Byzantine Hymn Recognition with Audio Fingerprints Resistant to Noise, Tempo and Scale Changes

In this study, an audio fingerprinting approach is utilized that was successfully used for Byzantine hymn recogni tion. The process is based on the extraction of time and scale invariant fingerprints using the two-dimensional Fourier Transform (2DFT). It also involves a matching process between fingerprints based on a distance metric calculated from similarity matrices. The proposed method was evaluated on a database of 351 live hymn recordings as well as different noise scenarios and exhibited high accuracy.