[vorbis] Ogg FLAC

rillian rillian at telus.net
Tue Oct 16 21:36:15 PDT 2001



On Tuesday, October 16, 2001, at 08:59 , Josh Coalson wrote:

> yep.  the format supports up to 32bps, 8 channels,
> and impractically-high sample rate (~1MHz).

Only 8? :-)

>
>> elsewhere Josh wrote:
>>
>>> my overall vision would be to have an -ogg option to
>>> flac that does what I mentioned, and a tool that can
>>> convert between Ogg-FLAC and standalone FLAC, which
>>> would also be very simple.  all of this I could do
>>> pretty quickly if it would be really useful.
>>
>> That sounds reasonable, since you have an estabilished user
>> community.
>> There's the extension-sorting issue to think of as well. It makes
>> more
>> sense to use oggflac when you're multiplexing with other data. OTOH,
>> as
>> jack mentioned you'd get our superior metatdata support. :^)
>
> how does vorbis handle it?  I am assuming .vbs means ogg-
> vorbis and the name for other_transport-vorbis is
> undefined?

Well libvorbis itself spits out raw data packets. (which includes the 
codebook and metadata headers) We either frame those with ogg, in which 
case it's a .ogg, or we packet them in rtp packets. At least those are 
the only two transports so far.

There are two types of metadata we're talking about though. There's 
vorbis-level comment header, and there's the mythical, knows all, 
records all, kitchen-sink metadata format (codenamed 'draycott') which 
would be multiplexed with the other data at the ogg level.

> the details are here (http://flac.sf.net/format.html) but
> basically, in ogg-speak, FLAC has only pages, no segments
> or packets.  pages may be variable length.  the page header
> also contains the basic stream-level data you need to play
> it, like sample rate, bits-per-sample, etc.

Interesting. Sort of chunk-based, but not really? You just made each 
metadata block or frame an ogg packet?

> I could drop the sync code, sample number, the CRCs and
> maybe the blocksize.  but this is maybe 0.2% or less of
> overhead on CD-rate data.

Right. Like with mng, it's mostly an aesthetic issue.

  -r

--- >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