Node:Current state, Next:Basic design, Previous:Why and how, Up:Introduction
Today, the development of programs and new technologies connected with speech synthesis under GNU/Linux is centered around two main points: visually impaired people and pure development. Although some fields are beginning to use synthesis for different purposes, like telephony servers, these are still like drops of water in the ocean. Here is a short (definitely not exhaustive list) software synthesizers, hardware synthesizers and applications known to work under GNU/Linux.
Hardware synthesizers are the devices, which may be connected to PC. Mostly they are external, connected via serial or parallel port. There are also some internal devices for ISA bus or USB. Application may send textual data to the port and the device converts it to spoken letters and words. Data may contain also several control sequences in the form of escaped characters as commands. The problem, we are facing, is that each of these devices uses its own communication protocol.
Festival is a multilingual Free Software text to speech synthesiser with high quality speech databases available. One of it's problems is that some of the most important databases are not free. (eg. the database for British English is non-free). Other problem is that Festival is intended rather as a platform for research and developement than as an end-user product and therefore is big and not-so-easy to install. The problem we face as Speech Deamon Developers is that it's too slow to be really useful for most aplications.
Flite stands for Festival Lite and it is a light fully free english speech software synthesiser with good quality of sound, developed by the authors of festival as an end-user product. It's very fast, however, we currently don't know how to configure it (it seems it is not possible yet) and it seems that the developers have some problems with importing the voices from Festival. Speech Deamon currently uses Flite as it's primary output module for English.
Odmluva is a simple (and very light) czech speech synthesizer available under the terms of GNU GPL. We are working on it's support in Speech Deamon.
Epos is czech synthesis. It is an academical project and it already gives quite good results, but some parts are covered by a proprietary license.
Free TTS is some JAVA-based text-to-speech system. We didn't checked it yet.
ViaVoice is a multi-lingual software synthesizer available for GNU/Linux. The main problem is that ViaVoice is not free (as in freedom). Until IBM changes it's licence, we can't use it in Free World / Free Operating System and therefore it's not and will not be supported in Speech Deamon.
MBROLA is a multi-lingual software synthesis available for GNU/Linux. MBROLA is not free as in freedom, although it's gratis. The same problems as with IBM ViaVoice prevents us to include it in Speech Deamon.
The EmacSpeak (by T. V. Raman <raman@cs.cornell.edu>) software package provides speech output for Emacs, and includes ,,speech servers" for the DECtalk speech synthesizers.
The package emacspeak-ss provides servers for several additional synthesizers. None of these programs are normally run by the user directly. Instead, they are run by Emacs. That is: Emacs runs the emacspeak code, which executes Tcl, which interprets the server code. This approach is too closely ,,wired" to usage with Emacspeak, so it can't be used for our general purposes.
This does not mean, that these servers are compleetly a bad idea and we can not use them. Thanks to the author Jim Van Zandt <jrv@vanzandt.mv.com>, we can learn from the sources and write the output driver modules for Speech Daemon (emacspeek-ss is GPL).
GNOME windowing toolkit library.
Windowing toolkit library.
Windowing toolkit library.
Windowing toolkit library.
Speakup is a kernel patch that provides low level speech output for visually impaired, so it works even if there is some problem in configuration and you can't run EmacSpeak.
Brltty is mainly a driver for different Braille displays, but also supports some kind of software synthesis.
We hope to be able to integrate Speech Daemon into these projects in the future.