proof that ogg isn't all it's cracked up to be?

[xeo]

I encoded some music by a metal band called helloween into 128k ogg vorbis format using media jukebox 8, and compared it to an mp3 version i downloaded which was also at 128k, and if I'm not mistaken the mp3 version sounded better!

Now don't get me wrong, i am all for the new codec, and i use it with my divx encodes, but is it only good when encoding at low bitrates like 80k, (1 quality) or can it still beat mp3 at 128k?

[canadian_fbi]

well it's entirely possible that to your ears, mp3 sounds better than ogg vorbis (at least for that one sample). or perhaps that was a pretty tough sample that vorbis might inherently have trouble with. keep in mind that there's only so much you can do with 128 kbps, and so audio quality is going to have to be compromised in some way, and it's just a matter of which way sounds best. so unfortunately it's a very subjective sort of competition, and one codec can't necessarily "beat" another all the time and for all listeners. certain varieties of mp3 can be pretty good, and vorbis isn't perfect by any means, but i think that ogg vorbis generally sounds better than mp3 at most bitrates and with most samples, and i think most people do too.

[spyder]

It's also possible that your Ogg sounded bad because maybe it was a re-encoded MP3.

[FDISK]

Here's the proper way to test: Rip one track from a CD. Encode using the LAME MP3 Encoder, preferably with the command line LAME -q 0 -d -b 128 infile.wav outfile.mp3. Then, encode using Vorbis RC3 (www.vorbis.com), manipulating the min and max bitrate values to get 128kbps CBR encoding, then compare the two files. Which do you prefer?

Personally, I find that LAME MP3 works much better for songs and such where I demand high quality, and Vorbis works better for stuff like movie sound tracks where small space is preferable. I've found that you can get EXTREMELY good quality audio in around 90kbps with vorbis.

[Neo Neko]

Yes always start from a clean uncompressed or lossless source and use the most current version of everything avalible to you. Vorbis(Not OGG) has it's problems as does MP3. They do things in different ways. They have different strengths and weaknesses. Does Vorbis always beat out MP3 at this point... No not at this point. There are isolated samples that give Vorbis some definate problems. But as a whole Vorbis does outperform MP3 in a vast proportion of instances. And if you are comparring a Vorbis encode made from an MP3 encode the original MP3 will win every time. Likewise if you compare an MP3 encoded from a Vorbis file then there is no way that the MP3 could compare. Transcoding is big no no #1.

[Koepi]

You might add that the psycho-acoustic model in OGG Vorbis is different to the one in mp3.

In mp3 the model is adopted that way that you have a biased signal after compression - you could best describe the effect with "loudness", that switch on your stereo. turn it off, and turn it back on.

Vorbis is targeting the most _identical_ and original signal, thus having no loudness effect (switch it off and listen to a cd.).
MP3 is for techno-kids mostly (sorry pals), thus biasing and disturbing the signal towards a _very_ noticable loudness effect (turn loudness on, lower the volume (!!!) and listenb to the cd again).

You'd propably find the loudness signal better (it's more charming for our ears). But it's not the original signal anymore.

[Nic]

Ogg (not vorbis ;) j/k ...whatever Garf calls it, I call it :)
(your right Neo it should be known as Vorbis, but few refer to it as such...the progs for example:- OggDrop, OggEnc, etc)

Ogg Vorbis RC3 does work alot better at quality settings than at abr ones....Vorbis _should_ beat MP3 at 128kbits in most cases, to most ears. You can normally trust www.ff123.net if you don't have time to test yourself.

Cheers,
-Nic

[Sven Bent]

Originally posted by FDISK
Here's the proper way to test: Rip one track from a CD. Encode using the LAME MP3 Encoder, preferably with the command line LAME -q 0 -d -b 128 infile.wav outfile.mp3. Then, encode using Vorbis RC3 (www.vorbis.com), manipulating the min and max bitrate values to get 128kbps CBR encoding, then compare the two files. Which do you prefer?

thsi is a total pointless test.
why handicap both encoder ?

if you wan to test it use the -alt-preset which gives way better quality then yours command line for mp3 encoding by lame.

