in case anyone's wondering, the binary around the logo corresponds to the ascii string "codetalker"
At this point, Codetalker has evolved to be a full tokenize-parse-translate solution, with an emphasis on ease and brevity without sacrificing speed of flexibility.
What is CodeTalker?
CodeTalker might be described as a compiler-compiler although that doesn't quite fit. I would probably call it a "compiler creation library", written in python (with a healthy dose of C for performance). With CodeTalker, I have written a JSON parser in 66 lines of code that beats most of the other python-based JSON parsers around.
...and why would I use it?
Currently, if you want to make a parser that recognizes a Context-Free Grammar, you can hand code it yourself (which is rediculous for all but the simplest of grammars), or use some sort of assisted parsing library. ANTLR is one that has been around for a long time. Yacc is another well known solution.
The cool thing about CodeTalker is that it doesn't rely on a specialized syntax for grammar definition -- it's all python -- so the grammar can take advantage of specialized "helper functions" which can make your grammar much cleaner and more readable.
The really awesome thing about CodeTalker is that it's fast. Fast to define, test and maintain your grammar (it is python, after all), but most importantly the parser it produces is fast.
Additionally, I know there's no accounting for taste, but I think the way that CodeTalker grammars + translators are laid out are more intuitive than other solutions...but of course I would; so take a look for yourself.
Here's my announcement, which is a pretty thorough into the whole thing...
And some benchmarks: http://jaredforsyth.com/blog/2010/jul/21/codetalker-doubles-in-speed/