A command embedded in textual input that a speech synthesizer interprets and applies to the pronunciation of the spoken version.