2P-07
Generating Homophonic Music with LSTMs Dedicated to Melody and Harmony
The goal of this research is to create a Neural Network model capable of generating music without any user input. The generated music should be “meaningful”, that is to say,
it should sound like it has a purpose.
Throughout the years, many attempts have been made at creating music procedurally. However, very few of those attempts managed to create music with “meaning”.
By first generating melodies with meaning, then chord progressions to fit a generated melody, as well as using intervals relative to the key of the piece for data representation, the model introduced in this research aims to generate music that is comparable to that of a human composer.
The generated music is homophonic, consisting of a melody accompanied by block chords. The output is in the MIDI format.
it should sound like it has a purpose.
Throughout the years, many attempts have been made at creating music procedurally. However, very few of those attempts managed to create music with “meaning”.
By first generating melodies with meaning, then chord progressions to fit a generated melody, as well as using intervals relative to the key of the piece for data representation, the model introduced in this research aims to generate music that is comparable to that of a human composer.
The generated music is homophonic, consisting of a melody accompanied by block chords. The output is in the MIDI format.