Extracting repeats from media streams

Many media streams consist of distinct objects that repeat. For example broadcast television and radio signals contain advertisements, call sign jingles, songs and even whole programs that repeat. The problem we address is to explicitly identify the underlying structure in repetitive streams and deconstruct them into their component objects. Our architecture assumes no a priori knowledge of the streams, and does not require a pre-trained database. Everything the system needs is learned on the fly. We demonstrate that using a modestly capable computer it is perfectly feasible to identify in realtime repeating objects that occur days or even weeks apart in audio or video streams. We outline the algorithms, enumerate several applications and present results from real streams.