1 Computers vs. Composers: A Modern War on Music

About the Author

Rebecca Baker is from Colorado Springs, Colorado, and Seattle, Washington. She is a freshman studying Music with an emphasis in Composition as well as Piano Pedagogy. She loves to do yoga and hike.

This essay was composed in May 2022 and uses MLA documentation.

SOME OF US ATTEND CONCERTS, listen to music as we walk to class, or curate the perfect playlist to listen to in the shower. Others jam to the radio in their car, walk to the music playing in the background of their weekly grocery shopping trip, or stumble upon snippets of songs in stories on Instagram. Music is undeniably all around us, and the influence music has had over the past few centuries on the average person’s life has dramatically changed. During Europe’s Baroque period (1600-1750), the middle class would hear music on Sundays during church worship. Eventually, the middle class would become more involved in public performance closer to the Classical period (1730-1820). Since then, music has become widely available due to radio, cassette tapes, CDs, and now streaming services right to a smartphone.

In antiquity, music was recited orally, and by 1400, it was recorded on tablets. Handwritten music notation and copying became popular in the 17th century. Printed sheet music became popular in the 19th century, and sheet music became digitized in the late 20th century. The availability of notated music increased heavily as technology advanced. However, since the Renaissance, composers and creators have taken different approaches to using algorithms to generate music. Some composers have used mathematical equations to develop a mathematically tonal atmosphere within the music. For example, Bach (1685-1750) used equations to create his three-part fugues, and Xenakis (1922-2001) used computer computation to create his music. Technology has revolutionized how composers generate music using algorithms by making a tedious process more efficient.

Thousands of different genres of music, millions of music creators, and billions of people who listen to music lead to billions of unique music tastes. However, nearly every listener seeks music that fulfills an emotional connection and expressive reward. Additionally, every song must fill a functional and aesthetic purpose. Not only can someone connect to the music being performed, but one may also connect with the performer and fellow listeners. Because of this, music can fulfill the human need for socialization.

Due to the advancement in technology, the composition and creative processes have been reformed. Technology assists composers in notating and generating music at a more efficient rate. However, with the implementation of this technology in major musical processes, music is at risk of losing its ability to connect people socially and diminishing the creative process. Despite this risk, technology can help a composer thrive in the environment by making it more efficient and forcing more creativity. From these points, the question arises, does music lose its meaning and connection to humanity when technology and computer generation are implemented in major musical processes?

Prominent composers have used algorithmic composition in the past in different ways, but they did the computing by hand as technology had not been an aide. In 1792, Mozart published an algorithmic piece labeled K. 294dK3. This piece was performed using numbered fragments with dice to roll the fragments randomly. Music written in the Renaissance often followed a strictly rule-based system, using isorhythms, which use a repeating rhythmic pattern throughout the piece, as well as using canon, which is generated in a round, such as in the children’s song “Row, Row, Row, Your Boat” (Povilioniene 4). Bach’s three-voice fugues follow a strict structure and system of rules. The subject is introduced in a soprano voice and then played in the alto, tenor, and bass. While the main subject is being played in any of the voices, the others play a contrasting and independent melody called a counterpoint. Additionally, notes moving in voices must be portrayed as independent by avoiding parallel fifths and octaves in the voices, meaning two parts may not mirror each other. Bach uses a system of rules so strict it is mathematical, like an equation to generate an entire piece using a mere two bars of music as a subject (Povilioniene 4). These examples through the Renaissance, Baroque, and Classical periods demonstrate that composers have been using algorithms written by hand for centuries.

Different models have been used to generate music following Bach’s strict set of rules, making it sound like Bach composed the music himself. Some programs require the composer to create a subject and the rest of the voices. Then, the computer program filters out the “unsatisfactory” material, after which the composer corrects and refines the piece. Another program is given the subject and generates the rest of the voices according to the rules. More revolutionary programs can treat Bach’s rules as a mathematical equation to be solved. One program, EcoComposer, considers the inputted voice, melodic content, chords, harmonization, independent voices, classical voice leading rules, and aesthetic fulfillment, such as the balance between repetition and variation, through intense computer programming to create a perfect solution to the “4-voice” problem. EvoComposer’s creativity and ability to compose allow the program to mimic Bach (Prisco).

