TrainC - Version 0.5 TrainC builds compact context dependency transducers for WFST-based speech recognition from acoustic training data. REQUIREMENTS This version is known to work under Linux and MacOS X using g++ (>= 4.1). OpenFst (http://www.openfst.org) is required. gperftools (http://gperftools.googlecode.com) is optional. CppUnit (http://cppunit.sf.net) is optional. INSTALLATION Follow the generic GNU build system instructions in ./INSTALL. Options: --with-openfst-include-path Path to OpenFst header files (default: /usr/local/include) --with-openfst-lib-path Path to OpenFst library files (default: /usr/local/lib) --with-google-perftools Path to gperftools (optional, default: autodetect) --enable-threads Enable support for multi threading (requires pthreads, default: no) --enable-tests Build unit tests (requires CppUnit. default: no) DOCUMENTATION See http://trainc.googlecode.com for documentation.