In a recent Inc. Magazine article, John Rampton touted the benefits of playing music, claiming of all things, that doing so helps your brain “more than any other activity.” According to various studies that scanned the brains of musicians and non-musicians alike, the corpus callosum—a large bundle of fibrous nerves connecting both sides of the brain, was in fact found to be larger in musicians. Specific to keyboardists, the category of musicians into which I happen to fall, areas of the brain involved in movement, hearing, and visuospatial abilities were also larger. Longitudinal studies determined that roughly fourteen months of musical training (not necessarily keyboard specific) had powerful and lasting structural and functional changes in the brains of children. Age at the start of musical training was identified as the most important factor in determining the extent to which the brain changes and the profound nature of those changes (Rampton, 2017).
As a classical pianist of thirty years, I am well aware of the benefits of music on a personal and social level. Music is an integral part of culture and is often seen as a source of national pride in some instances—many European composers are considered national treasures to the countries they resided in. Throughout my university studies, I read countless articles about the benefits of music, from simply listening to it, to studying it while learning an instrument and understanding the theory behind it. Music helps improve mood, reduce stress, alleviate anxiety, and provide comfort, among other various and sundry benefits (Heid, 2018). For example, many people are likely familiar with the “Mozart effect.” In a study conducted at the University of California, Irvine, college students were administered standard IQ tests following ten minutes of listening to Mozart, listening to a relaxation tape, or sitting in silence. The resulting test scores for the Mozart group were consistently higher on average than either of the other two groups (Harvard, 2011).
Music offers perhaps the most unique and transcendental experience of the arts and sciences. English poet and playwright Robert Browning once commented that “He who hears music feels his solitude peopled all at once.” Various studies document social isolation as being a rather marked cardiac risk factor. Some in the medical establishment even believe that social isolation is comparable to a near pack-a-day smoking habit (15 cigarettes per day) in terms of its toxic effect on overall health (Morin, 2018). Looks like Browning may have been onto something here. Shakespeare famously stated, “If music be the food of love, play on.” In a similar vein, Tolstoy once explained that “Music is the shorthand of emotion.” Such depth of insight conveys in a meaningful way advice similar to that we might hear from a psychologist or clinician about expressing emotions; music allows us to share the warmth of our humanity with others. Perhaps Plato said it best that “Music is a moral law. It gives soul to the universe, wings to the imagination, and charm and gaiety to life and everything else” (Harvard, 2011).
All of this is to say, in a passionate though rather roundabout way, that the prodigious effect of music in our lives can never be overstated. While I am undoubtedly grateful for every minute I spent practicing, performing, and otherwise occupying my time with the study of music, I realized after the economy bottomed out in 2008 that continuing to pursue a career in music—whether in performance or as faculty at a university—was no longer a financially viable option. Through much trial and error, mostly error, I discovered that I had a mild penchant for technology and data. This discovery in hand brings me very much to the point of this writing. Could I, through technological means, continue to enjoy the creative process of music making in combination with a stable career? Would it be possible to leverage my newly minted programming skills and understanding of artificial intelligence algorithms as an outlet for creative (and specifically musical) pursuits? Welcome to the wide and burgeoning world of computational creativity.
What is Computational Creativity?
Computational creativity refers to the “multidisciplinary endeavor that is located at the intersection of the fields of artificial intelligence, cognitive psychology, philosophy, and the arts” (Wikipedia, 2020). The field concerns itself with, on the conceptual side, crafting an official, comprehensive, and legitimate definition of creativity, while on the practical side, designing systems that are capable of autonomously exhibiting creativity or enhancing human creativity. Both sides work to inform the other. Despite its conceptual meaning, controversy still abounds as to what constitutes the illusive and virtually limitless beast that is creativity. The inability of the computational creativity community to settle upon a formal definition of creativity, one that is comprehensively satisfactory, has been described as “the elephant in the room (Jordanous, 2014).” Perhaps John McCarthy—American mathematician and computer scientist who is considered one of the founding fathers of artificial intelligence (Childs, 2011)—said it best,
To ascribe certain beliefs, knowledge, free will, intentions, consciousness, abilities or
wants to a machine or computer program is legitimate when such an ascription expresses the same information about the machine that it expresses about a person. It is useful when the ascription helps us understand the structure of the machine, its past or future behavior, or how to repair or improve it (Jordanous, 2014).
