Automatic Generation of Melodic and Chordal Chromaticism

Western popular and art music use the diatonic scale as their foundation in both melody and harmony, and restricting to diatonic scales can simplify both analysis and composition. However, even mostly diatonic genres contain some limited use of chromaticism or accidentals. Therefore I created this pipeline that, after creating diatonic music, auto- matically alters selected notes chromatically to enhance the melody and harmony. This project presents several deep learning models for generating this chromaticism given a purely diatonic input melody and accompanying chord sequence. The altered melody and harmony from the model are generally compelling and conform to musical syntax, making it a work a potentially useful tool for both human and computer agents to augment existing compositions or aid in the compositional process.

Diatonic input Example:

Chromatized Ouput Example:

Check the code out here: https://github.com/antchristou/Automatic-Creation-Of-Chordal-and-Melodic-Chromaticism