Are you a critical listener? Take the mp3 test

[Kodhifi]

Go to mp3ornot.com and take the test using the best headphones you can. See if you can tell the difference between a 128kbps and 320kbps mp3 file. They use a random selection of music, some of it's a little disingenuous as they pick some tracks that would tend to mask compression artifacts, so this is a real challenge even for an audiophile. I did it on my work computer using Dell on board sound and my HD201's, a $15 headphone. Nailed 3 out of 3 on my first try but it was difficult. Some tracks I listened to literally 20 times or more, concentrating on different instruments, their dynamics. 
 
What you want to do is understand what mp3 compression artifacts sound like, and where they might present in a mix. It will be more audible on higher frequency percussive sounds like cymbals, and it affects the stereo signature, so panning percussive sounds are even better for critical listening. Try to find something in each song that fits that description, and concentrate on it like your life depended on it. Try to find any difference in that sound from one to the other, then compare to X. Then post your results!
 
 
 
Here are my results
 

 

[HamilcarBarca]

Quote:

Go to mp3ornot.com and take the test using the best headphones you can. See if you can tell the difference between a 128kbps and 320kbps mp3 file.

 
While this test may be entertaining, the results of superficial tests like this can't be generalized to draw meaningful conclusions about MP3 sampling rates in general. Plus, it's unrelated to the topic of this forum, full-sized headphones.

 

[Kodhifi]

Quote:

 
. Plus, it's unrelated to the topic of this forum, full-sized headphones.

I disagree. The idea is to take the test on headphones. Also it's rare not to hear somebody mention the bitrate of your mp3's when recommending a headphone. IE " If your MP3's are 128 you're not going to notice much of a difference upgrading to HE500's"
 
Critical headphone listening in my opinion belongs right here.

 

[Tangster]

There's a volume imbalance between the two bit-rates. 320kbs also takes longer to load.

 

[HamilcarBarca]

Quote:

Quote:

 
. Plus, it's unrelated to the topic of this forum, full-sized headphones.

I disagree. The idea is to take the test on headphones. Also it's rare not to hear somebody mention the bitrate of your mp3's when recommending a headphone. IE " If your MP3's are 128 you're not going to notice much of a difference upgrading to HE500's"
 
Critical headphone listening in my opinion belongs right here.

 
It's quite common, as you point out, for people to make unsupported claims about MP3 bit- and sample-rates. Still, it seems more appropriate for discussion in the Sound Science forum, unless you're simply seeking anecdotes. However, I won't lose any more sleep if you feel the discussion belongs here. I apologize for intruding.

 

[viralcow]

Quote:

 
While this test may be entertaining, the results of superficial tests like this can't be generalized to draw meaningful conclusions about MP3 sampling rates in general. Plus, it's unrelated to the topic of this forum, full-sized headphones.

 
I think it's more like we already have a dedicated sub-forum for exactly this sort of stuff- sound science.
 
There's a lot of interesting stuff to read down there.
 
And yes, the difference between 128kbps, 320kbps, and lossless are actually quite slim if they are well-recorded to begin with. But most of us with good ear can tell the difference between the 128kbps and 320kbps (albeit not very easily). I'm not home right now but when I get back I'll take this test and see what I get.

 

[Kodhifi]

Quote:

There's a volume imbalance between the two bit-rates. 320kbs also takes longer to load.

Hmm, when you say load I'm not sure what you mean. Page gives me 3 buttons, clicking any of them instantly results in sound. Volume sounds the same between them both as well, but then again I'm in a noisy environment on very low end headphones.
 
"And yes, the difference between 128kbps, 320kbps, and lossless are actually quite slim if they are well-recorded to begin with."
 
I think that the encoding process matters for a lot here as well. Encoding an audio signal used to be so tricky people would farm the work out to other companies such as DVD mastering.
 
The mp3's on that page are pretty good even the low bitrate ones and as you said pretty close. I had to listen a lot before comfortably deciding. But I remember back in the day when mp3 was an underground kind of thing listening to a Police mp3 album and it sounded terrible at 128kbps. The cymbals especially were just this wishy washy mess, tons of compression artifacting in the songs etc. But this was all in the days before LAME, hell it was in the days of LPT connected ZIP disks........<shudders>
 
Now days it's much closer provided they have been properly compressed to begin with. I use 320 at home but I don't care about disk space and at that bit rate it is incredibly easy to get an artifact free encoding.

