Computational tools for knowledge-driven music browsing Gopala - PowerPoint PPT Presentation

Computational tools for knowledge-driven music browsing Gopala Krishna Koduri, Xavier Serra {gopala.koduri, xavier.serra}@upf.edu Music Technology Group Universitat Pompeu Fabra Barcelona, Spain

Part - I THE CURRENT STATE OF THE AFFAIRS

Data Sources

Data Sources

Data Structuring

Data Structuring

Data Structuring

Data Structuring

Entities • Also geographical regions, lineage etc…

How do we use this data?

Browsing the collections

Similarity measures for exploration

Application Programming Interface

Part - II WORK IN PROGRESS

Limitations: Disconnected sources

Limitations: Simplistic similarity

Next steps • Linking data sources – More insights! – Provenance and Trust • Machine-readable descriptions

Linked data example: Facebook

Linked data example: Google

How do they do it?

Machine readable descriptions

Machine readable descriptions • Definition • Classification • Association • … Semantics

Semantics of musical concepts: raaga

Semantics of musical concepts: raaga

Knowledge from community data Raaga Relation between raagas Musical form

What do all these entail? • List all the performances of Bhairavi and it’s allied raagas, of artists from Semmangudi’s lineage, at the music academy. • What are the distinguishing phrases of Pantuvarali raaga in the performances of artists from X and Y regions?

The ultimate goal Data gathering A data repository with varied sources Data structuring • MusicBrainz, Wikipedia, Kutcheris.com Musicological • Outputs from audio analysis validation • Semantic descriptions of musical concepts • Knowledge extracted from user generated data User pro fj ling Audio analysis Music exploration

THANK YOU!