[vorbis] never encodes same twice?

Monty xiphmont at xiph.org
Mon Jan 7 12:23:32 PST 2002



On Mon, Jan 07, 2002 at 02:57:17PM -0500, Karol Pietrzak wrote:
> Hello.
> As the subject of my email states, I've noticed that oggenc 
> never encodes the same twice.  i.e., encoding the same file 
> twice, using the same settings and same OS, never results in 
> identical files (checked using 'diff').  I know that a randomly 
> generated serial number is put it, but it gets better: a diff (-
> a -u0)  between these two encodings is ~120KiB.  Umm... why?  

diff only works properly on ascii text.  Use cmp for binary files.

> Should the diff be around 1KiB, at the most?

No, see above.  

> Also, I've compiled by own libao, libogg, etc. (in Linux 2.4.17) 
> and one of the files I've encoded using these utils has a 
> bitrate 2 bits lower than the same track encoded in Windows 95.  

MSVC will produce very different floating point optimizations from
GCC.  The differences are harmless, but enough roudoff behavior will
change to slightly alter the exact output.  It's harmless.

(Note though that there is a win32 bug when using -b in batch mode.
This is causing the incorrect bitrate after the first file encoded,
and appears right now to be a bug in MSVC)

Monty

--- >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