ROBUST AUDIO HASHING FOR CONTENT IDENTIFICATION (FriAmOR6)

Author(s) : Hamza Özer (Bogazici University/TUBİTAK, Turkey) Bülent Sankur (Bogazici University, Turkey) Nasir Memon (Polytechnic University, USA)

Abstract : We propose and evaluate three new perceptual audio hash functions. These hash functions are based on a concise description of the time-frequency characteristics. This information can be extracted from the time series of frame-by-frame fundamental period or from the singular value decomposition of the mel frequency cepstral parameters. Experiments show that the proposed perceptual hash functions are both robust and unique.

Menu