Digital Audio Phylogeny. A Computer Vision Analysis on Spectrograms

The possibility of sharing multimedia contents on the Web results in the existence of multiple similar but not identical copies of the same document. Multimedia phylogeny deals with the analysis of dependencies among these objects, in order to lead back to the original document and make sense of its evolution over time. This thesis proposes an innovative approach to the phylogenetic analysis of audio tracks, where time-frequency representations are used to convert audio contents into images.