Text To Speech in Read Etexts improved again

[James Simmons]

Hemant,

Does that mean you are making changes in the speech-dispatcher API 
itself?  That could complicate things, since I'd need your version to be 
on my test boxes.  Nothing I can't do, but that will make it more 
difficult for Ed to do his demos in Ubuntu.  It would have been nice to 
avoid threading when I wrote the TTS code to begin with, but now that 
the code is in there and working fine I don't see it as a problem so 
much.  I really need espeak to do a better job on the callbacks.  Or to 
replace espeak with Festival, which Hynek said worked better.

I already sent you my non-Sugarized version of Read Etexts.  That's what 
I would modify first to try out the new API, since that's much easier to 
test than the full version.  You could hack on that one yourself.  It 
might save you some time waiting on me.  Then I could modify the full 
Activity and test it.

Thanks,

James Simmons


Hemant Goyal wrote:

> Hi James,
>
> Absolutely not a problem at all. I am also working on finding a fix 
> for the problems that developers can have while using speechd api with 
> pyGTK as described here : 
> http://www.mail-archive.com/pygtk@daa.com.au/msg15868.html
>
> I discussed a solution to the problem with tomeu and erikos, and have 
> been making changes in the API implementation to make speechd support 
> asynchronous socket communication so that we can avoid  a thread to 
> perform network communication. We've figured out an approach to avoid 
> threads in the API, and hopefully I'll have the alternate solution 
> ready in 2-3 days. I'll be depending upon you to use the modified API 
> in your application for testing purposes :)
>
> Best,
> Hemant