Stochastic Model of a Robust Audio Fingerprinting System

An audio fingerprint is a compact representation of the perceptually relevant parts of audio content. A suitable audio fingerprint can be used to identify audio files, even if they are severely degraded due to compression or other types of signal processing operations. When degraded, the fingerprint closely resembles the fingerprint of the original, but is not identical. We plan to use a fingerprint not only to identify the song but also to assess the perceptual quality of the compressed content. In order to develop such a fingerprinting scheme, a model is needed to assess the behavior of a fingerprint subject to compression. In this paper we present the initial outlines of a model for an existing robust fingerprinting system to develop a more theoretical foundation. The model describes the stochastic behavior of the system when the input signal is a stationary (stochastic) signal. In this paper the input is assumed to be white noise. Initial theoretical results are reported ...

• Type: Conference
• Year: 2004
• Where: ISMIR
• Authors: Peter Jan O. Doets, Reginald L. Lagendijk
• Keywords: Audio fingerprint, ISMIR 2004, Signal Processing Operations, Suitable Audio fingerprint