Can you tell differences in file compression?

Old article, so feel free to delete this if it's old news.

I came across this NPR article and I'm interested to see if some of the better ears out there can do better than I did (2/6). I found the Coldplay song particularly difficult due to the lack of treble detail.

How well did you do?

 

interesting test... 2 out of 6 here also - and I'm not intimately familiar with these tracks. I still thought I might do better - but I'd be interested to find how much this changes when played on an ideal setup - like via KS or Wasapi on a good player on my minimal OS on my Mac Mini...

 

I wasn't familiar with the tracks either, which made it more interesting.

 

Unfamiliar music, non-ideal listening while eating breakfast, got 4/6... which is all kinda moot because I don't think these were volume matched anyways.

 

6/6. It's not a perfect test but it helped convinced me of what i felt... That i could tell 320 from lossless.

What surprised me was that it was out of my dells stock realtek dac/amp headphone out to HD600.

 

4/6, I retook it a few times after and I gotta say that cold play is nigh impossible I somehow picked 128 every single time.

 

5/6
Gungnir Multibit > Val2 > HD800s.

Its all in the timbre / treble. Failed the Neil Young one.

 

2/6, harder than I thought, but I probably didn't give it enough time/effort and listened to each track just once. That or I genuinely prefer 320kbps (my other 4 choices). Gungnir Multibit > Mjolnir 2 > HD800

 

Yeah, Coldplay recordings are pretty terrible. No surprise there.

 

It would be a more notable test if everyone took it a second/third/fourth time. I might try again tonight to see if I can do better. Wondering if some the high scores above may do more poorly a second time...

 

I took 4 times, First 4/6, Second 3/6, Third 4/6, Fourth 4/6. I seriously picked 128kbs cold play every single time. Neil young was the other consistent guess.

 

cold play is really the only one i cant pick out ever. 5/6 4/6 4/6. Katy Perry and Jay Z were the easiest for me.

 

Are six songs enough for a sample? There is relatively high probability that you'd nail it randomly.

 

Nice! Yeah, I found Jay Z to be the easiest at first listen. Coldplay and Neil Young are definitely the hardest due to the lack of treble detail. I'm surprised you found Katy Perry easy, I must have been listening for the wrong details in the song.

I find that compressed files sound a lot like over-mastered files, which makes some of the newer recordings harder. But again, old recordings aren't saved if they can't have treble in the first place!

 

http://abx.digitalfeed.net/list.html

I do think it's a bit funny that one 'good' test will make people instantly jump to the conclusion that they -can- hear a difference or that there is an audible difference at all without even questioning the validity of the test in question.

Dare I imagine how many people are convinced they can hear the difference because of that biased Tidal test.

 

Some of the differences are pretty noticeable in this test, it may also be biased.

 

Best way would just be to not trust any of these test sites, do your own proper lossy encode with a lossless file you have of a song or two you're familiar with and run the foobar ABX test a few times with consistent results, but then again that's too much effort and it's fatiguing etc etc until we run full circle and people in one crowd continue to scream about hearing the difference with the other screaming back at them to prove it.

 

3/6. But I have never convinced myself I can pick between FLAC and 320kbps so if I count the ones I picked 320, then I'm 5/6.

 

Yeah, def seems like these tests can only be set up so well - or in some cases might be intentionally biased by design. And if MP3 was specifically designed by somebody who was using Suzanne Vega's song as reference material - well, I can see how they might have eventually tailored their optimizations to it and MP3 might reproduce that track especially well.

I know to myself that I can hear the difference because I've gone back and forth between the two on quite a bit of music in my library. But at the same time, it's importance to me has declined as I've heard far bigger gains by paying attention to my overall chain. I've actually been quite impressed how immersive MP3 music can be on a good set up - and as far as I know, good mastering probably matters a lot more than MP3 vs lossless.

But I still prefer lossless and it's nice to have 24/96 when you can...

 

While I can't speak for the validity of this specific test, if you can't consistently pass a blind test with good scores, you can't tell the difference. It's as simple as that. No shame.

In the past, I've ripped a CD to WAV and converted to 320 using Foobar in order to do my own A/B. I didn't set up Foobar to do the blind (not sure how to do this) but I walked away from the experiment happy that I couldn't tell the difference. As soon as a snare sounded more crisp on lossless, I'd switch over to 320 and find it just as crisp. In the long run, this just means I can utilize a 320 streaming service and use less hard drive space.