This paper presents a statistical analysis of the abc music notation corpus. The corpus contains around 435,000 transcriptions of which just over 400,000 are folk and traditional music. There is significant duplication within the corpus and so a large part of the paper discusses methods to assess the level of duplication and the analysis then indicates a headline figure of over 165,000 distinct folk and traditional melodies. The paper also describes TuneGraph, an online, interactive user interface for exploring tune variants, based on visualising the proximity graph of the underlying melodies.
展开▼
