Got it right after switching between the two for 15minutes...using my ER4+Bass driver Customs with Yulong D100 and m-Stage
The 320 have a better decay on the treble, making the cymbals sound crisp.
I got it right straight away,but i couldn't hear anything above 12k on the test.Must be why my 325i's never sound harsh,always sweet. One advantage of being over sixty.
Having said that,i just did the klippel test and heard the distortion all the way throu to the end of the test -46db.So there can't be to much wrong with my hearing,i just can't hear a dog whistle
The answer to your question is,yes 320 does sound better,on the phones at least...
Well to me the difference was quite noticable both in vocals and cymbals I thought the 320kbps sounded just fine as I expected it and thought for myself the 128kbps can't simply sound like this. Especially at the time he sings "Which way you'll going" there's difference in the vocals especially at the first word "which" (128kbps sounds a bit more muffled) and the cymbal almost always sounds cleaner/crisper/less restricted (difficult to put in words). But yea it's a bad sample to use to demonstrate the difference for sure but that was probably made on purpose anyway.
EDIT: Also be sure to put max volume on that volume slider next to play button if you didn't, it's a bigger difference then vs using the default 2 sliders from max vol. So to those who thought 320kbps sounded very poor that might be it.
Songs that would demonstrate the difference a lot better would be for example this http://www.youtube.com/watch?v=zW9-YXpa2L8&hd=1 :D Or something like "Sailor - A Glass of Champagne" from the 70's which uses such a lot of different instruments.
Yea I have a feeling that a lot of those who have hard time to tell are only using Realtek onboard and then it's no suprise. The Audigy 2 ZS chip with kX Audio drivers which is older than the Realtek HD onboard (ALC889A) chip is already a lot more revealing in my case. I don't know how people can insist using Realtek onboard, buying a cheap 30~$100 DAC today already grants more SQ improvement than spending $150 on a new headphone.
With my Pro 900s, I listened to each clip twice. I did it once at low volume and then once at high volume. At first, both sounded exactly the same. It was at higher volume that I could hear more "crap" in the music, it just sounded more artificial, and instantly I knew which clip it was. It was an instant recognition once I heard the off sound of the noise. If I were not listening for a differences, the "garbage", and had of just been listening to either one individually without it being a comparison, I wouldn't have known...However, comparing the two side to side, it was unmistakable.
The difference in dynamic range is quite noticeable.. this test is nothing compared to this
I took that test and got down to -21db before I got some of the a/b comparisons wrong. Got -21db wrong twice, and then on third try got down to -30db, before it ended the test for me.
Got the answer right, on my uni rig: Asus netbook directly to BX8a speakers. The track wasn't too hard for the 128 encoder though, not too much going on.. listened to each track once completely, then once the beginning and must say that the difference was not that noticeable. What did it for me in the end was the faster response and better bass of the 320 (bass especially noticeable with speakers).
I completely failed with the gear I had on my head at the time - a cheap 30 euros headset ("full size" though) and a MBP.
But the sample track really isn't that hard to encode anyway. I tried with the same setup (cheap headset + MBP) the same kind of blind test with much harder to encode albums (classical music, MFSL ones, etc.) and the difference was quite obvious.
MP3 removes a lot of harmonics. So it is no surpise that a track that's not that rich in harmonics can be misleading with crappy headphones.
I got it right quite easily using my laptop speakers!!! I can tell there were better dynamics and the bass was a lot cleaner.
Either that or I got lucky...
I shall have my gf help me with a blind test with my normal setup.