Abstract: Methods and apparatus, including computer program products, for synchronizing segments in a signal stream having an audio component and a text component. In one aspect, a search is made in the text component for a marker text string consisting of at least one word, character or phrase; an audio pattern representative of the marker text string is generated; the audio pattern is located in the audio component; and, the audio and text components are aligned temporally. In another aspect, cue points delineating a start and/or end of a multimedia segment are found. In another aspect, a user can search for a multimedia segment and either read the text feed or listen to the audio feed.