Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Ogg killer track (Read 11990 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Ogg killer track

This track: 'QUEEN - Another One Bites The Dust' makes the OGG Vorbis 1.0 RC3 codec bite the dust badly even at quality levels as high as 9.

As many of you may know this song starts with a very agressive bass guitar that gets badly distorted.

I can upload the first 30 sec. compressed with Monkey's Audio but I don't know where, if somebody knows of a good free storage service or have the space please let me know.

Ogg killer track

Reply #1 –
Cut the thing down as short as you can... if the difference is noticeable within the first 10 seconds, keep it at that.  Or start the clip a couple seconds before noticeable distortion begins, and end it a few seconds after that.  I typically use cdex to rip segments, specifying, for example, from 6 seconds to 13 seconds of a given track.  This way I keep the wav files small enough to email out to people it they so desire.  Compress it in Monkey's, and maybe zip it after that.  But I'm only saying this because I have a phone line connection.  Don't know where you can post it... there are some sites that have problem clips posted, but I don't know how to get in touch with them.  Hopefully someone who checks this out will know.  Otherwise, email it to people who are interested... reach me at
clevelandrunnerboy at yahoo dot com.  Thanks.
God kills a kitten every time you encode with CBR 320

Ogg killer track

Reply #2 –
I have temporarily enabled anonymous ftp upload at:

ftp.ff123.net

userid = [a href='mailto:anonymous@ff123.net'][/a]
password = any

Ogg killer track

Reply #3 –
Good morning!

Very good, I'll upload it and also I'm going to cut down the wave a little with SoundForge to shorten the upload/download time.

Thanks

Ogg killer track

Reply #4 –
ff123

I uploaded the file to the incoming folder but still can't see it with the LIST command. Please let me know if it's really there, i'll upload it again if it's not

timcupery

ff123 can make it public on his site now

Regards

Ogg killer track

Reply #5 –
ff123, thanks for being willing to help here.  Could you (or gambito) provide a link for download once the clip is posted correctly?
God kills a kitten every time you encode with CBR 320

Ogg killer track

Reply #6 –
sample:  queen - another one bites the dust (9 sec.).pac can be found at:

http://ff123.net/samples.html

ff123

Ogg killer track

Reply #7 –
I tried at q5, but couldn't abx it.
The problem with your sample is that it gets higher in treeble from begining to end, sot it is very difficult for me to remind between two sample (even by switching very fast between A, B and X).
But i must admit that my equipment and ears are very cheap ;-)

Ogg killer track

Reply #8 –
it's very advisable to use earphones, it's dificult to spot the bass distortion with standard pc speakers

Ogg killer track

Reply #9 –
Using RC3, -q5 and my Grados, I can't hear what you're reporting.  More details please?

Ogg killer track

Reply #10 –
Quote
it's very advisable to use earphones, it's dificult to spot the bass distortion with standard pc speakers
Maybe the source of the distortion IS your headphones...

Ogg killer track

Reply #11 –
if the headphones are causing distortion, it will happen on both the original and encoded-decoded wav file, so speaker distortion won't be a source of false positives for ABX-ing.

But seriously, Gambito, I'm surprised that you can tell a difference up to -q 9.  Have you tried ABX-ing it?  There are some clips where I thought I could hear a difference between the original and encoded version, but when I actually test myself with ABX, I can't pick out the difference.  And if Monty can't hear the difference at -q 5... he's got pretty good ears and equipment.

BTW, for Monty and others concerned, I've got a clip of mostly noise... it's from a live recording, the fuzz before any music starts.  Left and right channels slightly diff so it's got some stereo action going on.  I was able to ABX it 15/16 (on my cheap speakers and soundcard) at -q 5.0.  I should note that -q 5.0 was _much_ better than -q 4.9, so that lossless stereo implementation means something for me.  At -q 6.0 I was no longer able to discern any difference.  The factor I focus on to differentiate between the ogg-encoded version and the original is, once again, the brightness which ogg adds.  This file probably isn't much different than whatever other white noise files are floating around.  But just my report on ogg...
God kills a kitten every time you encode with CBR 320

Ogg killer track

Reply #12 –
I wasn't able to spot any distortion in the bass at -q4 using my Grado SR325's either.