While McCarthy was speaking generally about artificial intelligence, it’s interesting to consider the notion that artificial creativity is both possible and valid, by imbuing a machine with, essentially, the exact elements that comprise the essence of a human being: beliefs, knowledge, intentions, and other abilities. In a 2017 Google AI paper titled “Attention is All You Need,” Google “introduced the concept of the transformer neural network architecture” for use in text data whereby stacks of encoders and decoders process sequences of arbitrary length and provide relevant outputs by “’attending’ to the right representation based on observed patterns in the training data.” Known as “self-attention”—the capacity for a machine to identify relationships within a sequence—the next step in computational creativity is to apply this neural net framework to “note generation in music, or pixel generation in images” (Rao, 2019). Is an understanding of computational creativity beginning to congeal? Let’s continue to explore this emerging field and look specifically at applications of computational creativity in music.
Computational Creativity in Music
Since the inception of computer music in the 1950s, primitive artificial intelligence algorithms have played an important role in the steady evolution of automated composition, improvisation, and performance of music (Lopez de Mantaras, 2016). The Illiac Suite—a string quartet composed by the Illiad computer via Markov chains (state dependent probabilities)—is one of the most well-known, earliest, pioneering works in computer music, created in 1958 by University of Illinois, Urbana-Champaign composers and professors Lejaren Hiller and Leonard Issacson (Wikipedia, 2020). While considered a remarkable accomplishment, the Illiac Suite was not particularly aurally appetizing as the random nature of Markovian processes simply cannot produce consistent melodic quality. In 1972, programmer, engineer, and mathematics savant James Moorer took a different approach to computational composition by simulating the human compositional process using “heuristic techniques” rather than “Markovian probability chains.” The result was a revelation, of sorts. Moorer discovered that the randomness of Markovian probability tended to “obscure rather than reveal the musical constraints needed to represent simple musical structures.” Using constraint-based techniques (heuristic techniques), Moorer was able to achieve more meaningful and tasteful melodies and underlying harmonies. He dubbed his programming constraints as “style templates” (Lopez de Mantaras, 2016).
The most successful application of artificial intelligence in the pursuit of computational creativity in music may be attributed to David Cope, an emeritus professor of music at University of California, Santa Cruz. In the early 1980’s, Cope was commissioned to write an opera (source of the commission is unknown). At that time in his career, he was suffering from what may only be described as a paralyzing case of “composer’s block.” Electing procrastination over forced productivity (or complete lack thereof), Cope began coding a music composition program called Experiments in Musical Intelligence (EMI) or “Emmy.” According to Cope, “Experiments in Musical Intelligence is an analysis program that uses its output to compose new examples of music in the styles of the music in its database without replicating any of those pieces exactly” (Garcia, 2015). In essence, Cope had created a computer program capable of identifying stylistic characteristics of various composers, taking a variety of their most recognizable characteristics and compositional devices, and create entirely new compositions whose style “matched” those of the composers they were based on. Coding EMI took roughly eight years. With EMI in hand, Cope wrote and delivered his opera in two days (Garcia, 2015).
