[vorbis] [new?] Streaming technique

[Wilson]

The way I understand it is that only a process that has the original WAV
file to examine can choose the optimum output at a given bitrate. You can
get close with "bitrate peeling", but it's always going to be an
approximation, because there's no way to reconstruct the original data set.
Actually, I don't see why a similar tool couldn't be developed for MP3..
There must be something basic that I'm missing.

--Wilson.

----- Original Message -----
From: "Joerg Bergmann" <bergmann at rcs1.urz.tu-dresden.de>
To: <vorbis at xiph.org>
Sent: Tuesday, June 05, 2001 12:42 PM
Subject: Re: [vorbis] [new?] Streaming technique

> On Tue, 5 Jun 2001 09:24:33 -0500, Aaron Plattner wrote:
>
> >Assuming bitrate peeling works correctly, what's preventing
> >you from just taking a 128kbps stream and splitting it into
> >a 64kbps sub-stream and a 64kbps "extra data" stream?  The
> >second stream won't be a valid Vorbis audio file, but when
> >combined with the 64kbps sub-stream, it'd be the original
> >128kbps stream again.  Right?  Am I misunderstanding
> >something?
> >
> >Aaron Plattner
> >
> As I assume, bitrate peeling itself is a nonlinear thing, too.
> It will do a kind of reconstruction the frequency distribution and
> (nonlinear) scaling down to a more restricted data set. This is
> not the optimum set (quality will be somewhat lower than encoding
> the original data with lower bitrate). But, after doing so, there
> is no _remaining_data_. Due to the nonlinearity, the peeled data
> is no subset of the original data. I'm not a ogg insider, but this
> is my experience with strongly nonlinear compression of XRPD data
> sets (a discipline of crystallography), where I'm programming
> for more than a dozen years.
>
> J"org Bergmann
> email at jbergmann.de
>
>
> --- >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.
>

--- >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.