A combination of multi-objective genetic algorithm and deep learning for music harmony generation

Automatic Music Generation (AMG) has become an interesting research topic for many scientists in artificial intelligence, who are also interested the music industry. One of main challenges is that there no clear objective evaluation criterion can measure grammar, structural rules, and audience satisfaction. Also, original contains different elements should work together, such as melody, harmony, rhythm; but most previous works, works only one element (e.g., melody). Therefore, this paper, we propose a Multi-Objective Genetic Algorithm (MO-GA) to generate polyphonic pieces, considering grammar listener In method, use three functions. The first function accuracy generated piece, based on theory; other two functions modeled scores provided by experts ordinary listeners. scoring listeners separately using Bi-directional Long Short-Term Memory (Bi-LSTM) neural networks. proposed generation system tries maximize mentioned new piece music, including melody harmony. results show method pleasant pieces with desired styles lengths, along harmonic sounds follow grammar.