Current Events



1. New Project: Music Mining Plant

Project funded by the Academy of Finland for the period 08.2009-07.2014, under Olivier Lartillot's Academy Research Fellowship

Computational tools are needed in order to efficiently manage and browse into the large amount of digital music databases largely available online. This project is aimed at conceiving an in-depth computational system for the analysis of musical databases, integrating all dimensions of musical representations -- low-level audio and symbolic descriptions, middle-level structures and configurations and high-level conceptualizations -- into a highly intercommunicating network.

  • At low-level, we propose to unify the seemingly separate areas of study of audio transcription and symbolic segmentation into a common methodology of auditory scene analysis, and to guide these processes based on the expectation already inferred from the higher levels.
  • At mid-level, detailed but nevertheless concise descriptions of the motivic content of music will be enabled within our proposed intensive pattern mining paradigm.
  • At high-level, musical concepts are modelled as weighted combinations of features and structures, automatically selected among a very large area of possible choices, represented in a connectionist knowledge representation, such as artificial neural networks.

This will require a strong cooperation between disciplines as diverse as signal processing, cognitive science, artificial intelligence, knowledge discovery, information retrieval, operational research, computational engineering, semiotics and musicology. The resulting platform will offer further understanding of music perception and cognition, and can be applied both for advanced musicological studies and as a leading-edge platform for intelligent browsing of music databases.