Re: [g-a-devel] Bug with festival-synthesis-driver

[Aditya Pandey <aditya kumar pandey gmail com>]

On 10/11/05, Bill Haneman <Bill Haneman sun com> wrote:

Aditya:

If you can find out what the festival-synthesis-driver process is doing
(via strace, perhaps) and what "festival --server" is doing (also via
strace) when this takes place, it might help diagnose.

I would try finding out the reason. I have not given up on this yet.
 

Can you explain what calls you are using to festival-synthesis-driver to

read the document?  I would hope that you are not sending it the whole
document in one 'say' call...

I am not sending the whole document but a few words at a time at the rate decided by "rate" parameter of voice. If there is callback available (for my festival driver, there is none), then I only send text when the last text sent has been consumed/spoken. It might be a problem with buffer size in festival/festival-synthesis-driver. May be there is hard coding of max buffer somewhere, or something gets freed/overwritten.
 

> No one has reported this behavior before as far as I am aware.

Might happen as gnome-speech is generally being used through gnopernicus, I suspect nobody would "stress" tested it continuously for 2-3 hours with making it read a large document at a high voice rate of 800.

Aditya