The EASAIER Sound Archive Integration and Enrichment Framework

  • Francois Scharffe, STI Innsbruck, University of Innsbruck, Austria
  • Michael Luger, STI Innsbruck, University of Innsbruck, Austria
  • Luc Barthelemy, SILOGIC, Toulouse, France
  • Christian Landone, Queen Mary, University of London, United Kingdom
  • Yves Raimond, Queen Mary, University of London, United Kingdom
  • A significant amount of archives in the cultural heritage domain has been digitized in the recent past. With the advent of the Web, it has become possible to publish such content to a broad audience. Still, such digitized assets typically exist isolated from each other, hindering meaningful cross-archive search over archives that share a similar thematic domain. The Semantic Web aims to overcome such issues, providing advanced possibilities for publishing information on the Web. The foundation for this are ontologies, providing a shared domain conceptualization. Archive content along with corresponding metadata can be made available in a structured way, taking into account its underlying rich semantic structures. This provides the foundation for precise and enriched querying possibilities, such as cross-media and cross-archive search. The EASAIER project provides a framework for enriched access to musical archive content, with libraries, museums, broadcast archives and music schools in mind. The framework applies recent advances in music and speech processing in combination with information retrieval that relies on Semantic Web technologies. We detail the architecture of the EASAIER system and the methodology for integrating existing musical data assets. So far, we have successfully integrated a traditional Scottish music archive. A mapping has been defined that allows the translation of editorial metadata from its original relational database scheme according to the Music Ontology. Furthermore, low-level feature extraction algorithms are applied to the audio files. Their results are described according to the Music Ontology, allowing meaningful queries over both editorial and acoustic metadata.