theora dev - Nov 2002 - Multiplexed Ogg and header problems.

I there was a discussion on the vorbis list about Real Audio
  SureStream (tm) format. The basic idea was that they put multiple RA
  files with different compression levels. So when they have to switch
  the bit rate due to bandwidth changes, they choose from the streams
  they have. Since peeling is a long time away, it could be implemented
  with ogg vorbis multiplexing several times the same file with
  different compression levels. My question (and I think Monty can
  answer it) is how should the the presence of multiple concurrent
  vorbis streams be handled. Because if you have multiple audio tracks
  for a movie, or you have multiple compression rates it's very similar.
  The only problem seems to identify what's each of those streams for.
  In other word, how should I know that stream A is Swahili and stream B
  is Speranto, or that stream A is the -q3 compressed and Stream B is
  the -q1 compressed?

  Regards,
  Alejandro Belluscio

--- >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
'theora-dev-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.

good question.  We have this functionality in our On2 server platform
(TrueCast).  It works great for limited/variable bandwidth situations ie
internet at say 150 - 400 kbits/sec.  Would love to see this in OGG.

<p>-----Original Message-----
From:	Alejandro G. Belluscio [mailto:baldusi@uol.com.ar]
Sent:	Fri 11/8/2002 2:13 PM
To:	theora-dev@xiph.org
Cc:	
Subject:	[theora-dev] Multiplexed Ogg and header problems.
  I there was a discussion on the vorbis list about Real Audio
  SureStream (tm) format. The basic idea was that they put multiple RA
  files with different compression levels. So when they have to switch
  the bit rate due to bandwidth changes, they choose from the streams
  they have. Since peeling is a long time away, it could be implemented
  with ogg vorbis multiplexing several times the same file with
  different compression levels. My question (and I think Monty can
  answer it) is how should the the presence of multiple concurrent
  vorbis streams be handled. Because if you have multiple audio tracks
  for a movie, or you have multiple compression rates it's very similar.
  The only problem seems to identify what's each of those streams for.
  In other word, how should I know that stream A is Swahili and stream B
  is Speranto, or that stream A is the -q3 compressed and Stream B is
  the -q1 compressed?

  Regards,
  Alejandro Belluscio

--- >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
'theora-dev-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.

<p><p><p>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: winmail.dat
Type: application/ms-tnef
Size: 3410 bytes
Desc: winmail.dat
Url :
http://lists.xiph.org/pipermail/theora-dev/attachments/20021108/61222a1a/winmail-0001.bin

hi Alejandro --
 
I'm not really an expert in OGG.  Monty will have to make this decision.  I
can contribute knowledge about, and perhaps some code that demonstrates,
multiple stream switching during playback, but it's up to the Xiph guys how
they want to make that work inside their format.
 

        -----Original Message----- 
        From: Alejandro G. Belluscio [mailto:baldusi@uol.com.ar] 
        Sent: Sat 11/9/2002 12:29 PM 
        To: Dan Miller 
        Cc: 
        Subject: Re[2]: [theora-dev] Multiplexed Ogg and header problems.
        
        

        Hello Dan,
          I've been thinking a bit about this. And may be we could come up
          with an XML based definition for this things. Each stream should be
          identified by it's serial number and have a description of
it's
          type. We then encapsulate this in an ogg file and concatenate it
          _before_ the multiplexed file. It's kind of custom header file. Is
          this too much against the idea of ogg encapsulating?
        
        Regards,
        Alejandro Belluscio
        
        Friday, November 08, 2002, 4:33:51 PM, you wrote:
        DM> good question. We have this functionality in our On2 server
        DM> platform (TrueCast). It works great for limited/variable
bandwidth
        DM> situations ie internet at say 150 - 400 kbits/sec. Would love to
        DM> see this in OGG.
        
        DM> -----Original Message-----
        DM> From:   Alejandro G. Belluscio [mailto:baldusi@uol.com.ar]
        DM>   I there was a discussion on the vorbis list about Real Audio
        DM>   SureStream (tm) format. The basic idea was that they put
multiple RA
        DM>   files with different compression levels. So when they have to
switch
        DM>   the bit rate due to bandwidth changes, they choose from the
streams
        DM>   they have. Since peeling is a long time away, it could be
implemented
        DM>   with ogg vorbis multiplexing several times the same file with
        DM>   different compression levels. My question (and I think Monty
can
        DM>   answer it) is how should the the presence of multiple
concurrent
        DM>   vorbis streams be handled. Because if you have multiple audio
tracks
        DM>   for a movie, or you have multiple compression rates it's
very similar.
        DM>   The only problem seems to identify what's each of those
streams for.
        DM>   In other word, how should I know that stream A is Swahili and
stream B
        DM>   is Speranto, or that stream A is the -q3 compressed and Stream
B is
        DM>   the -q1 compressed?
        
        
        --- >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
'theora-dev-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.
        

-------------- next part --------------
A non-text attachment was scrubbed...
Name: winmail.dat
Type: application/ms-tnef
Size: 6507 bytes
Desc: winmail.dat
Url :
http://lists.xiph.org/pipermail/theora-dev/attachments/20021110/506a9e4c/winmail-0001.bin

On Fri, Nov 08, 2002 at 04:13:06PM -0300, Alejandro G. Belluscio
wrote:
>   I there was a discussion on the vorbis list about Real Audio
>   SureStream (tm) format. The basic idea was that they put multiple RA
>   files with different compression levels. So when they have to switch
>   the bit rate due to bandwidth changes, they choose from the streams
>   they have. Since peeling is a long time away, 
It's not that it's a long way away, more that it's tricky and my
personal implementation model requires encoder support to make it work
really well.  I'm not sure that it's substantially harder than
implementing the multiple-stream model, it just requires more
coordination.
>   it could be implemented
>   with ogg vorbis multiplexing several times the same file with
>   different compression levels. My question (and I think Monty can
>   answer it) is how should the the presence of multiple concurrent
>   vorbis streams be handled. Because if you have multiple audio tracks
>   for a movie, or you have multiple compression rates it's very
similar.
>   The only problem seems to identify what's each of those streams for.
>   In other word, how should I know that stream A is Swahili and stream B
>   is Speranto, or that stream A is the -q3 compressed and Stream B is
>   the -q1 compressed?
Well... how specilised an implementation do you want?  The most
flexible solution (in the multi-stream vein) would be to have a
multiplexed file on the server side, and a server that is smart enough
to hide whatever stream was actually in use into a single output
stream and single header that switched back and forth without the
client ever knowing...  Or one could switch streams witht he client
fully aware of different headers, it just decodes whatever it's
getting at the moment....

More than one way to tackle this beast.  The Perl-like mixed-curse.

(Personally, I'd rather code peeling.  Certainly there's been much
more clamor for this feature recently.  It's time to start bumping it
up on my priority list.)

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
'theora-dev-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.