Dan Ellis, Columbia University, USA: "Mining for the meaning of music"

ABSTRACT:
Over the past several years we have been investigating ways to exploit the increasingly large amounts of music audio data that we have available, such as the hundreds of hours of music on the average iPod. One way of looking at a large body of examples is that they constitute an implicit definition of a category: given the right representations, and the right distance metric, other items that are “close” to one of the examples belong to the same class, and the remainder do not.  This talk will describe various aspects of our ongoing project to extract this “definition” of music, as the encompassing space defined by a very large number of examples. In this sense, we are using data mining to produce an operational definition of the term “music”, in so far as our set of examples covers the full range of music.

 

VIDEO ARCHIVE - DAN ELLIS

 
APA video citation:
Ellis, D. (2014, May 22). Mining for the meaning of music -
CIRMMT Distinguished Lectures in the Science and Technology of Music. [Video file].
Retrieved from https://www.youtube.com/watch?v=D5EJLQlUigM