Analysis by compression: Automatic generation of compact geometric encodings of musical objects

Research output: Contribution to conference without publisher/journalPaper without publisher/journalResearchpeer-review

191 Downloads (Pure)


MEL is a geometric music encoding language designed to allow for musical objects to be encoded parsimoniously as sets of points in pitch-time space, generated by performing geometric transformations on component patterns. MEL has been implemented in Java and coupled with the SIATEC pattern discovery algorithm to allow for compact encodings to be generated automatically from in extenso note lists. The MEL-SIATEC system is founded on the belief that music analysis and music perception can be modelled as the compression of in extenso descriptions of musical objects.
Original languageEnglish
Publication date2013
Publication statusPublished - 2013
EventThe Music Encoding Conference 2013: Concepts, Methods, Editions - Mainz Academy for Literature and Sciences, Mainz, Germany
Duration: 22 May 201324 May 2013


ConferenceThe Music Encoding Conference 2013
LocationMainz Academy for Literature and Sciences


  • music analysis
  • music information retrieval
  • music encoding
  • computational geometry
  • data compression

Cite this