REAL-TIME RHYTHM & TEMPO PERCEPTION<< back to Music Understanding & Tracking
One active research field at our institute is the automatic estimation of beat and downbeat (beginning of a bar) times from an audio signal. While these systems work mostly well in offline scenarios (where the system has access to the whole audio signal), they often fail in cases where this information is not available (online mode). This online beat/downbeat tracking is particularly important for interactive music systems, e.g. robots that play music with human musicians.
Recently, our institute constructed a working prototype of a drum "robot", that is able to accompany a musician by "listening" to what he/she is playing, detecting the rhythmic structure, and playing accordingly. However, it is limited in what it can do:
This context opens the possibility to work on multitude of topics. These topics range from practical implementations of algorithms with real-time constraints, to applied (offline) machine learning.
Note that musical knowledge is not strictly necessary to work on any of these topics.
The Automatic Ballroom Dance Instructor
Imagine you are on a wedding and the band starts to play. Everybody gets up, finds a partner and starts to dance. You always thought that ballroom dances are boring, but somehow the bands plays really great and you are getting bored sitting alone in the corner. You switch on the Automatic Ballroom Dance Instructor (ABDI) app on your smartphone, let the app listen to 20 seconds of a song and it immediatly tells you which dance style you should dance. Moreover, it also displays a sequence of dance steps synchronised to the music, making it easy for you to follow, join the crowd and dance through the night...
In this context, the following topics could be considered:
Contact: Florian Krebs
last edited by kf on Sep 29, 2015