Synchronization of Processed Audio-Video Signals using Time-Stamps

This paper proposes a system for synchronizing audio and video signals after they have been captured and processed in separate modules. The captured audio signal is run through a speech recognizer while the video signal, captured using a single camera, is run through an object detector. The synchronization system uses time-stamps from both processes to produce a synchronized output containing the detected objects and their relative recognized speech. The processed audio and video streams are buffered and the paper proposes an integration scheme such that a synchronized output is realized.