[vorbis] OGG Versus MP3

vorbis at papaya.altamente.com vorbis at papaya.altamente.com
Thu Mar 18 11:33:44 PST 2004



>> www.digit-life.com/articles/oggvslame/

Every time I do a search for ogg vs. lame I get this article, which I find
VERY lacking.  First, he relies too much on visualization of the music. 
Music is to be listened to.  All the damn cool edit spectral analysis in
the world will not tell you what it will sound like.  Hell, I'm going to
create a new music codec that will be based on jpeg.  I'll compress the
spectral snapshot of each frame with jpeg.  How do you think that will
sound?  It'll look really good though.

Once again, you will need to listen to tell the difference.  For myself, I
am quite happy with Garf's Tuned Beta 1 or 2 at quality 7, although, I
still think oggvorbis beats everything at its default ~112, which rocks
and I'd compare to lame ~160 or perhaps a tad better.

As you go up in quality, I don't believe oggvorbis scales so well for
higher bitrates, that is, the improvement with respect to increase in
bitrate tapers off above q 5.  Enter Garf's tuned encoder for higher
bitrates.  I like it, it sounds great to me.  For classical music, I hate
mp3's joint stereo business.  Downside of GTbeta1 or 2?... Garf's a bit
grouchy *waves*.

However, I agree with the previous message that there is becoming less
difference between lame --preset extreme and oggenc(GTBeta2) than there
was before, quality-wise especially with lame 3.95.

I just encoded a particular song with lame --preset extreme and oggenc -q
7 (GTBeta2) and although I can STILL detect a slight difference between
them, it's not much.  The lame one still suffers from that mushiness
(pre-echo) to which I am particularly sensitive ( I guess).  The piece in
question is dance/europop/remix stuff, so stereo separation isn't a
problem with either encoder.  The oggvorbis sample just sounds sharper,
and drum beats are cleaner and snappier, more in line with the original. 
Could be my imagination though.  I'll have to ABX the samples to be sure.

Anyway, the difference isn't much.  The lame --preset extreme is 10%
bigger than oggenc(GTBeta2) -q 7, but IMHO sounds just a tad worse.

I put up the samples on a bittorrent just in case anyone is interested. 
Why bittorrent?  I'm playing around with it *G*.

http://celine.homelinux.com:6882/

--James

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.



More information about the Vorbis mailing list