Algoritmisk mønsteridentifikation: Nogle betragtninger omkring computeranvendelse i musikanalytisk øjemed

In the field of computer-assisted music analysis, similarity studies are widely recognized as the main areas of investigation comprising a number of computational tasks such as extraction, recognition, and comparison of musical patterns. A ‘musical pattern’ refers to any kind of recurring material (melodic, rhythmic, harmonic etc.), which is measurable through symbolic modelling. In this article, several issues associated with automated (algorithmic) procedures of musical analysis are introduced and discussed. First, preliminary historical and epistemological conditions of computational musicology are presented, followed by a few basic distinctions between different methods of application. Second, two central problems are evaluated: 1) the difficult choice of the form of symbolic representation, and 2) the challenge of finding musically adequate methods of comparison.