A method performed by one or more processing devices includes receiving data indicative of dictated speech that has been spoken by a user during speech dictation; causing speech recognition to be performed on the data to obtain units of text; selecting a unit from the units, wherein the unit selected corresponds to a portion of the data received at a time that is more recent relative to times at which others of the units are received; and generating, based on an output of the speech recognition, data for a graphical user interface, that when rendered on a display device, causes the graphical user interface to display: a visual representation of the dictated speech, wherein the visual representation includes a visual indicator of the unit selected; and a control for performing dictation correction on the unit selected in real-time during the speech dictation.