Jeffrey Cruz


Advisor: Professor David Paul Benjamin Major: Computer Science

Document Type



Artificial intelligence is one of the fastest growing fields at the moment in Computer Science. This is mainly due to the recent advances in machine learning and deep learning algorithms. As a result of these advances, deep learning has been used extensively in applications related to computerized audio/music generation. The main body of this thesis is an experiment. This experiment was based on a similar experiment done by Mike Kayser of Stanford University in 2013 for his thesis “Generative Models of Music” where he used Hidden Markov Models and tested the quality/accuracy of the music he generated using a music composer classifier. The experiment involves creating Markov models for music generation and then creating new models that use deep learning algorithms. These models were trained on midi files of piano music from various composers and were used to generate new music in a similar style to the composer it was trained on. In order to compare the results of these models quantitatively, the music generated by these models was passed to a classifier in order to see which technique create a model that makes music that is correctly classified as being from the composer they were trained on. The results of this experiment showed that the classifier was able to more accurately label music generated by the deep learning model than the Markov model as being from the composer the model was trained on.