[vorbis] Ogg Vorbis v1.1 suggestion

[Carsten Haese]

On Tue, 2003-11-25 at 16:37, Robert Jessop wrote:
> ABR mode works by making several versions of each packet and then choosing the 
> one that gives the best quality/bitrate compromise over a number of packets.
> The number is versions of each packet it encodes is defined in 
> codec_internal.h
> #define PACKET_BLOBS 15
> 
> If you reduce it and recompile ABR mode should be quicker, though if you go 
> too low quality will be really bad. (I haven't tested this)
> 
> Robert Jessop

As ugly as this suggestion sounds (as I explained in my previous reply),
it's actually worth giving a try as long as you know what you're doing
and know that you're making a very specialized version that shouldn't be
used for anything other than streaming ultralow bitrates.

So, with this big disclaimer up front, for what it's worth, here are the
results of my experiment.

With PACKETBLOBS set to 7, oggenc --resample 32000 -b 50 -M 36 seems to
produce quite solid results (as much as anything at 32kbps can be called
solid :) at about 4x real time, compared to about 2.7x that it takes
without the packetblobs hack. That's of course not as fast as the 6x
that my computer can do at pure VBR, but it's a definite improvement.

At PACKETBLOBS=5, the bitrate manager's control over the constraints
began to slip, and the encoding speed only went up to 4.5x. With
PACKETBLOBS set to 3, the bitrate manager became totally useless and was
unable to fulfill the requested -M constraints.

Hope this helps,

Carsten.

<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-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.