CCT Special Speaker: Joachim Ganseman
November 1st, 2016
Composition and Computer Technologies Presents:
Vision Lab, Dept. Physics
University of Antwerp, Belgium
Wednesday, November 9, 2016
B11 Old Cabell Hall (aka the VCCM)
(moving to 107 Old Cabell Hall, if necessary)
Abstract: Source separation is one of the classical challenges in audio analysis: given a mixture of sounds, we wish to extract and isolate one or more of the original sound sources. In general, this problem is intractable. However, music scores can give us ample of information to help guide the process of separation: it tells us what instrument should play what note when. In this talk I give a short overview of the existing approaches to score-informed source separation, the different components of such a system, their upsides and downsides, and open problems. I’ll also demonstrate a new approach that is currently still under development, inspired by methods that are used in the domain of remote sensing.
Biography: Joachim Ganseman received B.Sc. and M.Sc. degrees in Computer Science from the University of Antwerp, and is pursuing a Ph.D. degree on the topic of score-informed source separation at the VisionLab of the University of Antwerp. Part of this research has been conducted as visiting scholar at Stanford University’s Center for Computer Research in Music and Acoustics and Queen Mary University of London’s Centre for Digital Music. His academic areas of interests are music information retrieval and machine listening. From 2014 to 2016, he developed the IT systems and new music library catalog of the Royal Conservatory of Brussels. His free time is spent playing lots of piano and getting involved in way too many extracurricular activities for the promotion of Computer Science education in high school.
Hosted by Luke Dahl.