Do 320kbps mp3 files really sound better? Take the test! Do 320kbp, Maybe unscientific, but interesting results Options

[2Bdecided]

128kbps vs 320kbps samples here:

http://www.noiseaddicts.com/2009/03/mp3-so...y-test-128-320/

EDIT: If you want to try the test yourself, DO NOT read on - there are spoilers further down this thread































































The reason I posted it (and the reason it's interesting IMO) is because most people got it "wrong".

43634 votes for 128kbps being better
vs
39578 votes for 320kbps being better

FWIW I thought both sounded like low quality vinyl (a bit muffled, a bit noisy), but the 128kbps version had extra artefacts.


Cheers,
David.

This post has been edited by 2Bdecided: Feb 23 2012, 19:58

[Porcus]

The reason I posted it (and the reason it's interesting IMO) is because most people got it "wrong".

Well ... announcing the results will lead to bias among future voters.

[pdq]

How will that lead to bias? Will people start picking the one that sounds worse because they assume that is the 320kbps?

Edit: The problem that I have with this test is that the file that "sounds worse" may, in fact, be closer to the original (which was not made available).


This post has been edited by pdq: Feb 23 2012, 18:32

[KMD]

I plugged in a pair of ear buds into my laptop to do the test and it was imediately obviouse that the second clip had horrible artefacts. so I'm suprised so many people got it wrong. The second clip was louder so that does fit in with the classic loudness A B test loudness bias concept. By the way the hearing test on that site is dangerouse - playing 20Khz sine waves into your amp can fry your tweeters.

[Porcus]

How will that lead to bias? Will people start picking the one that sounds worse because they assume that is the 320kbps?

That is a very plausible placebo effect.

[2Bdecided]

I plugged in a pair of ear buds into my laptop to do the test and it was imediately obviouse that the second clip had horrible artefacts. so I'm suprised so many people got it wrong. The second clip was louder so that does fit in with the classic loudness A B test loudness bias concept.

Now that will ruin the test for anyone else. Couldn't you have said "one of the clips had horrible artefacts"?!

Where's a mod when you need one?

Cheers,
David.

[2Bdecided]

How will that lead to bias? Will people start picking the one that sounds worse because they assume that is the 320kbps?

That is a very plausible placebo effect.

They're already doing something strange already. The difference is too great to be chance, but favours the wrong result.

I suspect it's the simple perception bias that the second one sounds better (very common), or that the higher numbered clip has the higher bitrate.

Oh **** it, I've done it now! Sorry for criticising KMD!

EDIT: edited first post, problem solved.

Cheers,
David.

This post has been edited by kode54: Feb 23 2012, 20:13

[Kohlrabi]

This was ridiculously easy. If the voting results they state after the test are correct, I'm surprised and shocked.

The comments on that site are especially brilliant. Some "audiophiles" who chose the wrong sample and try to argue around their simply bad hearing. Also one comment from a quite clever fellow:

CougarMan said:

I chose Clip 1 because it took longer to load.

Reply
# 4 February 2012 at 5:31 pm

This post has been edited by Kohlrabi: Feb 23 2012, 20:36

[onkl]

The hihats/snares/percussions (what ever it's called) are obviously smeared, so obvious that the used encoder probably isn't the best.

This post has been edited by onkl: Feb 23 2012, 20:27

[zima]

Another possible factor in such distribution of votes: people might be getting used, after a decade+, to how "worse" bitrate sounds - we have a precedence of this effect with vinyl, and I remember also a thread here (linking to an article in some high-profile ~newspaper) about such & compact cassettes.

PS. I seemed to remember reading somewhere about some studies of such effects... luckily, it turned out to be in the most straightforward of places:

A test given to new students by Stanford University Music Professor Jonathan Berger showed that student preference for MP3 quality music has risen each year. Berger said the students seem to prefer the 'sizzle' sounds that MP3s bring to music.[43]
http://radar.oreilly.com/2009/03/the-sizzl...d-of-music.html

Plus the "this sounds right & just like I'm used to" effect could even be sort of more prominent, I imagine, on typical cheap headphones, speakers, poor listening conditions.
(is there still expectation of censoring such this low in the thread? )

This post has been edited by zima: Feb 24 2012, 01:09

[matt the cat]

Well I noticed the difference there.