Though EMI was a remarkable achievement in its own right, I couldn’t help but wonder…were the compositions any good? Since computational creativity is all about discovering whether or not machines are capable of human level creativity, how would EMI’s works stack up against the works of the composers that EMI was programmed to emulate? Like any good sceptic, Dr. Douglas Hofstadter—distinguished professor of cognitive science at Indiana University, Bloomington, published author, and recipient of the 1980 Pulitzer Prize for general nonfiction for his first book called Gödel, Escher, Bach: An Eternal Golden Braid (Somers, 2013)—organized a musical version of a Turing test. Winifred Kerner, pianist and instructor at the University of Oregon School of Music, was enlisted to perform three works: two in the style of J.S. Bach (of which one was composed by EMI and the other by a distinguished University of Oregon professor of music), and an actual work of Bach. Following the performance, the audience was asked to match each piece with its composer. Overwhelmingly, the audience mistakenly thought the work created by EMI was actually Bach. In an interview with the New York Times, Hofstadter remarked that “EMI forces us to look at great works of art and wonder where they came from and how deep they really are. Nothing I’ve seen in artificial intelligence has done this so well (Garcia, 2015).”
Looking to the Future
As we look to the future, it’s worth pointing out that algorithmic composition is no longer strictly a product of computational scientists, researchers, and various and sundry academics. In 2016, Sony’s AI research team released an algorithmically composed song called “Daddy’s Car” in the style of the Beatles. Google has also been exploring the computational creativity space with its project “Magenta,” which is all about creating music and art through machine learning. In addition to the behemoths that have taken an interest in this field, there are also two European startups that have entered the fray, Aiva and Jukedeck. Both companies create music using algorithms; Aiva uses human musicians to record the AI compositions while Jukedeck’s platform allows users to create their own tracks based on input parameters (genre, mood, etc.). The Jukedeck platform is particularly appealing for content creators such as podcasters or YouTubers in need of music for their media creations. Aiva has branded themselves as customizable AI composers of “memorable and emotional soundtracks” to enhance the storytelling of movies, commercials, games, and trailers (Euler, 2017).
The exciting reality for a guy like me, and others like me, is that it turns out there are opportunities to engage in a creative capacity to produce works of “art” through programming. One of the most arduous career paths a person can choose is to dedicate one’s life to a career as a musician (or frankly, any of the performing arts). Through computational creativity, musicians, artists, and various other creatives among us have a unique opportunity to acquire a new skillset in programming, develop an understanding of and appreciation for artificial intelligence, and remain connected to our passion for creative pursuits. Seemingly gone are the days of either/or. Perhaps we really can have our cake and eat it too.
Childs, Martin. (2011, November 1). John McCarthy: Computer scientist known as the father of AI.
Computational Creativity. (2020, March 3). Wikipedia. Retrieved April 15, 2020, from
Euler, Thomas. (2017, June 28). The Current State of Creative AI: Artificial Intelligence in Music, Writing,
and Movies. attention.econo.me. https://attentionecono.me/creative-artificial-intelligence-music-writing-video-movies-2017-7e9c192594a3
Garcia, Chris. (2015, April 29). Algorithmic Music – David Cope and EMI. Computer History Museum.
Harvard Men’s Health Watch. (2011, July). Music and Health. Harvard Health Publishing.
Heid, Markham. (2018, April 26). You Asked: Is Listening to Music Goof For Your Health?. Time.
ILLIAC. (2020, January 20). Wikipedia. Retrieved March 26, 2020, from
Illiac Suite. (2020, January 21). Wikipedia. Retrieved March 26, 2020, from
Jordanous, Anna. (2014, April 10). What is Computational Creativity? The Creativity Post.
Lopez de Mantaras, Ramon. (2016). Artificial Intelligence and the Arts: Toward Computational Creativity.
Markov Chain. (2020, March 25). Wikipedia. Retrieved March 26, 2020, from
Morin, Amy. (2018, June 18). Loneliness is as Lethal as Smoking 15 Cigarettes Per Day. Here’s What You
Rampton, John. (2017, August 21). The Benefits of Playing Music Help Your Brain More Than Any Other
Rao, Prashanth. (2019, March 5). Computational Creativity: The Role of the Transformer. Medium.
Somers, James. (2013, November). The Man Who Would Teach Machines to Think. The Atlantic.