On Fri, 2008-06-06 at 20:46 -0400, Nick Stockton wrote: > *nods* I remember trying that but it didn't seem to work. > it was making the sound start out speaking a few words very fast then it > would freez and not speak at all. Speaking fast in gnome-speech might be due to the fact that gnome-speech seems to be able to reach much higher speeds than speech-dispatcher does (on my system I need to set speech-dispatcher to full (99) in orca to achieve what seems nearly fast enough where as whith gnome-speech I am dropping it down to about 60 (although I may speech it up once I have tuned it a bit better for pitch and such like)). > I kind of wonder if it was because I have my USB sound card set to output > 44,100HZ in /etc/asound.comf and viavoice would be using 22,000HZ. What was the rate of speech when using gnome-speech's test application (test-speech)? If this seemed the fairly standard slow default speed applications normally set then sample rate conversion is probably occurring. It's quite common for alsa to resample for you, any device you set up as plug or to use dmix will automatically resample for you. > When ever I get a faster computer I'll probbely try running gnome-speech > again but at the momentgnome speech is not very responcive on my computer > and speech-dispatcher is quite a bit more responcive. > I kind of wonder what is causing it to spell some words but not others > though... Very strange. >From what I can tell, the following causes speech-dispatcher (when using orca, when using speechd-up I think I got slightly different results): If the word contains non-alphabetic characters (eg. speakup-request at braille.uwo.ca gets some parts spelt) If something has capitals in the midst of it (eg. java variable names or method names, although the method names will have non-alphabetic characters after it)) Those above rules probably could be improved upon to be more specific but it's a starting point. Some words get spelt by gnome-speech, but this is less frequent and I am still working when this occurs. I don't remember espeak doing this so much so I don't think it's in the main speech-dispatcher code just the ibmtts module or ibmtts itself. Michael Whapples
