[orca-list] Test installers for speech-hub in Ubuntu 13.04

[Bill Cox <waywardgeek gmail com>]

I managed to get speech-hub working with the version of Orca that ships with Ubuntu 13.04.  I've done no testing in 13.10 yet, though I think it will probably work.  I am pretty confident it will not work with previous Orca releases.

This installer supports 4 speech engines: espeak, voxin, pico, and Mary TTS.  You have to have voxin installed on your computer for the this engine to work, and it's called ibmtts in Orca.  The other three in theory should work out of the box.

There are two separate installers, both of which can be found in this download directory:

http://vinux-project.org/downloads/

I had to right click on them and do a "save as" to save them.  There are two self-extracting installers that need to be run as root.  speech-hub-install.sh simply copies the speech-hub files to /opt/speech-hub, and installs two scripts in /usr/bin.  The first script is speech-hub, which just launches speech-hub.  The second script is sh-say.  After running speech-hub-install.sh, you can test if speech-hub is working with:

sh-say "this is a test"

If that works, and if you feel pretty handy with Linux and don't mind if Orca stops working, feel free to run the second script as root: speech-hub-orca-install.sh.  This script does exactly three things:

1) It modifies /usr/lib/python3/dist-packages/orca/settings.py, adding "speechhubfactory" to the list of available speech factories.  I do not know of any way around this - I had to edit the Orca python code to add a new speech server, but it's just one line.  If my installer script breaks Orca, try copying settings.py.orig back to settings.py.

2) It copies speechhubfactory.py to /usr/lib/python3/dist-packages/orca.  This is a modified version of speechdispatcherfactory.py.

3).  It copies speechhubfactory.py to /usr/lib/python3/dist-packages.  This is just the low-level Python interface to speech-hub.

The speech-hub code under /opt/speech-hub should be fairly Linux distro independent, as Isaac Porat wrote most of it in Java.  The C executables that interface to the speech engines I wrote to be as portable as possible.  The only non-portable part is the Orca interface, since many different versions of Orca exist out there, and I have to interface to Orca in Python.

I'm hoping that speech-hub's Python interface for Orca can become part of Orca in the future, just as speechdispatcherfactory.py is part of Orca.  That way, you would always have a compatible version with whatever version of Orca you install.

Isaac released speech-hub under a GPL v2 license.  Some of the other code is written by other authors under different licenses, but it's all documented in SpeechHub_license.txt.

Isaac already released the Windows source code.  He prefers to release tar-balls rather than having a public git repository.  The Windows source and licenses can be found on:

http://speechhub.org

I'll release the Linux tarball as soon as Isaac has a chance to review the source tree.

Thanks,
Bill