FreeTTS on steroids

I have been playing with FreeTTS and Festival for a while now, and it always bothered me a bit that I could not use a British English voice in FreeTTS. While looking into how difficult it would be to add support for that, I found out just how different the speech engine in FreeTTS differs from Festival. That's not a real surprise, since FreeTTS is based on Flite rather than Festival. But in all, I didn't really see much reason why adding in a lot of the features of Festival would pose a problem (or a performance degradation).

So, what started as a project to add a different dialect to FreeTTS turned into a fairly extensive rewrite of FreeTTS, to the point of no longer really being the same speech engine. That's why I decided to do a spin-off named jTTS. It's aim is to be a free speech synthesizer written entirely in Java, implementing a speech engine that is based on Festival. It supports both American and British English, both for voices and lexicons, and adds features such as a probabilistic Part-Of-Speech processor and better post-lexicon phoneme processing. It also features a more optimized lexicon format (making it unnecessary to split works into syllables at runtime for words in the lexicon).

As part of my doctoral work (albeit a side-project), I am working on some web pages about it as the jTTS project. The code will be downloadable from SourceForge.net as soon as a project is created there with the same name.