I am sending this message again, because the mailing list returned it to me the first time saying that my message was too large... Sorry, I did not know there was a 5 KB limit! ************************************** > Hello: > Yes, I am using FC2. So, here are the details: > > 1. My kernel: > I downloaded a copy of the 2.6.8.1 kernel.org kernel source, applied some > performance patches from http://members.optusnet.com.au/ckolivas/kernel/ > and used the "checkout" script to patch speakup into it. I am not sure > the performance patches do anything for speakup, but they might help with > the software synth and the audio subsystem (just a guest). Then, of > course, I configured and compiled the kernel with the sftsyn module. > Important: > you have to create the device /dev/softsynth to use software speech... Do > this by typing: > > # mknod /dev/softsynth c 10 26 > > Make sure /dev/softsynth is created properly. I do something like: > > # file /dev/softsynth > > This should say something like: > /dev/softsynth: character special (10/26). > > > 2. Flite: > At first my problem was getting flite to compile properly. For some > reason I was getting tons of errors and I could not think of anything to > help me. So, I checked out the Speakup mailing list and got the address to > some flite RPM's and those worked wonderfully. Just make sure you get the > flite rpm and the flite-devel rpm or else speech-dispatcher will not > compile. > > > 3. speech-dispatcher: > This is important! After you extract the sources for speech-dispatcher: > you have to create a symbolic link in the speech-dispatcher source > directory pointing to /usr/include. The link has to be called "flite". > If you don't create this link it will not work... > Then, run ./configure, "make all", and "make install" and you should have > speech-dispatcher installed. > You can test speech-dispatcher by typing these commands: > > # speech-dispatcher // runs the speech service... > # pidof speech-dispatcher // should print the process ID for > speech-dispatcher > # spd-say "hello world" // tells speech-dispatcher to say hello world. > > Flite should speak the words "hello world"! > > 4. speechd_up: > I had no problems setting this up. I just extracted the source and ran > build.sh, then make and then make install. > > 5. running speakup with the new setup: > I removed my ltlk speakup module from the running kernel. If you forget > to remove the ltlk module and try to load the sftsyn module your system > might crash... At least that is what my system did! > Anyway, make sure speech-dispatcher is running. Then, type: > > # modprobe speakup_sftsyn // loads the speakup_sftsyn module > # speechd_up // The speechd_up thing > > By now, your machine should be talking... > > 6. Using another synth? > I paid the $50 and got the DECtalk 5.0 software synth from fonix. Here is > my impression: > The software is very fast, at least it is much much faster than flite. > Honestly, it works more quickly than my hardware double talk lt! At $50, > I don't think it is a bad investment if you are serious about software > speech. Just remember that software speech is dependent on your computer's > capabilities and if you try it on an i386 it might not be the same as what > I experienced. I have an overclocked Athlon system (2.5 GHZ) with 1 GB of > RAM and this might explain DECtalk's speed. Although, I can still tell > that it is much faster than flite... > Unfortunately, I was not very impressed with the quality of DECtalk's > voice. It was not what I had hoped for at all. It is not as good as it is > advertised by Fonix. Other than that, it was not bad at all... > > > I am sorry if I mist something. I am in Windows right now and I can check > any of this to make sure it is all correct. I could have forgotten > something. however, I don't think I did. > > David Bruzos > > ----- Original Message ----- > From: <willem at top.health.gov.za> > To: "David Bruzos" <david at bruzos.org> > Cc: <speakup at braille.uwo.ca> > Sent: Tuesday, November 02, 2004 1:22 AM > Subject: Re: writing spk howto (?) > > >> Hi, >> Go ahead with the software synth howto. >> As you say the docs is a little outdated. >> Do you have speakup running using software speech under fedora c2? >> If so, I would like to have the details, howto or no howto. >> Regards, Willem >> >
