very very right, so very real).  My poor little RS325is is just a pale shadow of what the GS1000 and PS1000 does for spacial cues and the sense of space with the Rodrigo y Gabriela.

 

So I think particular headphones are going to make this particular listening test easier or harder depending on how well the headphones do at getting the sense of space in this song.  I'd suggest headphones like the big Grados or the HD800 for this test if you've got them.  HD600/HD650, D2000, and similar will make the test more difficult.  I don't know about the K701/702, I haven't tried it with this music.

 

I'm going to use my SR325is for this one cause it's what I've got.  If I had access to a HD800 or GS1000 I'd use that.

 

I already gave the samples a quick listen.  I think I know which is which and it seems I'm able to hear a difference.  It is subtle.  I'll have to plug into the Foobar ABX test to confirm.  But won't be able to take the time to do so right away.

 

Edit to add a spoiler for my initial guess:

Warning: Spoiler! (Click to show)

Spoiler for my initial guess based on listening with the SR325is

A is lossy

B is lossless

The difference is the "space" in the sound and recording, but it's subtle


Edited by Ham Sandwich - 10/18/10 at 11:01pm
post #6 of 88

I'm pretty sure "B" is the lossless.

post #7 of 88

B is lossless.  I hear the guitar on the right distort a little at the end of that loud note right at :30 sec...other than that they sound similar.

post #8 of 88

Too easy haha. Definitely B. Only need the first 2 seconds. Btw I do own the Rodrigo y Gabriela CD, so maybe that has something to do with it, but I doubt it. Still, as usual with AB tests, I must pick up the difference right away (playing both files no more than twice each) or else I will either get the answer incorrect or just not hear a difference at all. Using Grado HF-2 btw.

post #9 of 88

After listening each track several times, I could hear a slight reduction in frequence response in file 1_b. I would say that this is the mp3 encoded at 128 kbs and 1_a would be the lossless flac. Both files were very similar through the Earsonics SM3.

post #10 of 88

1_B is the lossless.

post #11 of 88

 

Nice try but you can't fool me. The lossless file is in fact 1_C.

 

I'm fairly confident I can hear a difference between the two, but identifying which is which is another matter. B sounds a bit more ragged than A, but by the same token, A sounds a bit less ragged than B. Meaning, I'm not sure if A is lossess and the encoding is making it sound more ragged or if B is lossless and the encoding is making it sound less ragged.

 

I think a better test would have been to have the lossless file known, and then compare between B and C, one of which would randomly be lossless and the other lossy.

 

Anyway, if I were to hazard a guess, I'd go with A being lossless.

 

se

 

post #12 of 88

I made a careful effort to not read any comments before deciding, and my verdict is B is lossless.


Edited by dmac03 - 10/19/10 at 8:28am
post #13 of 88

Failed miserably, even though I know which is which (I cheated) I could not DBT them in FooBar 2K. Not music I am familiar with or particularly like but that is no excuse,  I could not to pick them out, 52 year old ears and improved codecs I guess ....ho hum. The biggest differences are in the higher octaves where my hearing is age-related poor and at ~ 30 seconds there is a segment where the mp3 high frequency roll-off is more pronounced as the lossless maintains the higher frequencies at relatively high amplitudes.

 

Interesting, thanks for that.

 

However, rather than folks opining it is def A or it is def B put them both through the ABX comparator in FooBar 2k and score how often /20 you correctly identify A = X or A = Y , that is much stronger evidence of detection, mine was a random 50% with two options correctly guessing once is not great evidence...


Edited by nick_charles - 10/19/10 at 9:04am
post #14 of 88

 

 

added spoiler thing also.

Warning: Spoiler! (Click to show)

I put my headphones on for this, and i am CONFIDENTLY saying that it is 1_B, 1_A is lossy encoding through and through.

 

edit: using sennheiser HD555 out of my Aune.

 

 

edit: using sennheiser HD555 out of my Aune.

 

edit2: and indeed, tested using a program.


Edited by googleborg - 10/19/10 at 9:04am
post #15 of 88

Am using some random philips iems lying around. Should be B for lossless?

Did a abx on foobar and got it 7/9 times. Shall try again with proper iems later.

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › Blind Test - Take Now!