Vorbis - Jun 2003 - Vorbis decoding from scratch and the old MP3->.OGG story (new thread, was: Please confirm your message)

Hi,

<p>> I have a question for you.  If I want to write a vorbis client
from scratch,
> should it attempt to read the input file from start to finish, or should it
> attempt to start at the end first
> in order to use statistical analysis to optimize the conversion process?  I
have
> been playing
> around with oggenc and ogg123 and enjoying the way my music sounds, but I
can't
> help thinking about how nice it would be to have a portable device that
plays my
> ogg vorbis files.  :)
Uh, well .. getting Vorbis on a portable, non-laptop/-PDA player is a
different matter, I guess. It should be happening not so long from now
on the Neuros Audio Computer, though.

If you want to look at something to start with, check out the Tremor
sources from CVS - this is an integerized decoder, optimized for devices
with low memory and low CPU power.
> In other words, should I "top-post" my input file to the
algorithm, or should the
> algorithm
> "bottom-post" it in order to optimize the decompression
algorithm?
I'm not a developer, but I know that much: To start decoding a Vorbis
stream, you don't need to know about the end of the stream (only to
calculate some statistics, like stream length and such). All the
necessary information to start decoding is in the three header packets,
that have to be read and processed in order first. If this wasn't the
case, streaming Vorbis would be troublesome, if not impossible. Your
questions should be answered completely in

http://www.xiph.org/ogg/vorbis/doc/Vorbis_I_spec.html

If something in the specs is very unclear or even wrong, documentation
bug reports to http://bugs.xiph.org/ are very appreciated. (Vorbis Core
-> Documentation) People implementing Vorbis from the specs only are
ideal to iron out those things and make Vorbis better!
> Also, does anyone have a recommendation for an existing *nix client for
> "streaming ogg-vorbis" files other than
> xmms, which I like? 
ogg123 is able to play .ogg streams coming via http on the command line,
for example.
> I just used oggasm to convert my entire .MP3 collection to ogg-vorbis.
I hope you are aware that by doing so, you created suboptimal .ogg files
- they won't sound better than your MP3s, maybe even worse. Vorbis can't
magically recreate the audio information that got lost during the
encoding to MP3. Vorbis will even waste bits on trying hard to sound as
close to the source MP3 as possible, with all possible problems, quirks,
blurbs, swirls, etc.

Good advice would be to leave the MP3s as they are and re-rip your music
to Ogg Vorbis, i.e. "simply don't make MP3s anymore".

<p>Moritz
--- >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@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

gtgbr@gmx.net wrote:
> > I just used oggasm to convert my entire .MP3 collection to ogg-vorbis.
>
> I hope you are aware that by doing so, you created suboptimal .ogg files
> - they won't sound better than your MP3s, maybe even worse. Vorbis
can't
> magically recreate the audio information that got lost during the
> encoding to MP3. Vorbis will even waste bits on trying hard to sound as
> close to the source MP3 as possible, with all possible problems, quirks,
> blurbs, swirls, etc.
>
> Good advice would be to leave the MP3s as they are and re-rip your music
> to Ogg Vorbis, i.e. "simply don't make MP3s anymore".
Although I was aware of this, my informal audio tests caused me to be quite
impressed
with
the results from oggenc at default (~128k) settings.  Maybe it's just
psychological,
but many of my homemade .ogg
files sounded better to me using ogg123 to play them than the original .MP3
files
did.  In other words, I liked the sound of
the Ogg Vorbis files.  So the margin of error, i.e. the lost information, was
not
critical to me.

Although I am a professional musician, I am not an audio purist.  My current
equipment
does have limitations, although it
sounds good to me.  But I was literally amazed at my perception of the quality
per
bandwidth ratio I subjectively experienced with
ogg123 and xmms.   Many of my original .MP3 files were encoded at 384k because I
found
128k .MP3's to be somewhat limiting.
Yet when I listen to those files after they have been converted to .ogg format,
they
sound just great to me even at ~128k. I noticed
quite a bit of variation in the bitrate of the output stream, and found that I
was
able to enjoy even the converted .MP3's that were
originally encoded at only 128k.  So my conclusion was that in my case, it is a
moot
point.  I didn't delete my .MP3's , but I generated
a full tree of .ogg files from the .MP3 source tree and was actually surprised
to be
satisfied with the result.

Thanks for your informative response to my initial post.

<p>Regards,

Glendon Gross
>
>
> Moritz
--- >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@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.