Machine Learning project to generate classical music based on previous scores using Tensorflow.
This project is my Research Work on Artificial Intelligence. The goal of the work was to learn about how Artificial Intelligence works and to program a Machine Learning algorithm able to compose music.
The work is divided in two parts: the Theoretical Frame with my Research about Machine Learning and the Practical Frame which is the code of the ML algorithm programmed in a Jupyter Notebook with explanations about what the code does. It implements a Long-Short Term Memory Recurrent Neural Network using Keras and Tensorflow and preprocesses the data of the scores (found in Data) using music21, Numpy and Pandas.
The full work can be found under the title "Algorithmic Improvisation".
This project has been designed to work with Tensorflow 1.10, so there may be problems if the code is run over another version of Tensorflow.