wordperfect for linux uses an xterm window in which to run. If some way exists to pipe keystrokes into an xterm window and pipe screen output out of an xterm window to stdout and stderr when necessary possibly wordperfect might work for speech under linux.