By: Linda Mao RIG Inc Intern Researcher
Figure 1 Piano Staves
The emergence of advanced technology and artificial intelligence has made our lives easier and more productive. Virtual assistants, like Apple’s Siri, can easily accomplish tasks from setting up an alarm to navigating through thousands of search results. Even banks and hospitals utilize artificial intelligence to help serve their clients and patients in a more efficient and friendly way. Interestingly, the music industry seems to get involved in such trends as well.
From Chinese Yuan Qu to Baroque choral works, the passion and enthusiasm for music grew and flourished independently among distinctive continents. Even five-year-olds nowadays may be able to name a few famous and gifted composers, such as Beethoven or Mozart. Yet, the ability to find the most harmonious and dulcet voice among different sounds and instruments was considered a skill that only pertains to humans. However, in present society, the music industry has begun to rely on advanced technology to create new pieces.
In 1957, Lejaren Hiller and Leonard Isaacson advocated for the possibility of using an electronic computer in the process of creating music pieces (Alex Di Nunzio). At the University of Illinois at Urbana–Champaign, generated the first composition for the string quartet, named Illiac Suite, by an electronic computer. In preparation for this composition, they carried out a selection of diverse musical elements and styles and conducted a series of careful analyses and examinations. They also gathered and published all the details of the process in a book titled Experimental Music: Composition With an Electronic Computer. This accomplishment creates the foundation for many future investigations on the use of technology in the field of music composition.
In 2017, the songwriter Taryn Sothern released an entire album, I Am AI, with artificial intelligence platforms. She inputted music elements and selected beats and genres. As a result, the platform returned discontinuous music chunks, which she arranged into a complete composition. In the interview, she admitted that it was at first troublesome to maintain consistency between songs, yet she later found the uniqueness and benefit of such trouble because “it might lead to some pretty heavy whiplash for listeners.”
Figure 2 I AM AI
In the year of 2019, SKYGGE, a music group, combined the melody of Pete Seeger’s Black is the Color and a Bossa nova song with artificial intelligence tools to create a new version of the composition. The combined product sounds like an entirely new and singular piece according to the reviewers. Indeed, the advantage and specialty of AI is demonstrated by this album.
Currently, many open-sourced AI tools provide access to generating different music genres to the public. For instance, Google’s Magenta utilizes machine learning models to return sequences of notes based on inputs. It implements the music VAE (Variational Autoencoder), where all the original data is summarized and recreated by a decoder to construct new pieces. Features of the composition, including the number of new sequences and the randomness of the notes, can be altered with ease.
Google’s Magenta is not the sole music-making AI platform in the market now. Back in 2012, Sony CSL has launched the project Flow Machines as a research and development approach in the music industry. Flow Machines Professional, the developed AI-assisted music composing system, was a part of this project. It helps arrange notes and inspires users and artists in their process of creating art. In the lab, the researchers have created two entire pop songs with artificial intelligence, Daddy’s Car and The Ballad of Mr. Shadow. Daddy’s Car is a piece in the style of The Beatles and is arranged by the French composer Benoît Carré. The Ballad of Mr. Shadow is written in the styles of several American songwriters. Interestingly, the company emphasizes the implausibility of using the platform to generate the entire music piece and advocates for its use as a source of ideas and creativity (Alex Paretski).
Figure 3 Flow Machines
Although the appearance of artificial intelligence in the process of composing a melody may shock many people, one can observe that human composers still play a major role by selecting genres and arranging verses. Most software still relies on human’s aural skills and music talents to produce appealing pieces. Yet, we cannot deny that the AI system benefits the art world by contributing ways to fuse seemingly dissonant music styles together. This allows for more variation of the pieces and may introduce future trends and types of music. One day, Machine Learning artists may become a new career in our society, yet for now, its possibility to replace human composers is still minimal.