A ranking technique for fast audio identification
Frank Kurth · 2004
We introduce a novel ranking technique for fast and robust audio identification. In this approach, identification is performed by evaluating certain parts of correlation sequences. Starting form a general framework for content-based audio identification, we show to integrate our new algorithm into a fast index-based search algorithm. We demonstrate the capabilities of our approach by considering the tasks of identifying highly distorted audio material and search fragments in large scale MP3 data bases.