Multimodal music information extraction

Consists in the development and continuous improvement of tools, algorithms and computational models for extracting acoustic and kinematic parameters from both the audio signal and the three-dimensional spatial position of musicians, to support the requirements of internal research projects.

  • Models for extraction and parameterization of musical expressiveness content
  • Content analysis of note attack and note transitions in music performance
  • Segmentation and parameterization of musician’s physical gesture