Research
I am employed as Ph.D. student within the
Intelligent Sound project
(funded by the Danish Technical Research Council (STVF), STVF no.
26-04-0092). The purpose of the Intelligent Sound project is to combine
signal processing and machine learning with database technology to make
it possible to do intelligent search for sound.
My part of this is to find suitable, mathematical representations of sound
streams. The representations (features) must convey relevant
information about the sound, must be computationally tractable, and
must be of a suitable size for both searching and storing.
Currently, I focus on features to represent music. Much of my time is
spent cosidering how to actually evaluate the performance of potential
features, since this is not always trivial. I have placed Matlab
software that incorporates some of my attempts to evaluate properties of new
features on my page with
miscellaneous
stuff.