[vorbis] Technical MP3->OGG transcoding question

[Seth de l'Isle]

On Thu, Dec 05, 2002 at 03:22:11AM +0100, gtgbr at gmx.net wrote:
> TobiX wrote:
> > CAN it be done in a better way than Poor Man's Transcoding?

It sounds from the responses we've seen so far, that this is not an easy or
reasonable project.  However, I think that the question boils down to: 

  "Is there information, relevant to audio that is lost in the MP3 -> WAV
  step."

For example, if we know that a sample contains only voice, then we can use a
voice only format to achieve a better quality/compression ratio than if we
didn't know that the sample would contain only voice.

It sounds like the answer is "not much."  However at a theoretical level, and
without knowing much about audio formats, there is definitely _some_
information that is lost in the MP3 -> WAV step.  The obvious lost information
being that the audio was compressed with an MP3 encoder.  The compression level
of the MP3 is another piece of potentially interesting information.

So I'm confident that at least at a silly, pathological level trans-coding could
be improved over MP3 -> WAV -> Vorbis.

> > (i.e.: mpg321 -w - file.mp3 |oggenc -o file.ogg - )
> > CAN it be done in a lossless way?

Lossless means, if I convert my MP3 to Vorbis I can get a bit for bit identical
MP3 back from my Vorbis data.  Since both formats are lossy, information is
thrown away at each encoding step, period. So I'm confident in claiming that
transcoding MP3 -> Vorbis or vice-versa can not be achieved losslessly.
Otherwise, we are all just mistaken in calling MP3 and Vorbis lossy formats.


-- 
pgp public key http://ubertechnique.com/seth/pgp_key.txt ID: 0x60F9B67A
--- >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.