:) I wrote one of these modules for a project a few weeks ago. It's not in great shape (i.e. it only performs very basic recognition and buffer insertion (none of that understanding the english language stuff), but it's good to serve as a prototype for connecting ViaVoice on Windows to XEmacs. The only interesting bit that is necessary in the XEmacs source code is to modify the Windows event loop (mswindows_wnd_proc, in event-msw.c) to add something like this under the WM_COMMAND case: #ifdef HAVE_SPEECH switch(wParam) { case SPEECH_CONNECT_ID: SmDispatch(lParam); return 0; } #endif SPEECH_CONNECT_ID is the constant that you send to the Speech API when you initialize it. This SmDispatch function will call any of the proper callbacks you've registered with the Speech API without having to modify the XEmacs source more than necessary. You can find info on the API at: http://www-4.ibm.com/software/speech/dev/ Andy Begel