Ημερομηνία: Δευτέρα 3 Σεπτεμβρίου, 2012 Ώρα: 11:30 - 13:00
Τοποθεσία: Aίθουσα Συναντήσεων "Αλκιβιάδης Χ. Παγιατάκης", ΙΤΕ, Ηράκλειο, Κρήτη
Host: Γιάννης Στυλιανού
In this talk recent approaches will be summed up that Holzapfel proposed in his cooperation with OFAI in Vienna and INESC TEC in Porto. One focus will lie on the estimation of rhythmic similarity between two pieces of music. Approaches were compared regarding their tempo robustness, and it will be conclusions will be presented regarding how far tempo is an important criterion for rhythmic similarity. Secondly, an overview will given about a system based on mutual agreement to assess the quality of beat sequences in relation to a piece of music. We presented an approach that is capable of selecting pieces that are difficult for beat tracking without any prior knowledge of the ground truth. Insight will be given in how far automatic beat tracking agrees with human beat perception, and which signal characteristics lead to problems for algorithms and for humans.
Andre Holzapfel completed his PhD at the Department of Computer Science at the University of Crete in 2010, with focus on computational methods for ethnomusicology. From September 2010 to February 2011 he worked as a post-doctoral researcher at the Austrian Research Institute for Artificial Intelligence (OFAI), focused on the improvement of rhythmic similarity measures for music. From April 2011 until March 2012 he joined INESC TEC in Porto for a post-doctoral research on beat tracking and rhythm synthesis. He is now working for the Compmusic project at Universitat Pompeu Fabra in Barcelona, exploring the rhythmic properties of classical Ottoman music.