Sign in to follow this  
Followers 0
rhythmboy

Wow 128kbps mp3 really is dodgy... Gibbs Effect

18 posts in this topic

I don't claim to be the first to discover this, but I've just had my whole sense of what's wrong with mp3 clarified beyond doubt tonight.

Big thanks go to Glen our tech maestro at Uni who had the curiosity to try this out, and I've just repeated the experiment tonight. Such a simple and obvious example of the "Gibbs Effect":

http://www.dspguide.com/ch11/4.htm

I'll just let your ears be the guide... check out the top end ;)

For playability the following examples are all in 320kbps mp3 but I've listed their original rendered format. Processed in Pro Tools and converted in itunes:

STEREO TO MONO SUM - 16/44.1 WAV - IN PHASE

[mp3=400,30,0,center]http://www.soundpunk.com/downloads///home/RBstuff/monowavin.mp3[/mp3]

Good - the kick, snare, bass and vocals are all there... now let's flip... 

:P

STEREO TO MONO SUM - 16/44.1 WAV - INVERTED R CH PHASE

[mp3=400,30,0,center]http://www.soundpunk.com/downloads///home/RBstuff/monowavout.mp3[/mp3]

OK, that's what I'd expect... now let's try some mp3! 

;)

STEREO TO MONO SUM - 320KBPS MP3 - IN PHASE

[mp3=400,30,0,center]http://www.soundpunk.com/downloads///home/RBstuff/mono320in.mp3[/mp3]

What do you think? Satisfactory? Wonder what happens if...  :

;)

STEREO TO MONO SUM - 320KBPS MP3 - INVERTED R CH PHASE

[mp3=400,30,0,center]http://www.soundpunk.com/downloads///home/RBstuff/mono320out.mp3[/mp3]

Not hard enough! Go harder with the compressionisms!  :bang:

STEREO TO MONO SUM - 128KBPS MP3 - IN PHASE

[mp3=400,30,0,center]http://www.soundpunk.com/downloads///home/RBstuff/mono128in.mp3[/mp3]

(Average punter): "Well! Not too shabby! I'd buy that..."  :cans:

STEREO TO MONO SUM - 128KBPS MP3 - INVERTED R CH PHASE

[mp3=400,30,0,center]http://www.soundpunk.com/downloads///home/RBstuff/mono128out.mp3[/mp3]

Oh 

:(

Once myspace are back from routine maintenance I'm going to download the same tune off my page and do it to that one - can't wait for the results!

Share this post


Link to post
Share on other sites

I did this myself a while ago but slightly different, I posted it on ITM  8)

Mine was different in that I converted the stereo file to an MP3 then inverted the whole file and summed it with the original WAV. This leaves audiable only what was in the WAV but not in the MP3. The way you did it shows how bad MP3s encode difference information... Very badly...

What surprises me more than how bad 128kbps are is that 320kbps is actually quite alright as long as the very HF sounds aren't too important to the mix.

Here is my post on ITM. The sound files aren't there anymore, they are on a hard drive sitting on my desk so I may grab them off one day but not now. I did do a spectrum analysis though and they are still there.

http://www.inthemix.com.au/forum/showthread.php?t=215550

Share this post


Link to post
Share on other sites

^ Great idea Anders, really effective demonstration.

I have some new lab exercises for audio theory class next semester ;)

BTW I did download the tune from myspace today, but interestingly the download sounded heaps better than the flash player they use - basically the same as the original mp3 I uploaded. Never bothered before so I never knew it was that way...Anoth

Another phase test being made as I write - will post back in mojoe's thread later:

http://www.soundpunk.com/index.php?topic=1490.0

Share this post


Link to post
Share on other sites

Following from the "Anders method" of phase-testing the mp3, I tried it out and here's the result. Used the same audio as above for reference.

WAV summed to mono, mixed with 128kbps equivalent, phase inverted:

[mp3=400,30,0,center]http://www.soundpunk.com/downloads///home/RBstuff/mono128fullinvert.mp3[/mp3]

This seems to answer a suspicion many have that they sense a loss of bottom-end fullness out of mp3 perhaps. A strong presence of kick and bass here says to me there's a loss proportional to it in the normal mix - or at least enough phase drift present that is capable of doing this damage.

Share this post


Link to post
Share on other sites

Wow that had alot more bass in it than my test... Its probably because that song has alot of difference information in the Mid and HF but bass is mono. When you mono the track, the bass will jump in volume but the rest won't. Just a thought. Try leaving the file stereo.

Also i've found that different styles of music are processed better than others. Tracks with long sustained notes like guitar, piano, pads and long cymbal hits often loose alot of clarity as the tail end of these sounds are left out of the compressed file. Try this difference test on that kind of material and you will hear the sustains clearly in the difference file.