[JJZolx]

That was pretty obvious, even played through my computer's audio system.

You selected Clip# X …Congratulations!

Congratulations for what???

[hellokeith]

I picked correctly. Stereo imaging is better.

[_if]

That test was all right for what it was, but after browsing some other articles, I have to say, that website is truly awful. Everyone should be able to get a good laugh out of this, in which the author complains about the poor quality of MP3s and says the one given in the test of 128 vs 320kbps was a bad example because the "recording has no 'HIGHS'!" (despite a very present hi-hat), and makes some hilariously foolish statements: http://www.noiseaddicts.com/2012/02/sound-...ssed-audio-mp3/

The big kicker is the final paragraph:

Lastly–I have spent hundreds of hours doing tests with HD music and conversion formats and I have noticed something amazing [!!!] I have been able to make a wave file smaller than an mp3 at 128kbs by changing the codec to mpeg layer 3 and 16 bits 44htz stereo and sound identical to the original wave file. Some of the music files were only 3 megabites. Try to do that with an mp3 recorder!"

Impressive.

[Mach-X]

huh, well, for the guys who were able to pick it out based on 'obvious' artifacts, congrats, i wonder if you didn't cheat by eq in'g your system before hand to reveal flaws....heh heh....kidding, I didn't detect any obvious difference, and although after around 20 or so listens including rapidly switching back and forth, I WAS able to detect the correct sample, but again it took a lot of repeated listens, and no, I would not be able to abx them reliably. I am _almost_ inclined to believe I could hear the different lowpass, but I have to be careful not to violate TOS #8 in speculating about it...

[LithosZA]

That was really easy, but I think they might of used some really old/bad MP3 encoder for the 128Kbit/s version. Can hear an obvious annoying flaw between 5 and 6 seconds.

[kritip]

I've never done well at sound tests and ABX, but without reading anything other than the intro and link, I did the test on my laptop speakers with the TV on in the background and got it right. Surprised so many people prefer 128kbps! Of course that may have just been chance

[yourlord]

Just did this test without reading the rest of this thread and got it right. Did it on a POS cheap laptop with a pair of Sennheiser HD-202's.

The 128Kbps clip had a very low level noise that was audible in the quieter parts and the high hats had a fairly obvious warble.. I listened several times before deciding to make sure the effects were not present in the other clip.

[thrillscience]

Well, I picked the 320 Sample, but I was listening with headphones, and I've had a lot of experience being a critical listener to audio compressors. That being said, they're both probably pleasant enough to listen to for a typical pop/rock type of track on YouTube, etc.

However neither sounded top-notch, compression aside. It sounded like one big blur, either way. I think if it had been a more "crisp" recording, more people would have got it right.

Most of my listening / recording is with solo piano music, and I've become very good about telling what's going on with any recording I hear, from compression to type of piano, to microphone type and placement. With other types of music, I'm not as good.

This post has been edited by thrillscience: Sep 16 2012, 20:55

[pisymbol]

Well, I picked the 320 Sample, but I was listening with headphones, and I've had a lot of experience being a critical listener to audio compressors. That being said, they're both probably pleasant enough to listen to for a typical pop/rock type of track on YouTube, etc.

However neither sounded top-notch, compression aside. It sounded like one big blur, either way. I think if it had been a more "crisp" recording, more people would have got it right.

Most of my listening / recording is with solo piano music, and I've become very good about telling what's going on with any recording I hear, from compression to type of piano, to microphone type and placement. With other types of music, I'm not as good.

This is the best comment of the bunch and really explains the results in a nutshell.

Let's see:

a) The recordings are both crap so does it really matter?

b) The 5-6 second distortion in the 128kbps sample threw me for a loop....without being familiar with the material at all I wasn't sure if the top one was missing it or it really was distortion? How do you know?

c) I need to hear the uncompressed version as a reference OR the whole thing is bogus to me...you must establish a reference in order to establish metrics, e.g. "Which one sounds more true to the original uncompressed version?" To my ears, thats the only issue that matters.

Finally, storage is cheap. Long live FLAC. :-)