[vorbis-dev] vorbis-utils features ;)

[Kevin Marks]

At 2:22 am -0500 22/3/01, Keith Wright wrote:
>  > From: Kevin Marks <kmarks at apple.com>
>>
>>  The quantisers used in the frequency domain obviously have a big role
>>  to play here, but making pre-filtering a necessary separate stage is
>>  going to disappoint.
>
>Disappoint who?  A programmer is not going to be upset to find
>that when ey presses the GO button, it runs a script that does
>
>    getaudio | filter -20db | oggenc
>
>instead of
>
>    getaudio | oggenc -filter "20db"
>
>and nobody else will even know the difference if it is installed
>for em along with the rest of the programs.

I'm speaking from experience here. It will disappoint the people who 
want encoding to be simple and transparent, not some specialised 
black art the way vidoe compression is at the moment.

>  > There is a huge industry of hardware products that do this kind of
>>  pre-filtering for MPEG video.
>
>Which seems to be proof that nobody is too disappointed.  What
>you are asking is to have one pre-filter chosen at random
>and compiled into the basic codec.

No, I'm asking for intelligent filtering within the codec so that 
endless pre-filtering isn't necessary. It sounds like this is more 
viable with Vorbis than with some other codecs.

The only input parameter that most users want to vary is the target 
bitrate. The codec needs to make the tradeoffs internally to hit this 
rate. Gregory's comments on this make a lot of sense.

At 1:12 am -0600 22/3/01, ben at slumber.dhs.org wrote:
>Perhaps a solution would be to run a noise gate on audio with "noisy
>silence" before it is encoded, if the person encoding it wants absolute
>silence in those places (as you said, it's subjective).

Its not that I want absolute silence; I just don't want bits wasted 
on encoding the people coughing the audience and not having enough 
left for the tutti chords that come afterwards. My (possibly 
incorrect) understanding of the plans to hit target bitrates for 
vorbis is to truncate the variable length packets so that they are a 
more uniform length. For a given bitrate there is always an 
integration window where you can collect multiple packets together to 
construct the average bitrate; the trick is getting the encoder to 
make the right decision here. It is a second order problem, but 
competing codecs are already doing this.

>But most music I've heard doesn't have enough silence in it for this to
>be that big of an issue..

Then you need to broaden your musical horizons and listen to more 
classical music, or unaccompanied voice.

How about speech? Should vorbis be usable for (say) the 6 hours of 
Harry Potter I have on a tape currently, and the Goon shows and Monty 
Python records I have seen on the net in MP3 currently or is it 
targeted primarily at music?

--- >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.