Detection and Correction of Lip-Sync Errors Using Audio and Video Fingerprints

A method for measuring and maintaining time synchronization between an A/V stream (A/V) is described. A/V fingerprints are used to create a combined A/V synchronization signature (A/V Sync Signature) at a known reference point. This signature is used at later points to measure A/V timing synchronization relative to the reference point. This method may be used, for example, to automatically detect and correct A/V synchronization (i.e., lip-sync) errors in broadcast systems and other applications. Advantages of the method described over other existing methods include that it does not require modification of the audio or video signals, it can respond to dynamically changing synchronization errors, and it is designed to be robust to modifications of the A/V signals. Although the system requires data to be conveyed to the detection point, this data does not need to be synchronized with, or directly attached to, the audio or video streams. This method uses fingerprints; it also enables other fingerprinting applications within systems, such as content identification and verification. In addition, it may be used to maintain synchronization of other metadata associated with A/V streams.