Because a computer program can mimic a composer so perfectly if given the necessary attention to detail, the question lingers if the computer may overtake the role of the composer. Frequently, music reflects the raw emotions of a composer and humanity in general. The music reflects this emotion by creating tension through imperfection, such as a note being played on the wrong beat, a note not fitting the chord, or the piece generally sounding atonal, random, and dissonant. The computer program-generated music is at risk of losing its creativity and passion if always generated by numbers and perfect mathematical equations. Additionally, with computer programs becoming more accessible and better funded, computer-generated music will be cheaper to produce compared to paying a composer to create their music. Due to money, the ability to mimic composers, and perfect computation, music risks losing its meaning and connection to humanity if technology and computer generation are implemented in major musical processes. Additionally, the use of technology may diminish the role of the composer or music producer and, in a more extreme case, eliminate them.

Xenakis is an excellent example of modern composers using algorithms, specifically computer programming, to generate music. Xenakis describes a process in which he uses computation to analyze the most “successful” parts of the music he creates that generate an expressive reward and emotional connection, thus revolutionizing the creative process to be more efficient and generally have more quality (Hamman 120). To achieve this effect, Xenakis would analyze the similarities between the music passages that people seemed to enjoy the most. He was also an engineer and appreciated mathematics. Today, many associate math with being highly detailed and structured, leaving no room for creativity. However, Xenakis saw the bridge between geometry and art/music as an opportunity for enthusiasts to appreciate its validity and beauty (Hamman 120). He believed mathematics could find the perfect balance between settling on an idea and being able to elaborate and create, just as a composer would typically consider. Xenakis claimed that using artificial intelligence to compose the more tedious algorithmic fragments of his works allowed him to flourish in creativity rather than having to compute everything himself (Hamman 121). Artificial intelligence and computing can become the base from which composers may flourish in aesthetics and creativity.

Using computers and artificial intelligence has revolutionized the connection between art and mathematics. Additionally, composers have reached a “highly mathematized tonal space” (Povilioniene 16). For instance, Bach’s canon from Musikalisches Offer, BWV 1019 is highly similar to a Möbius strip found in mathematics. The equation and the canon are similar because they are both, in essence, palindromes. Composer Gary Lee Nelson (1940-present) uses fractal algorithms to create minimalist three-voice compositions. Some composers have used equations and fractions to shift metric accents and thus the beat and sense of pulse in the piece of music. All of these samples are composers using mathematics to generate a creative solution to a proposed problem in the music.

Mozart, Bach, and Xenakis each use algorithms to generate a piece of music. These processes were initially done by hand, and in Xenakis’s time, by computer to increase efficiency. Generated music is not much different from humans using algorithms anyways, as computers merely speed up the process of computing. Even in Xenakis’s case, who used some artificial intelligence, he could calculate the mathematics of the AI by hand. In considering this, computer-generated music is just what a composer could do by hand, just revolutionized to be more efficient.

Computing the logistics of a piece, such as form, chords, and harmony, eliminates a painstakingly long part of the composition process. For example, if a computer lays a piece’s foundation, a composer has the opportunity and time to explore melody. Alternatively, if the computer generates a melody, a composer has the time to spare to explore chords and harmonies. This framework applies to all aspects of creating a piece without considering performance interpretation and cues. Because the computer offers the composer a strict framework, they must be more creative in completing the other aspects of the piece and inserting their interpretation and voice into the work. Laying out different tedious elements of the creative process makes it more efficient, allowing composers to thrive in their creativity.

While a computer program can allow a composer to compose creative pieces more efficiently, the question remains if the computer can take over the role of the composer. Returning to EvoComposer’s ability to mimic Bach offers a primary focus for such a discussion. This program is strictly modeled after an existing composer with an established voice and style. Most other computing and algorithmic methods supplement a composer during their creative process. Thus, the software is only successful because it is copying a clearly existing style, so the software is nearly an extension of Bach himself. Now, most composers have a voice and a style, but they are not nearly as strict as Bach’s was, so it is safe to assume that modern composers will not be replaced by a computer that can mimic them. Additionally, the program follows a set of rules and guidelines that a composer already followed. The algorithmic program is not likely to be better than a human composer because that composer must program it according to their existing ideas and processes. The algorithm could never produce a piece that is better than what the person creating the algorithm could. However, it could produce more efficiently, so it can be a powerful tool in helping a composer who can describe their composition process in algorithms. The creative process is reformed with advancing technology to help composers become more efficient and innovative in their music.

