The speech plugin "reads" song title as provided by input plugin through speech synthesizer.

Current version supports output through Festival Speech Synthesizer server. You must install this synthesizer before using speech plugin.

Festival can be downloaded from its homepage at http://www.cstr.ed.ac.uk/projects/festival.html
Don't worry, prempiled binary distributions are also available. It's highly recommended to use MBROLA voice database providing much better voices for both british and US english (other languages are also supported, please refer to Festival's manual).

• festival server host and port
• preferred voice (including MBROLA if installed)
• speech speed

1. The most of source code is "stolen" from song_change.c and festival_client.c
2. No header files for Festival connectivity are needed. Plug-in uses standard unix sockets. It's able to connect to any Festival server running on any host (refer to Festival server-mode configuration).
3. The most users run the Festival server on the same host as XMMS. Plug-in stops XMMS and release /dev/dsp for Festival while need to speek. So you DON'T need any audio output sharing (e.g. Enlightened Sound Daemon known also as EsoundD).

• Source
        speech-xmms-plugin-0.1.0.tar.gz

You can get more information about XMMS at http://www.xmms.org.