If you want to use a speech command that does not have a speaker-independent template, you must have the user train the word. If the word is not trained, the speech command is not included when the user asks for voice Help. If you add a speech command without a speaker-independent template, and the user has not trained the word, short voice Help is disabled and long voice Help defaults to the following string, read after the other long voice Help strings: “You need to train the custom command <command> before you can use it.”