Installation

This document provides information on how to install and configure Speech Dispatcher to work with LSR.

1. Download and install Speech Dispatcher.

2. Download and unpack festival-utils if Festival speech engine is to be used. Copy *.scm files to appropriate location as outlined in README file.

Speech Dispatcher Configuration

1. Edit speech-dispatcher/speechd.conf. Choose the output module you desire in the "OUTPUT MODULES CONFIGURATION" section and set the DefaultModule just below this section.

LSR Configuration

1. Some speech engines, such as festival, may need to be started first. IBMTTS does not need to be started.
2. Start speech dispatcher using speech-dispatcher. Check output to make sure output module was correctly loaded and connected.
3. Start LSR and use the key combination Alt-Shift-q to bring up settings dialog.
4. Go to the "User Profile" and select the "Profile Devices" tab. A list of available device will be shown.
5. Select "Speech Dispatcher" and "raise" it above all speech devices.

Troubleshooting

Festival is a freely available speech engine so the troubleshooting section will focus on this speech engine. Although the test scripts may differ, the same methodology should apply. Troubleshooting should be done in an incremental manner. First, make sure the speech engine is working properly, then make sure speech dispatcher is communicating with the speech engine and then LSR is communicating with the speech dispatcher properly. Use the following steps:

1. Start festival-server and let it run in the foreground.

2. Start festival and at the prompt type (SayText "Hello, World!"). You should here "Hello World". Alternatively, you could use test-speech to test festival. If there is no speech output and no debug output for festival-server then festival-server might not be configured properly. Check the configuration file for host, port etc. If there is no speech output but there is debug output on festival-server, make sure your sound system is functioning and repeat test. If output is heard go to step 3.

3. Start speech-dispatcher in the foreground and check the debug output to make sure festival connected. If it did not connect, check speech-dispatcher/speechd.conf to make sure festival is the default module (DefaultModule) and that it was added using AddModule.

4. Go to speech-dispatcher-0.6.x/src/tests and run ./run_test basic.test. The test will connect, configure the speech engine and then output several pieces of text at various settings. If you got output go to step 5, otherwise examine speech-dispatcher/speechd.conf for errors and repeat test.

5. Start lsr using lsr -l debug. Examine the output and make sure SpeechDispatch loaded properly. If it did not, run lsr -s -p <profile>, where <profile> is the name of the profile you are using. Make sure SpeechDispatcher is shown. If not, run lsr -a SpeechDispatcher -p <profile> --index=0. Restart LSR and check that SpeechDispatcher loaded properly.