[Koepi]

I'd say use a post-rc3 binary to test vorbis.

You should find such encoders at the "rare warez" page, http://www.inf.ufpr.br/~rja00/


Have fun,
Koepi

[FDISK]

Sven Bent: You aren't handicapping anything, you're leveling the playing field. You're making sure that both encoders have exactly the same number of bits to work with, and seeing how well they work with the space they have.

[tenebrenz]

@FDISK, I kind of agree with Sven here. Both Lame MP3 and Ogg get a lot of their power by using VBR, making both CBR doesn't allow either to produce their best and may give one an unfair advantage. I think it would be much fairer to encode in a VBR MP3 mode, then encode the same file in OGG and adjust the quality control until both files were very close in size. That way you should be able to tell which one sounds best for a particular average bit rate.

[gnoshi]

I love the way you use the word 'proof' when the effect being described is subjective anyway.
It has been said 100mil+ times before, but just because you 'like' one format better than another doesn't necessarily say a damned thing about the quality of the respective formats.
You want to prove one format is more transparent (less different perhaps is a better way to say that) is to ABX it. easy.
You want to prove that one is 'better' than another in any other way; good luck. Opinion != proof.

gnoshi

btw. fiddling with the -q of vorbis to match a given bitrate of a vbr encode with lame (or reversed) is the only real way to test with any fairness (unless you are using abr in your everyday encodes, in which case I say... FREAK!)

[MaTTeR]

Moving the thread to the audio forum where Vorbis is discussed. Thread is mistitled somewhat.

I have to agree with gnoshi here. If you want to compare codecs properly then ABX them with around 16 different samples. As has been said many times before, each codec has it's strength and it's weakness. It just so happens that in my opinion Vorbis works much better for 1CD rips. If it's a 2CD rip then I'll obviously keep the AC3/DD.

Edit- I think it's about time we add the definitions of Ogg and then Vorbis to the FAQ. The terms are thrown around way too loosly for my taste, so perhaps the FAQ will clarify.

[Sven Bent]

Originally posted by FDISK
Sven Bent: You aren't handicapping anything, you're leveling the playing field. You're making sure that both encoders have exactly the same number of bits to work with, and seeing how well they work with the space they have.

no you are not elvleing they playing filed at all.
you are just testingn in an artificial situation.

cbr might hurt og more then mp3 who know ???
and since none if these cbr will be used you are just teting something totaly wrong.

evne if X came up on top in cbr
it culde easyli be Y that better when using the encoding you woulde REALLY use.

besides testing only on one track is uselss

just use spahm or fatboyt as vorbis sound will TOTALY scred them up

this is agian an artificial test whcih does NOT show the average quality of the two encoders.


actuall i have brogut up the spahm/fatboy weak spot of vorbis sound up at HA and doom9 before.

[ookzDVD]

@xeo,

I prefer to stay with mp3 for music encoding,
lame -alt standard is still my fav. for music encoding,
and for trading.... mp3 is more acceptable. ;)

and stay with ogg for movie soundtrack, small size but not bad! ;)
quality setting 2.0 is enough for me for DVD to 1 CD ripping,
ogg about ~80mb, so the movie will be 620mb for 700mb cd, nice!!

[dado023]

mp3(lame) in general somehow doesnt compreses corect because time length of the song(in this scenario) isnt the same when you decompress it back to wav, it is few ms(milisecond) deviation like the file size(few kilobytes)..and it amplifiyes the signal so you get more cliped samples

but when you decompress ogg back to wav time length is identical and file size( to the bit) is corect like the original


if you want some real proof do the anylse test in cool edit 2000, there you can see the number of the cliped samples,frequency cutoff...etc