Golden Ears result

[mtillman]

OK, here's what you've all been waiting for...

I think mp3 wins almost outright

Thank you all for taking part. I think only Tim D came anywhere close - I reckon he more or less spotted the 128k mp3, but also heard quite a few things that weren't actually there, or at least weren't 'faults'.

I can't agree that the opening is clipped. Certainly the wav isn't, and I'd be staggered if the original recording was. I'd also be a bit surprised if it hit any limiter or compressor in the recording chain, to any significant degree anyway. I think what you're hearing is the real sound of the real piano in the real room...

For those of you who want to buy this wonderful piece of music, you'll find it here

Can anyone translate what I guess is Breton?

OK, now I'm off to find some cymbals

 

[Tim D]

hehe, Wow, actually my finding really does mirror your results. I just didn't realize you had both Mp3 & alt-preset encoded Mp3 thrown in.

I took less than 10 minutes to do it with AKG240S hooked up to my Onkyo USB soundcard and most of the time was just downloading it and playing the beginning (or whatever finished downloading). I can tell you that I still think there IS some clipping and warbling plainly evident at the points I mentioned from my setup, wheter or not some other artifacts came in during transfer or my soundcard can't handle severe transients I don't know.

The means I used most to distinguish besides the easy warbling (well easy for me to hear at least) in the 20-30 is what I call "dead air". Mp3 typically cuts off at above 16khz giving recordings a sense of dead air...I *could* detect air in the first 10 seconds but also clipping...but I know that Mp3 doesn't cause clipping if properly done but normal recordings can have it. Shortly after I felt dead air...that the 16khz was removed. I felt air restored after 1 minute...removed after 1:35...and I just got bored afterwards.

I didn't know there was two levels of Mp3 used...but that would explain why I felt that although some segments both lacked air and warbled...some didn't warble but still lacked air.

Mp3 surely didn't win this test for me cause your results were what I heard.

BTW if you throw something in like powermetal or speedmetal...mp3 gets slaughtered.

Since I got pretty darn close...what do I win...Max...Orpheus?

I didn't even bring out the heavy guns and burn the thing on CD, use an amp, or take out the Ety's...in which case I'd tell yah the height and weight of the singer, his haircut, the color of his eyes.

 

[PodMan]

LAME's alt-preset Standard cuts off at 18.671-19.205khz.

 

[Tim D]

I still detected the 1 minute transition point which went from alt preset standard to source.

I'm pretty sure the 128kbit mp3 cuts shortly above 16khz...if not cut, than it doesn't accurately do high frequencies or at least attenuates before having a severe cutoff.

Even if alt-standard cuts off only at 18khz+, I still heard it.

 

[PodMan]

yeah, the 128kbps cutoff is around 16khz. I think cymbals would probably be much harder for mp3 to compress than a piano.

 

[mtillman]

Quote:

Originally posted by Tim D Since I got pretty darn close...what do I win...Max...Orpheus?

Close, but no cigar

And no prize at all for English comprehension

You missed the bit in the instructions about two different mp3 encodings, AND you missed the bit about the prize - which is a video of me eating your hat.

 

[Tim D]

Bah...I only read the instructions if something starts smoking and smelling really bad.

Quote:

AND you missed the bit about the prize - which is a video of me eating your hat.

I've eaten in the UK before...I believe we're talking about my prize...not yours!

 

[mtillman]

Hey, that was genuinely really funny

I get the last laugh too, because I can't hear anything above 16k.

 

[lan]

Hmm, so the end part really was the WAV part. I thought it was the higher type of MP3 but um, your setting for alt preset standard isn't one I'm used to. I don't know what's different but it just sounds worst than it should.

The noise floor and noise shaping kind of disturbed me so it was tough choosing the WAV portions in the quiet parts. My selection of WAV in the middle was kind of limited because of variations on the voice. I guess the singer was ok one moment and more straining / distored another.

I don't believe I could select the beginning WAV because it just lacked openness which I correlated with MP3.

 

[Pepsione1]

Wow I can't believe I actually got the most of the wave parts right. I was really begin to question myself when I saw all the votes for mp3 in the first ten seconds of the clip. Good thing my judgement remained unchanged. All of the listening were done on the my computer speaker at moderate level. Maybe it was luck but I sure am surprised by my own ears.

 

[Squalish]

I only detected the opening change around 10 seconds(there WAS a bunch of distortion at the beginning, WAV or not), and a bit before 1 minute(makes sense that I didn't detect @ 50 seconds, there isn't much volume until about 58 seconds). Later on, I knew there was SOME distortion, whenever the voice hit a peak note, but I couldn't differentiate between them.

Also, frmo your orig post, I thought you were mixing 128k, alt-preset-standard, orig wav, and alt-preset extreme or something like that.

All this shows is how badly 128kbps can be done. Most stuff you DL is nowhere near the fidelity of the stuff in this clip.

 

[mtillman]

BTW, I've just spotted that the hf cutoff for the 128k mp3 is 17500Hz.