 

[pp312]

Quote:

I did it on my work computer using Dell on board sound and my HD201's, a $15 headphone.
Here are my results
 
 

 
I've never found the differences between lossless, lossy and all the different bitrates to be anything like as great as is usually made out here. In fact to me to insist on FLAC over 320 MP3 is ridiculous. However, the real mystery here is where you got the HD201 for $15.

 

[Kodhifi]

Quote:

 
I've never found the differences between lossless, lossy and all the different bitrates to be anything like as great as is usually made out here. In fact to me to insist on FLAC over 320 MP3 is ridiculous. However, the real mystery here is where you got the HD201 for $15.

Frys had a sale about 3 years ago, I bought 1 just for work headphones, and I liked the sound VS cost so much I bought 3 more pair just so I'd have enough to last to old age should they be discontinued.

 

[pp312]

They are pretty good performance v. cost, and twice as good at $15. Pity this isn't the HD201 thread.  

 

[chewy4]

Quote:

Hmm, when you say load I'm not sure what you mean. Page gives me 3 buttons, clicking any of them instantly results in sound. Volume sounds the same between them both as well, but then again I'm in a noisy environment on very low end headphones.
 
"And yes, the difference between 128kbps, 320kbps, and lossless are actually quite slim if they are well-recorded to begin with."
 
I think that the encoding process matters for a lot here as well. Encoding an audio signal used to be so tricky people would farm the work out to other companies such as DVD mastering.
 
The mp3's on that page are pretty good even the low bitrate ones and as you said pretty close. I had to listen a lot before comfortably deciding. But I remember back in the day when mp3 was an underground kind of thing listening to a Police mp3 album and it sounded terrible at 128kbps. The cymbals especially were just this wishy washy mess, tons of compression artifacting in the songs etc. But this was all in the days before LAME, hell it was in the days of LPT connected ZIP disks........<shudders>
 
Now days it's much closer provided they have been properly compressed to begin with. I use 320 at home but I don't care about disk space and at that bit rate it is incredibly easy to get an artifact free encoding.

They're not volume matched. This is something you can test for, it's too small of a difference to easily notice by ear but a slight change in volume can result in one file sounding better for unexplainable reasons.
 
When compressing MP3's the gain is lowered after a certain bitrate, which is done to avoid clipping I believe. A 320kbps mp3 will always be a bit louder than a 128kbps mp3 until you change the gain on one.

 

[dolor]

Interesting comment from a spanish speaking dude from that site "Well I downloaded the two files, and I edited and are exactly the same, just change the 128 is the 320 is stereo and Joint Stereo. Are otherwise identical, and no difference audibly. But my daily experience, I do appreciate differences between recorded at 128, 192 and 320. Although between 192 and 320 the difference is minor. Greetings to all."

 

[wolfetan44]

That song is horrible.

 

[Kodhifi]

Which one? I loved the Flamenco guitar but I can't stand Juliette Lewis. /gag
 
I re-did the test last night on my new DT 990's and it was much easier but partly because I already knew what to listen for.
 
Julette Lewis song I listen for the cymbal at the very end of the sample.
Spanish Guitar I listen to the stereo percussive noises that go left right left right to the end of the clip.
The other one I'm listening to the drums again, the jazzy one.
 
I got 100% after 8 plays. It's just the same songs repeating but the X is random every time. The differences are incredibly slight, you really have to put on your analytical hat in order to tell the difference because both samples were well ripped and well recorded using exact audio copy. You can get a good sounding 128kbps but you have a much smaller margin of error in the encoding process where as 320 will consistently sound good regardless of how carefully you encode.
 
 
If any of you do graphic art type stuff you can see the visual equivalent of mp3 compression artifacts when saving from lossless bmp, tiff, or png, to jpg if you know what to look for. Text for instance, especially very small font text against a contrasting background, will lose some of it's contrast because the edges will be summed a bit.
 
Mp3 is similar but instead of edges, it's quiet percussive sounds against a louder fore ground sound, and it's affected more when it's stereo due to joint stereo encoding than mono.
 
The flamenco track for instance is perfect for listening because the percussive guitar sounds go left, right, left, right, and are quieter than the guitar strings. On the 128kbps they lose some stereo separation and a little clarity, similar to text looking fuzzier or less contrasty in a .jpg.
 

 

[extrabigmehdi]

I thought the test flawed, because I  found the difference too obvious (done only 2 test,  but I  didn't hesitate at all).
They didn't specify which encoder was used, and I'm thinking there's perhaps a volume imbalance too.
I know I  reach relatively easily transparency, by  using lame encoder.
Unless using a high end headphone makes a difference (the guys at hydregenaudio, claim it doesn't matter).