Grammatical Evolution with Zipf's Law Based Fitness for Melodic Composition

Created by W.Langdon from gp-bibliography.bib Revision:1.4524

  author =       "Roisin Loughran and James McDermott and 
                 Michael O'Neill",
  title =        "Grammatical Evolution with Zipf's Law Based Fitness
                 for Melodic Composition",
  booktitle =    "SMC 2015 The 12th Sound and Music Computing
  year =         "2015",
  address =      "Maynooth, Ireland",
  editor =       "Joseph Timoney",
  month =        "26 " # jul # " - 1 " # aug,
  organisation = "Music Technology Group in the Departments of Music and
                 Computer Science at Maynooth University, Ireland",
  keywords =     "genetic algorithms, genetic programming, grammatical
                 evolution, Zipf distribution, music",
  URL =          "",
  size =         "8 pages",
  abstract =     "We present a novel method of composing piano pieces
                 with Grammatical Evolution. A grammar is designed to
                 define a search space for melodies consisting of notes,
                 chords, turns and arpeggios. This space is searched
                 using a fitness function based on the calculation of
                 the Zipfs distribution of a number of pitch and
                 duration attributes within the given melodies. In this
                 way, we can create melodies without specifying a key or
                 time signature. We can then create simple accompanying
                 bass parts to repeat under the melody. This bass part
                 is evolved using a grammar created from the evolved
                 treble line with a fitness based on Zipf's distribution
                 of the harmonic relationship between the treble and
                 bass parts. From an analysis of the system we conclude
                 that the designed grammar and the construction of the
                 compositions from the final population of melodies is
                 more influential on the musicality of the resultant
                 compositions than the use of the Zipf's metrics.",
  notes =        "",

Genetic Programming entries for Roisin Loughran James McDermott Michael O'Neill