Re: [orca-list] Need help with gnome-speech.



Hi Will,
I took your suggestion, and grabbed the gnome-speech 04.21 sources and compiled them. The speech servers compiled and installed fine. However, for some reason I don't understand neither the Cepstral Swift voices or Dectalk is talking to me via the test-speech utility. When I test Callie with swift I am able to get her talking just fine. With Dectalk running the say utility also works. However, when I try to run them via test-speech it says server couldn't be loaded and the error was unknown. Now, one thing about Cepstral that may have something to do with this when installing Callie it said something about adding some lines to ld.so.conf. What lines I am suppose to add I am not sure because that text flew right passed me, and when I tried to read it with Orca it had scrolled off the screen. I think, but am not completely certain of this, it was asking me to put the libs on the path like
/opt/swift/lib
in /etc/ld/so/conf?
Just for future reference I am running Ubuntu 8.04 which was released last April with on a few updates installed. I just installed it over the week end, and haven't really had the time to do a lot of upgrades, updates, etc to it yet. One final note I did install Cepstral Callie 5.1 to the default location/opt/swift. I'm not sure why gnome-speech is unable to initialize it. Any ideas here how to resolve this?
Thanks.

Willie Walker wrote:
Odd. Looks fine to me. The missing-prototype thing you're seeing is a switch being sent to the compiler. I'd suggest trying gnome-speech-0.4.21, thhough:

http://ftp.gnome.org/pub/gnome/sources/gnome-speech/0.4/gnome-speech-0.4.21.tar.bz2

Here's a summary of the changes from 0.4.16 to 0.4.21:

gnome-speech 0.4.17:

* Fix for bug 503058 - Swift driver build failure on Solaris.  With
  this fix, Cepstral/Swift works on Solaris Express Community Edition
  build 79.

gnome-speech 0.4.18:

* Fix for bug 508439 - Gnome speech Espeak driver will not compile
  with Espeak 1.30 (Thanks David Csercsics!!)

gnome-speech 0.4.19:

* configure.in: be smarter about automatically finding espeak

* drivers/espeak/Makefile.am: remove libstdc++.  It is not needed
  and makes the Solaris build fail.

gnome-speech 0.4.20:

* drivers/espeak/*: fix for bug 535493 - eSpeak driver can
  block.  The fix adds a list mechanism to handle utterances
  to be spoken and then processes them on the gidle thread.
  Also raised the maximum speech rate for eSpeak.

gnome-speech 0.4.21:

* drivers/espeak/espeaksynthesisdriver.c: fix for bug #545896 - eSpeak
  segmentation fault.

After you compile/install, I'd also suggest running test-speech to make sure the speech stuff works.

As a note of warning, in Intrepid, Ubuntu has wrapped the synthesis drivers in shell scripts that then execute the binary. So, it might be that your build blew away the shell scripts that were in place. In particular, I believe the pattern Ubuntu uses is something like this:

1) The gnome-speech drivers are named /usr/bin/*-synthesis-driver, where the * is one of espeak, festival, etc. This is what bonobo looks for.

2) Ubuntu moves the /usr/bin/*-synthesis-driver binaries to /usr/bin/*-synthesis-driver.bin. It then creates /usr/bin/*-synthesis-driver shell scripts that call the *.bin files.

The build/install of gnome-speech is going to blow away the /usr/bin/*-synthesis-driver files.

Having said all that, I think "sudo apt-get install libgnome-speech7" package should get you all you want without needing to build gnome-speech from sources.

As a final note of warning, Linux audio sucks and it's especially flakey on Ubuntu Intrepid these days.

Will




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]