xml-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stefano Mazzocchi <stef...@apache.org>
Subject Re: AW: AW: Reverse XSL:FO
Date Tue, 25 Apr 2000 21:55:38 GMT
Henning von Bargen wrote:
> 
> > -----Urspr√ľngliche Nachricht-----
> > Von:  Stefano Mazzocchi [SMTP:stefano@apache.org]
> > Gesendet am:  Donnerstag, 20. April 2000 23:05
> > An:   general@xml.apache.org
> > Betreff:      Re: AW: Reverse XSL:FO
> >
> > Henning von Bargen wrote:
> > >
> > > There's a tool called ps2edit (an extension to ghostscript) that
> promises
> > > to convert files from PS or PDF to SVG, but I didn't test it.
> > > >From SVG, why should you then want to use XSL:FO ?
> > > SVG is a XML format in itself and you can use XSL to do whatever you
> want.
> >
> > Like, for example, do neural OCR on the SVG vectors to get FO text using
> > XSLT stylesheets? good luck :)
> 
> Why would this be necessary? SVG stores text elements as text, see some
> examples
> at http://www.adobe.com/web/features/svg/samples.html
> for example this one (etchedGlass.svg), which displays the letters SVG with
> a lot of
> effects added (notice the <text> element at the bottom of the source):

Wrong. SVG "could" store text elements as text. Just like PDF could but
this is not "required" to be done.

We are talking generally and if you read what I wrote "do neural OCR on
the SVG _vectors_". I know that extracting text from SVG is dead simple
with XSLT, but how do you know your vectors are not composing your text?

> > Converting PS to FO is not different from converting JPG to MathML, or
> > WAV to VoxML.
> >
> > XSLT is great but not magic :)
> >
> > --
> > Stefano Mazzocchi      One must still have chaos in oneself to be
> 
> Hey Stefano, you're right!
> It's impossible (or at least, very very very difficult) to do such things
> with pure XSLT or FO.
> But theoretically, one could use an external program to do voice2text
> recognition and
> let it generate a text file, this could be included via uri:include ;-)

Totally. In fact, I'm currently using Cocoon as a VoxML engine using
Motorola's latest mobile software development kit: the browser does
voice recognition and generates http requests to Cocoon that transforms
the pages into VoxML marked-up pages.

So I'm simulating a phone conversation with Cocoon on my laptop.
Impressive for demos :)

And this works very well if you have a short list of choices.. something
like "would you like to go to *weather* or *stocks*"... and even the
most stupid speech recognition engine is able to recognize the
difference between "weather" and "stocks" and get it right 99% of the
time even with background noise.

But still, this is _far_ from what we were talking about, like
simulating a full form upload using speech recognition sent marked-up to
the server... and what for? browsers are already great at doing their
job :)

-- 
Stefano Mazzocchi      One must still have chaos in oneself to be
                          able to give birth to a dancing star.
<stefano@apache.org>                             Friedrich Nietzsche
--------------------------------------------------------------------
 Missed us in Orlando? Make it up with ApacheCON Europe in London!
------------------------- http://ApacheCon.Com ---------------------



Mime
View raw message