On June 3, the seniors in the music industry program here at Drexel University presented their senior capstone projects that they had been working on for eight months in Studio One at One Drexel Plaza. There were various types of projects, some based in audio production, some business plans, some research based and some that were a combination of the three.
One of these hybrid projects was a research paper and EP called “Artificial Feelings” done by students Erin Hession and Matt Kopyt. Hession and Kopyt used their project to explore the growing applications of artificial intelligence in the production of music. The result was a five song EP that will be released June 7.
Both Hession and Kopyt are in the business concentration of the music industry program and, therefore, have only received minimal training in music production and theory. They were inspired to explore how new artificial intelligence programs can be utilized by musical amateurs to create a commercial ready project.
“Erin and I aren’t very technically trained musicians,” stated Kopyt in the vlog series the two created to give a behind the scenes look at the project “We both have some type of musical inclination, but neither of us have been trained in a lot of music theory, but artificial intelligence will give us the ability to allow the computer to make a lot of those higher level decisions.”
The idea behind the project was to examine how the barriers for entry into the music industry could be drastically lowered with future improvements in these technologies. They also researched the history of artificial intelligence being used by musicians.
Throughout the production process of their EP, the two used open source scripts created by Google Magenta and Amper Music. These are two of many programs in beta testing available for use by the general public online.
Google Magenta is a suite of 13 artificial intelligence applications that can be used to make different aspects of a song. In their project, Hession and Kopyt utilized the Multitrack Chords application.
Amper Music is a program that allows users to set multiple parameters like mood, instrumentation, tempo and key to create a more customizable track that fits the creator’s vision. It was originally developed by composers in Hollywood to accelerate the soundtrack process.
Hession and Kopyt used these programs to create the chord progressions and melodies that their songs were built around. From here, they made arrangements in Logic from the Musical Instrument Digital Interface (MIDI) files exported from the artificial intelligence programs.
Originally, the duo had expected artificial intelligence to serve a much bigger role in the creation of the album, but as they did further research and started working with the programs, they realized that the programs readily available to the public have limited outputs.
“We had set out to find programs that could compose an entire piece of work on its own with parameters set by us. It became apparent very early on that the technology necessary to achieve that goal is not available at this time,” the two wrote in the research portion of their project.
Hession and Kopyt composed vocal melodies and lyrics but relied on artificial intelligence for their initial song structures. The two composed and produced their songs over the course of winter term and mixed and mastered them throughout the spring.
During their presentation, Hession and Kopyt played samples from the EP and the initial file that was given to them by Google Magenta or Amper. The final product is an impressive collection of electronic R&B songs.
“I am extremely happy with the end product. Making music has always been a goal of mine. This project was the perfect opportunity to combine my goals with technology that is impacting the industry. There are very few artists exploring AI,” Hession stated about their EP.
Be sure to listen to the “Artificial Feelings” EP June 7 on all streaming platforms.