Music Information Retrieval

Music Information Retrieval

Music Information Retrieval represents an interdisciplinary field of research, which exists already since 1960. It has been established with the incremental distribution of the world wide web and corresponding digital formats like MP3 . One of the core problems is to enable efficient access to large music collections.

In contrast to traditional text based information retrieval, music retrieval requires the consideration of numerous different representations on both parts, providers (databases) and consumers (digital audio: Wav,MP3, etc.; symbolic formats: MIDI, Score Notation, etc.; lyrics: ASCII; ontologies: RDF(S), OWL).

Our approach aims to develop a holistic system, which is capable to serve any mode and to activate numerous, distributed multimodal knowledge sources. In doing so, we focus on the exact and partial search for particular genres, songs and musicians as well as on a reference service based on similarities in compositions.

Our research also entails the sociocultural compatibility of such MIR-systems by looking at music as a subject matter and a result of cultural processes which – from the creator to the customer - consider all concerned. Thereby we examine in particular the effects and consequences of “virtuality”.

Our interdisciplinary approach entails

- digital signal processing
- information retrieval
- pattern recognition, machine learning
- knowledge representation
- web mining, semantic web
- human computer interaction
- cultural interfaces