Flatland transcribing (was Progress update)

[robert b <http://dummy.us.eu.org/robert>]

--- "Jessica P. Hekman" <http://www.arborius.net/~jphekman> wrote:
> * I generated some new transcriptions. Get them at
> 
>   www.arborius.net/~jphekman/sphinx/texts.tgz
> 
> That uncompresses into a directory with bin/ (containing some useful 
> scripts) and flatland/ (containing text in transcription format, and a 
> fileids file). Yes, the text is Abbott's Flatland.
> 
> * To make wav files, run
> 
>   ../bin/record-wavs.pl flat1.transcription
> 
> It will print a line. Hit <enter> when you are ready to dictate the line. 
> Dictate the line. Hit control-c. It will present you with another line but 
> will not record until you hit <enter> again, so you will get into this 
> <enter><control c><enter><control c> pattern. I'd rather it just be a 
> series of <enter>s, but ecasound will record until it gets control-c. If 
> you feel like fixing the script, let me know.
> 
> * I am working on training the results of the above now. I notice that the 
> generated time.dic has this:
> 
> SIL                 S IH L 
> 
> That's not really what we want, is it? It should know that SIL means 
> silence. So that's something to think about.
> 
> j
> 

> 
> -------------------------------------------------------
> This SF.net email is sponsored by: ApacheCon, November 18-21 in
> Las Vegas (supported by COMDEX), the only Apache event to be
> fully supported by the ASF. http://www.apachecon.com