Speech Dispatcher

common interface to speech synthesis

Maintainer: Hynek Hanke <hanke@brailcom.org>

Index:

Introduction
Future
People
Documentation
Source code
Mailing-lists

Quick links:

Introduction

The goal of Speech Dispatcher project is to provide a high-level device independent layer for speech synthesis through a simple, stable and well documented interface.

Key features:

Common interface to different TTS engines
Handling concurrent synthesis requests – requests may come assynchronously from multiple sources within an application and/or from more different applications
Subsequent serialization, resolution of conflicts and priorities of incomming requests
Context switching – state is maintained for each client connection independently, event for connections from within one application
High-level client interfaces for popular programming languages
Common sound output handling – audio playback is handled by Speech Dispatcher rather than the TTS engine, since most engines have limited sound output capabilities

What is a very high level GUI library to graphics, Speech Dispatcher is to speech synthesis. The application neither needs to talk to the devices directly nor to handle concurrent access, sound output and other tricky aspects of the speech subsystem.

Supported TTS engines:

Festival
Flite
Espeak
Cicero
IBM TTS
Espeak+MBROLA (through a generic driver)
Epos (through a generic driver)
DecTalk software (through a generic driver)
Cepstral Swift (through a generic driver)

Supported sound output subsystems:

The architecture is based on a client/server model. The clients are all the applications in the system that want to produce speech (typically assistive technologies). The basic means of client communication with Speech Dispatcher is through a TCP connection using the Speech Synthesis Interface Protocol (See the SSIP documentation for more information). High-level client libraries for many popular programming languages implement this protocol to make its usage as simple as possible.

Supported client interfaces:

C/C++ API
Python API
Java API
Emacs Lisp API
Common Lisp API
Guile API
Simple command line client

Existing assistive technologies known to work with Speech Dispatcher:

speechd-el
Orca (see http://live.gnome.org/Orca/SpeechDispatcher)
Speakup (through Speechd-Up)
Yasr
LSR
BrlTTY

Future

Currently, only speech output is supported. However, in near future, the task of handling the speech synthesis devices will be passed over to the new TTS API Provider. Handling of braille output can be easily achieved through BrlTTY. Speech Dispatcher will then concentrate on providing a very high level API, advanced synchronization and prioritization of messages, history and client specific configuration. Output for both speech and braille will be joined under this new version of API so that the applications only need to send messages once and so that both outputs can be coordinated to provide better comfort for the user.

People

Core team:

Hynek Hanke <hanke@brailcom.org>
Milan Zamazal <pdm@freebsoft.org>
Jan Buchal <buchal@brailcom.org>
Tomas Cerha <cerha@freebsoft.org>

Contributors:

Lukas Loehrer <listaddr1 at gmx dot net> ... eSpeak output module
Gary Cramblitt <garycramblitt at comcast dot net> ... IBM TTS output module, ALSA audio output
Olivier Bert <obert01 at mistigri dot org> ... Cicero output module
Jacob Schmude <jschmude at adelphia dot net> ... generic output module configuration for DecTalk software synthesis.
Steve Holmes <steve at holmesgrown dot com> ... generic output module improvements
Gilles Casse <gcasse at oralux dot org> ... PulseAudio output
Luke Yelavich <themuso at themuseo dot com> ... audio output subsystem fixes
C.M. Brannon <cmbrannon at cox dot net> ... bug fixes
Rui Batista <ruiandrebatista at gmail dot com> ... Java client API

Documentation

Documentation is available as a part of the project and is distributed together with the source code.

The HTML version can be accessed on-line:

Speech Dispatcher Documentation ... includes both user's and developer's documentation of Speech Dispatcher and its client interfaces
SSIP Documentation ... communication protocol specification

Source code

Latest released version is 0.6.7.

The source code is managed using GIT. There are the following options to get the current development version:

Clone the git repository from git://git.freebsoft.org/git/speechd
Use GIT web interface.

Please send patches to project maintainer.

The current development version is maintained by Luke Yelavich at:

git://git.themuso.com/git/speechd.git

Java API is developed separately in CVS module speechd-java:

Download Java API CVS snapshot.
Use a web interface to Java API CVS.
Access the CVS module speechd-java directly.

Anonymous CVS access (login with an empty password):

cvs -d :pserver:anonymous@cvs.freebsoft.org:/var/lib/cvs login
cvs -d :pserver:anonymous@cvs.freebsoft.org:/var/lib/cvs checkout speechd-java

Mailing-lists

There is a public mailing-list speechd@lists.freebsoft.org for this project.

This list is for for Speech Dispatcher developers, as well as for users. If you want to contribute the development, propose a new feature, get help or just be informed about the latest news, don't hesitate to subscribe. The communication on this list is held in English.

See also the listing of all Free(b)soft mailing lists.