Xavier Serra gives two seminars at the University of Rochester (New York) on work done at the MTG on CompMusic. On March 2nd he gives a seminar at the Department of Electrical and Computer Engineering and on March 3rd another seminar at the Eastman School of Music.
March 2: Music Information Retrieval approaches for Music Understanding
Music Information Retrieval is a research field that involves the use of information processing methodologies to model music. This field has had an amazing growth and success in recent years and many music companies are using the resulting technologies to develop products and services for the creation, distribution and interaction with music and music-related information. However, an important question is weather these MIR methodologies are capable of helping us to better understand music. Most experts would say not that much and they would also acknowledge that this is a general problem of most machine learning approaches. But we want to do better, we want to take advantage of these powerful methodologies to also contribute to the topic of music understanding. In this talk I will present and discuss issues around this by going over some of the recent research I have done with my team. I will specially cover the work done in the CompMusic project, in which we aimed at advancing in the automatic description of music by emphasizing cultural specificity.
March 3: Computational studies of several non-western musical repertoires
The use of computational methods in the study of music through the processing of digital artifacts goes back several decades, but it is only in the past few years that this type of research has started to obtain some musicologically relevant results, mainly from the approaches taken in the field of Music Information Retrieval (MIR). With these approaches we have been able to validate some existing musicological knowledge, but we are still at the very beginning. In this talk I will go over several of the studies done at the MTG in the context of CompMusic, a project in which we have focused on five non-western music repertoires, Hindustani (North India), Carnatic (South India), Turkish makam (Turkey), Arab-Andalusian (Maghreb, or North Africa), and jingju (Beijing Opera; China). We have created corpora for each repertoire, have developed computational methodologies with which to study their melodic and rhythmic characteristics, and have obtained some musically relevant results to help us better understand these music traditions.