cgstmad

For people interested in articulatory speech synthesis, GnuSpeech is a very useful tool, because it is a complete, free software articulatory text-to-speech (TTS) system. This project was started in 2007, when the GNUstep port of GnuSpeech was not in a good shape. The objective was to reproduce the essential parts of GnuSpeech's functionality. To minimize the effort, only the GnuSpeech's tube model (written in C) and the articulation data (XML) were copied. Then a controller was written, and an external front-end adapted.

Some years later, with some adjustments the GNUstep port of GnuSpeech was improved. Now it is not as good as the OS X port, but you can load the default articulation data, enter a text and get the synthesized speech through the audio device. For this reason, this software is only for people who doesn't want to use GNUstep or who want to try all the alternatives.

The current project status is alpha.

Note: cgstmad stands for "Controller + GnuSpeech Tube Model And Data".

Overview

Samples

Voice Type Front-end Target Formats
en1 male Festival en_US WAV OGG
en2 female WAV OGG
en3 child WAV OGG

Tools

Download (SourceForge.net)

Cgstmad is written in C++ and is released under the GNU GPL. It was developed on a GNU/Linux-based system but it may possibly work on other POSIX-compatible systems.

The latest source code is available on the Subversion (SVN) repository.

Links

Contact

marcelo at mym.eng.br