![]() |
Department of
Computational Perception
|
![]() |
MUSIC UNDERSTANDING & TRACKING
Feature Extraction, Structure Recognition, Music
Understanding
| In general,
we are interested in any project that aims at developing methods for
the automatic recognition of musical entities or structures (melody,
rhythm, segmentation, ...) in music recordings, for the recognition and
categorisation of music, new means of interaction with music, new ways
of displaying and visualising musical information, controlling or
modifying music in interesting ways, etc. The following list of example
projects is only a suggestion; we are open to any creative ideas you
may have. The little video to the right of the list visualizes certain
sound/rhythm characteristics in different kinds of music (from work by
a former Ph.D. student of ours). Example projects:
|
Feature extraction video by former PhD
student Elias Pampalk
|
Real-time Music
Following and Interpretation
| At our
institute, we are developing very efficient and reliable algorithms for
"listening to" and tracking music in real time, and for aligning music
audio streams to the printed score. Based on that, there is a lot of
exciting possibilities for real-time applications. The following list
gives a few examples. (Again: if you have other ideas, please let us
know!) The little video on the right (and Prof. Widmer as amateur pianist) shows an application of our real-time music tracking algorithm: an automatic page turning device for musicians that listens and knows when it is time to turn the page. Example projects:
|
Our automatic reactive page turner
|
Real-time Music
Interaction and Control
Although computers offer virtually unlimited freedom in the control, synthesis, and processing of sound, this potential has hardly been unlocked to performing musicians, because of the lack of intuitive control interfaces. Devices such as the Wiimote or Kinect would enable sound control by natural gestures of the body, relieving performers from the necessity to use computer keyboard and display. Gesture-tracking could also open up new possibilities for non-musicians, for instance by inferring intended musical character automatically by capturing arm gestures to create new interfaces to music searching and browsing or by capturing characteristics of the user's movement, and transfer them to a playing piece, in such a way that the user has real-time control over (for example) tempo and loudness of the music. The video on the right shows Prof. Widmer contactlessly controlling an expressive piano performance on a Bösendorfer CEUS Grand Piano using a MIDI Theremin. Example projects:
|
CEUS and the MIDI Theremin
|
last edited by pk on May 3, 2012