[vorbis-dev] oggenc questions

[Stan Seibert]

On Sun, 2003-07-27 at 15:58, Dominik Kuhlen wrote:
> 1 why is it called oggenc? (vorbisenc would make more sense)

No idea.  oggenc was shorter probably.  :)  We've talked about this
issue, but haven't reached a consensus beyond "leave it unless we have a
compelling reason to switch".

> 2 Is it true, that oggenc uses only some predefined codebooks (depending on -q)?

Yes, they are predefined.  (Keep in mind most of the answers here apply
to the reference vorbis encoding libraries, which nearly every tool
uses.)

> 2.1 Are they just "random" books, or were they "optimized" in any way

They are optimized, though this is somewhat of a black art.

> 2.2 Is it possible, that some codebooks are stored in the header, but are never used (even in long (>3min) files)?

I don't think so, but I don't know the answer for this one for sure.

<p>> 4 Are there any docs about the psychoacoustic model?

The specs here:

http://www.xiph.org/ogg/vorbis/docs.html

mainly regard decoding.  I don't know of any precise documentation of
the psych model used.

> 5 Does it compress using two passes (whole file)?

No.

> 5.1 Is it necessary to use twopass to do "real" VBR (1. pass: estimate BR; 2.pass: compress)?
>  (ups, this is a general question about VBR)
> 5.2 Same applies to windowsizes: It has to check the file before choosing optimal windowsizes

Because of the previous answer, these are also no.  The standard
encoding libraries do single pass VBR.


---
Stan Seibert

<p><p>--- >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-dev-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.