The CMU Sphinx Group Open Source Speech Recognition Engines

Check out the code:

https://cmusphinx.svn.sourceforge.net/svnroot/cmusphinx/trunk/cmuclmtk

The repository also contains binaries for Windows and Linux. Binaries are stored by OS, in bin/x86-nt/ and bin/x86-linux/.

Configure and compile in linux:

To compile for a 32-bit word ID space, add "--enable-32bits" to the configure command line. It's not a bad idea to always include this flag.

aclocal && autoheader && automake --add-missing --copy && autoconf
./configure CFLAGS="-g -Wall -O0" CXXFLAGS="-g -Wall -O0"
make && make install

Configure and compile for Windows:

You should do your development underCygwin since the makefiles are autobuild-oriented. Be sure to download the developer and mingw components. To compile for a 32-bit word ID space, add "--enable-32bits" to the configure command line. You want to compile with mingw since this will give you Windows-native binaries and will therefore be portable from computer to computer.

aclocal && autoheader && automake --add-missing --copy && autoconf
./configure --enable-mingw CFLAGS="-g -Wall -O0" CXXFLAGS="-g -Wall -O0"
make && make install

For Development on Linux

1. build with autoconf
2. commit
3. do an update from a windows box with cygwin
4. compile under cygwin
5. commit again

For Development on Windows

1. build with cygwin
2. commit
3. do an update on a linux box
4. build with autoconf
5. commit again

If you're working on a filesystem shared by linux and windows then you can skip the middle commit and update steps.