Comparative Study of Genetic Algorithm and Ant Colony Optimization Algorithm Performances for the Task of Guitar Tablature Transcription

The problem of guitar tablature transcription is the conversion of a song in standard music notation (music sheet) to an alternative notation known as guitar tablature or tab. A guitar tablature consists of indicating each string and fret of the guitar needs to be played to produce a particular note. However, considering that each note can be played in different positions of the guitar, this conversion is not a straightforward process. In this paper we address the problem by categorizing it as an optimization problem, as not only we want to generate a playable guitar tablature, as we also want to make the guitar tablature easier to play. For these reasons, in this paper we present two novel evolutionary approaches for this task. The proposed approaches are based on the Genetic Algorithms with subpopulations and the Ant Colony Optimization algorithms. Our experimental results with a novel dataset of 148 songs show that the Ant Colony Optimization approach produced the best results for this task.