Machine Learning Mandolin

Musicians combine their knowledge with intent to compose new musical pieces. Artists are endlessly creating more music, even though instruments have a limited set of sounds. Now, computer programs like Google’s Magenta project use machine learning to assist musicians in creating new songs. This is achieved by exposing a program to large amounts of music, and having the program learn patterns on its own rather than needing prior knowledge about music. The scope of this project will include researching how machine learning is applied to music generation and using a model to achieve this. Machine learning has been applied to music generation in many different styles, from classical piano, to jazz, and to video game music. This project’s focus will be on mandolin music, specifically Vietnamese folk music. The objective is to input audio files into a program that after training on the data, will generate new music using the patterns that were detected