Festival is a software that helps to build speech synthesis systems with pre-built modules. It provides a comprehensive framework and module examples.
If you are looking to build your own speech synthesis system, Festival provides a great general framework to get started, as well as a range of example modules. This comprehensive system offers text to speech capabilities through a number of different APIs, including those for shell level, Scheme command interpreter, C++ library, Java, and even an Emacs interface. It currently supports multiple languages, including English (British and American) and Spanish, with other groups releasing additional languages for the system.
What's more, if you need to create a voice specifically for your system, Festival offers full tools and documentation for you to build your own voice, which are available through Carnegie Mellon's FestVox project. The system is written in C++ and uses the Edinburgh Speech Tools Library for low level architecture, while a Scheme (SIOD) based command interpreter provides control. Finally, documentation is provided in the FSF texinfo format, meaning that you can generate a printed manual, info files and HTML as needed.
It's worth noting that Festival is completely free software, distributed under an X11-type license, which allows unrestricted commercial and non-commercial use. The software has some strong features, including backwards compatibility with Festival 1.4.3, and waveform synthesizers including those based on diphones and MBROLA database support. There are also externally configurable language independent modules, including for phonesets, lexicons, letter-to-sound rules, tokenizing, part of speech tagging, intonation, and duration. Festival is highly portable, supporting Unix distributions, and features SABLE markup, Emacs, client/server (including Java), and scripting interfaces.
Looking to the future, the latest release of Festival includes an HTS hidden Markov model based synthesis engine from Nagoya Institute of Technology, as well as the Multisyn general purpose unit selection synthesis engine from CSTR. Support for gcc 3.2 and 3.3, and Intel 8.0 has been added, as well as clunits unit selection improvements and Apple OS X support. Numerous bug fixes have also been included in the latest version.
What's more, if you need to create a voice specifically for your system, Festival offers full tools and documentation for you to build your own voice, which are available through Carnegie Mellon's FestVox project. The system is written in C++ and uses the Edinburgh Speech Tools Library for low level architecture, while a Scheme (SIOD) based command interpreter provides control. Finally, documentation is provided in the FSF texinfo format, meaning that you can generate a printed manual, info files and HTML as needed.
It's worth noting that Festival is completely free software, distributed under an X11-type license, which allows unrestricted commercial and non-commercial use. The software has some strong features, including backwards compatibility with Festival 1.4.3, and waveform synthesizers including those based on diphones and MBROLA database support. There are also externally configurable language independent modules, including for phonesets, lexicons, letter-to-sound rules, tokenizing, part of speech tagging, intonation, and duration. Festival is highly portable, supporting Unix distributions, and features SABLE markup, Emacs, client/server (including Java), and scripting interfaces.
Looking to the future, the latest release of Festival includes an HTS hidden Markov model based synthesis engine from Nagoya Institute of Technology, as well as the Multisyn general purpose unit selection synthesis engine from CSTR. Support for gcc 3.2 and 3.3, and Intel 8.0 has been added, as well as clunits unit selection improvements and Apple OS X support. Numerous bug fixes have also been included in the latest version.
What's New
Version 1.96: N/A