User Tools

Site Tools


Conducting Music - Mapping Movements to Measures

Authors Harald Grohganz, Sebastian Böck, and Sebastian Stober
Affiliation University of Bonn, Johannes Kepler University Linz, Western University London (Ontario)
Code Github Link

In this project, we contributed some Machine Learning techniques which may lead to an advanced “conductible” MIDI player. At the end, this player might provide tempo control by hand gestures in real-time.


The interface mainly consists of a MIDI player which provides the actual position within each measure during playback. To detect the conductor's movement, we use Dense Optical Flow - hence it is not necessary to track a conductor's hand or baton. This is some previous work, written entirely in Java.

Our goals at HAMR:

  • Test some movement features.
  • Estimate the conductor's position in measure.

Future work:

  • Compare both positions in measure to estimate the tempo change rate.

The Model

Will be presented.

conducting_music_-_mapping_movements_to_measures.txt · Last modified: 2014/10/26 05:21 by superbock