Regarding Speech-Dispatcher, Speechd-up, and ESpeak with Speakup.

[Ralph W. Reid]

I'm just taking a shot in the dark on this one, so don't flame me too
much if I am way off base--I am naked and weaponless...

Are the two systems you mentioned using hard drives with similar cache
sizes?  I _think_ .wav files are created, played, and then deleted
during speech operation.  If this is happening as you are keying in
information, these files will be created and deleted at a manic pace,
and a hard drive with a large cache will never actually spend time
writing these files onto its platters--they will be crated and deleted
entirely in the drive cache.  A drive with a smaller cache will take
more time to write and delete the files because the mechanical
components have to physically move into position to get the job done.
Maybe this extra drive time is the culprit in your case???

The size of the drive cache on my Slackware system can be found in

/proc/ide/ide0/hda/cache

and is listed in KB as 7936 (nearly 8 MB).

I vaguely recall something from long ago about a file system called
'tmpfs' (or something like it) which could be used to set up a /tmp
area in RAM--maybe this could help?

HTH a little, and have a great day.

On Sat, Feb 17, 2007 at 02:24:14PM -0700, Keith Hinton wrote:
> Hi List,
> I'm more convinced then ever now that speechd-up is what is crashing. I have to keep typing "Speechd-up" to restore communication.
> The question is: Why?
> What exactly was that backtrace about?
> Could a chron job be setup to monitor if speechd-up is running, and if it crashes or whatever, instantly reboot it the instant it crashes?
> I need a temporary workaround, short of turning off keyboared echo in Speakup with ESpeak. This, as I have said is on AMD64 machines, spisifically. Speechd-up
> does not crash  on my Pentium box, no matter what I do, doesn't matter how fast I type with keyboared echo on, somehow it's able to keep running. On my
> AMD Athlon server, however
> it will crash after a period of time, wich I cannot define. It just will suddenly crash. As long as ESpeak is being used (Sinse it is so responsive compared
> to a laggy Flite)
> On my AMD that has one gig of RAM, it is severely responsive.
> So quick, that I believe Speechd-up gets overwelmed with the generic module reloading, etc. So, Speechd-up chooses to crash. What do you all think about my question, suggestions?
> Regards,
> --Keith.

-- 
Ralph.  N6BNO.  Wisdom comes from central processing, not from I/O.
rreid at sunset.net  http://personalweb.sunset.net/~rreid
...passing through The City of Internet at the speed of light...
TAN (x) = SIN (x) / COS (x)