ff123

Ogg killer track

Reply #13 –
sorry i was changing room

the distortion i ear (with earphones) was phase shifting like and i was able to ear it up to qual 9 although very reduced, but at qual 5 was very clear, also i don't ear this type of distortion on the original wav, an aac (psytel -normal), a mpc (normal) nor a mp3 (alt-preset standard) only on ogg, also i'm using the latest plugin from pp

tonight i'm going to test it again anyway, you all make me doubt

Ogg killer track

Reply #14 –
gambito-
    tell us about them if you can ABX cleanly.  don't know if you understand ABX; I didn't when I started this forum.  But the double-blind test structure forces you to differentiate between the files, not just think that you're able to tell a difference.  You can get software and info at http://www.pcabx.com
God kills a kitten every time you encode with CBR 320

Ogg killer track

Reply #15 –
i repeated the test at qual 5 and i was able to ear the distortion again and very clearly, now i'm pretty sure about it.

also tested psytel 2.01 -normal, mppenc 0.90s --standard and lame 3.91 --alt-preset standard and none of them produced this type of distortion

i really can't describe exactly what i ear, it's phase shifting like but not exactly as in mp3, it's some type of whistle sound that appear originated from the bass guitar

hope you'll be able to spot it also

regards

Ogg killer track

Reply #16 –
I ABXed -q 5 pretty quickly 16/16 with first try and think I hear what gambito means. There's this fluttering/phasing effect in the backround associated with bass, but it's not the bass base-frequency fluttering, it's a bit higher, but no high-freqs.

I hear it especially well at around about 3-3.5s and 7.5-8s.
Juha Laaksonheimo

Ogg killer track

Reply #17 –
Quote
i repeated the test at qual 5 and i was able to ear the distortion again and very clearly, now i'm pretty sure about it.

To get the developers to investigate your report seriously, you need to demonstrate that you really *can* hear this artifact (and I'm not saying you can't). The easiest way is to use an ABX utility (such as that at http://www.pcabx.com ), and report the score you get. If it's significant, then we can all be sure that there actually is something worth investigating.

Ogg killer track

Reply #18 –
Uhm, I ABXed 16/16 and I can hear distortion which fits the description. It's definitely there.
Juha Laaksonheimo

Ogg killer track

Reply #19 –
I just ABXed -q 9 19/20.
This was also with first try, but certainly wasn't very easy. Concentrated almost only on phasing near 3.5s.
Average bitrate is 278kbps.
Juha Laaksonheimo

Ogg killer track

Reply #20 –
thanks god! i'm was not crazy!

i spoted this distortion from beta 4 (was worse) and i thought that developers would fix it soon but i was surprised that it survived until today

> To get the developers to investigate your report seriously, you > need to demonstrate that you really *can* hear this artifact
> jon ingram,

you're right jon, that's b/c i downloaded abx 1.75 and tomorrow i will post my results

regards

Ogg killer track

Reply #21 –
This "phasing" seems to be mostly connected to the dropouts visible in spectral view.

Also if one can't still hear this, create about 350Hz highpass and filter the decoded clip, so that the bass sound isn't masking too much here.

Now it should be very easy to hear the distortion from about 2.2s to 8s.

Then when "learned", it should be easy to hear the distortion with normal -q 5 encoded clip.
Juha Laaksonheimo

Ogg killer track

Reply #22 –
Quote
Uhm, I ABXed 16/16 and I can hear distortion which fits the description. It's definitely there.

Did I say it wasn't there? No. Did I say I *thought* it wasn't there? No. Did I give the *impression* that I thought it wasn't there? Perhaps, but if so it wasn't intentional.

There have been a lot of people saying 'this sounds bad' recently, without demonstrating that they can actually hear a difference. I was just explaining that one of the easiest ways of objectively demonstrating that a difference exists is to try and ABX a sample (as you have subsequently done). BTW: quality 9? You must have the ears of a god!

Ogg killer track

Reply #23 –
test done with abx at qual 5, scored 30/30, next test at qual 9
we'll see...

Ogg killer track

Reply #24 –
test done at qual 9, scored 30/30 also!
very surprising even for me!

can be this a bug on the psychoacoustics model?

experts have the word

regards