[vorbis] Ogg Vorbis files can be compressed ?!?

Nescafe nesca at wanadoo.fr
Sat Jan 18 08:26:24 PST 2003



> >   136,140 sample.ogg
> >   110,284 sample.zip
> ~25 kB saved.
> >   3,243,113 sample.ogg
> >   3,223,435 sample.zip        (0.6% savings vs ogg)
> ~19 kB saved.
> Pretty close.  Coincidence?  Probably not. :)

I searched a bit, and found that page :
http://homepage.ntlworld.com/jfe1205/
It has a link to this :

oggenc2.2: Command line encoder - allows direct input of '.flac', '.ape',
'.wv', '.pac', '.ofr' and '.shn' files + 'rehuff' option - 2003-01-01 - ICL6
compile - 245Kb - 1.0 - by John33

http://homepage.ntlworld.com/jfe1205/oggenc2.2.zip

Which seems a nice version of oggenc
(c) 2000-2002 Michael Smith <msmith at labyrinth.net.au>
& portions by John Edwards

So, I've been using this one, and it produced a valid
rehuffed stream, just about the same size as my failing
rehuff compiled version. The problem remains that
the 126'755 bytes file compresses to 98'694.

But how would you explain an mp3 for the same
precise sample being not any further compressible
with either ZIP or RAR ?

Is MP3 coding (not transforms, but final coeff coding)
so different that it adds way too much entropy for
the lossless algorithms to deal with ? In this case,
why would OV let a part of the job to the archivers ?

JPGs don't store raw selected DCT coefficients.
These coefficients certainly might be even further
compressed if the final user applied a LZ based
algorithm instead of the inlined Huffman codec...
But this is part of the JPG codec.

My questions aren't any technically speaking.
I am just wondering how a "good" codec should
behave with rare samples.

This could even be taken as a good point for
Vorbis, depending on your point of view.
Let me use a picture :
Which is better ? The place cruising at mach 2,
ot the one cruising at 1.5 and capable of 2.5
with Post-Combustion activated ?

My final questions is :
What is to blame for this sample ?
Redundancy, or maybe window size, or what ever ?
Shouldn't Vorbis take this redundancy into account ?
Or maybe is this a planned Vorbis extension...

Nescafe
- In fact, don't feel yourself blamed for anything, I
really do like Vorbis AND this point puzzles me,
so, the passion is back ;-)
- Should I cross post to vorbis-dev ?

--- >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