GERAÇÃO DE MÚSICA COM APRENDIZADO DE MÁQUINA

ABSTRACT – Machine learning is a concept that has been a part of the day-to-day life, being used in applications like social networks, e-commerce, smartphone assistants, among others. In the music area it can be used to inspire composers, produce music based on a specific style or generate music in real time for games or VR applications. This article will evaluate the ability of a neural network to generate satisfactory results in the music area, using Magenta libraries, a toolkit for the use of machine learning in artistic applications. The songs are generated based on a dataset of Bach compositions and procedurally checked for plagiarism, comparing the obtained results to the dataset.