If you leave the files stereo and use a decent listening room (not my room) you could listen for stereo image issues caused by encoding too. Let me know if anyone tries this.

Share this post


Link to post
Share on other sites

Actually that last example makes a pretty cool shaker.

Nice work on the experiments guys. It's just such a pity that your average person really can't tell the difference between a wav and 128 kbps mp3.

Share this post


Link to post
Share on other sites

Hey come on now... Thats just casting the line!

Don't you mean "The average person doesn't really care about the difference"?

Share this post


Link to post
Share on other sites

It's just such a pity that your average person really can't tell the difference between a wav and 128 kbps mp3.

Don't you mean "The average person doesn't really care about the difference"?

I don't have an answer ;)

Just reminded of a project Amz-Star from round these parts did for me a couple of years ago for a project at Uni. Amz performed WAV vs mp3 listening tests on people of different age ranges. From memory the outcome was that younger people cared a lot less about the difference than older people (ie in their 60's) did.

For the generation who grew up on vinyl records and valve radios, the reactions to the difference were stronger and more negative than those already well accustomed to mp3 and ipod playback.

It may be safe to assume that older people would notice less differences because we'd assume their hearing is less optimal than in younger people. However,  an older person with decent hearing will have had a lot more experience listening to different musics than a young person. This experience contributes significantly to our ability to discriminate difference.

Share this post


Link to post
Share on other sites

Hey!

Yeah, I did do a project on WAV Vs mp3 audio.  What I did was had a survey and used two songs, 'Trouble' by Pink and 'Leave a Note' by Missy Higgins.  Pink has alot of 'ear candy' and Missy is just a piano and guitar (for that song) so therefore it made it interesting to do the test.

I used a set of reasonably cheap computer speakers to do the testing as that is what the music would be played through, except for mp3 players, but i didn't think it would be hygenic to have all of the participants using the same ear buds!

The results that I came out with were that the majority of the participants could not hear the difference between the WAV file and the mp3 128 kbps file.  When I then played the 64kbps file around half of the participants could hear a difference.

I had two participants that were in the 60 + bracket, and one of them loves music, the other is not really into music.  The one that was into music could not really hear any differences in any of the tracks, but the other one could hear some differences (if I remember correctly RB, as I dont have my assignment in front of me right now!)

It was a very interesting process though

I even found out that when I converted the audio in iTunes and played it back in Pro Tools the LE version it was fine in, but the HD system I played it on too kept peaking!  So the conclusion that RB and I came up with was that the HD system was alot more susceptible to quick changes in the audio wave form.

If anyone was interested I could get my essay out and put it up as a PDF, but only if anyone wants a read.

Amz :cans:

Share this post


Link to post
Share on other sites

Would be fascinated to have a read Amz, I've been following this thread keenly!

Share this post


Link to post
Share on other sites

No worries,

Just send me a PM with your email address, or should I add it here?

I don't know the protocol for this!

Thanks

Amz

Share this post


Link to post
Share on other sites

^ It's your work Amz, you can put it up if you want. Just attach it to a post in here, and only members can get it that way ;)

Share this post


Link to post
Share on other sites

I'll definitely give it a read... I will no doubt come up with some one in a million exception to the rule in your conclusion to make you second guess for ten minutes then realise im talking #$%@  :;)

Share this post


Link to post
Share on other sites

Hey,

Well here is my essay and the graphs and survey that I did for the essay

Please be kind as I am not a great writer, but I did pass the course!

Let me know what you all think

Thanks

Amz :cans:

Share this post


Link to post
Share on other sites

i just an ultralight and can say for a fact this is the first time ive heard a difference between wavs and 320kpbs mp3s.. oh any my 190kpbs mp3 sound HORRIBLE. gesus. ignorance is bliss. have shit gear and you never have to worry about this stuff

also we just got a HD tv and its the same thing all over!! i was watching sympathy for lady vengence on our crudy 70's tube and it looked all nice and gothic, dark, grimey. they swapped tv's mid movie and it looked like it was shot entirely on HD handicams. it was horrid. like every single person was green screened. grr. too much focus perfection, information overload!!!

i think i like analogue and warm over this ultra clear world we are walking into

Share this post


Link to post
Share on other sites

Don't you mean "The average person doesn't really care about the difference"?

Also I tried this with my boyfriend, 1. he didnt care and 2. he said it all sounded the same??? It so dosen't but if you call a tradesman "the average person" then he as an example, dosn't hear the difference nor does he care.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now
Sign in to follow this  
Followers 0