Xenakis described using algorithms and mathematics in his music because he wanted to understand music on a more fundamental level, such as why certain frequency combinations may be pleasing to the ear. While most performers study the technique of their instrument for years to produce a beautiful, colorful, and expressive sound, Xenakis sought to understand the technique of music and, further, sound itself (Hamman 120). He began to search for the truth in the beauty of music. Michael Hamman explains, “Xenakis sought to engineer a collision between the centripetal force of sonic matter and the centrifugal force of human meditation” (Hamman 120). Essentially, this means he strived to take the material, primarily mathematical, and to make it matter, an actualization and manifestation of the concept through sound. Xenakis describes his creative process by using artificial intelligence, stating it would not be possible to formulate these highly detailed compositions without the help of competition because one would be “bogged” down with quantifying the data by oneself (Hamman 120). Hamman describes the process:

Programming projects self-referential activity into an all-referential realm, allowing humans to clearly realize (or perhaps knowingly reject) otherwise invisible qualities of their own thinking and activity (122).

He can hone his creativity because artificial intelligence takes care of the tedious parts of computation in algorithmic composition. Returning to the question, does his music still generate an emotional connection and reward? Xenakis’s music may still generate connection and reward as he was passionate about achieving the highly mathematical tonal space and studying the bare foundation of sound in his composition. Xenakis recognized the benefits of advancing technology in his creative process, and none of his highly lauded work would be possible without computer generation.

In the past, when music was not recorded, most performances were live and very social events. Since then, music has not lost its social aspect. For example, music is often played in the background at a dinner party, or attending big concerts and festivals is still widely popular. In the past, some participation in the production of music occurred, such as requesting a theme to be improvised or playing duets together on the piano. In modern times, the listener participates by choosing the volume or order of songs when listening. In 2007, De Jaegher H and Di Paolo E researched the connection between music and the listener (Maes 81). They found that music provides the most influence and connection when listeners use their senses and interact socially. An experiment, later called “Soundbikes,” was created based on this concept in which two people cycled together. Computer-generated music was produced and played in real-time, depending on their biking patterns. This experiment demonstrated the importance of music affecting multiple senses, such as the physical exertion of cycling and producing music (Maes 91). The Soundbikes experiment also demonstrated the social aspect of participating in music with others, despite the music being created by artificial intelligence. Overall, this study shows that even computer-generated music can have meaning and fulfill a person’s need for the social enjoyment of music just as well as a composer can. However, different steps may need to be considered to achieve this.

China is experiencing a need for more composers, but not enough people to fill the demand for music. There has been a higher demand due to the fast production of media and content, such as video games, animation, and video platforms which all require original music (Wang 1). A rising issue is that the high demand for music means more commissions for composers, so the production cost goes up, and only the wealthier companies can afford original music. Because of this issue, a Neural Network algorithm was created to analyze generated music that used artificial intelligence for quality control (Wang 2). This program has the potential to lighten the load that composers bear, such as relieving the tedious work so they may work more creatively, also making the process more efficient. It also can allow non-composers to create their own music more scientifically to ensure the integrity and quality of the music they are producing. This model goes into extreme depth and calculations, so further research must be done to make the program more efficient and genuine. This research ensures that computer-generated music can mimic the emotions composers want a listener to feel, so there will be no loss in quality or intention behind the music.

We have discussed four significant points in considering if music loses its meaning and connection to humanity when technology and computer generation are implemented in major musical processes. One, algorithmic composition has already existed for centuries, and computer programming merely speeds up the process. Two, algorithmic composition and computer generation can allow composers to thrive in their creativity. Third, artificial intelligence can make the process more efficient, thus allowing for greater quality and quantity of music to be produced. Fourth, computer generation can still fulfill an emotional and social connection, leading to no loss of intention behind a song. In these discoveries, it is also recognized that composition will become more accessible to everyone, and more voices and perspectives may be heard. With artificial intelligence, a composer’s job is more efficient so that a composer may be less overworked and quality of life and music may increase. In considering the effects of advancements in technology on the human race, Andrew Feenberg says, “We should engender for ourselves a nation of technology that empowers a broader range of human participation, in terms of design, philosophy, education, and application.” While most people will continue to listen to their favorite popular songs, computer generation has the potential to affect popular genres of music as well. Advancing technology and computer generation has the potential for increasing meaning and connection to humanity when implemented in major creative musical processes.

Works Cited

Hamman, Michael. “On Technology and Art: Xenakis at Work.” Journal of New Music Research, vol. 33, no. 2, 2004, pp. 115–23. Crossref, https://doi.org/10.1080/0929821042000310595.

Maes, Pieter-Jan, et al. “Embodied, Participatory Sense-Making in Digitally-Augmented Music Practices: Theoretical Principles and the Artistic Case ‘SoundBikes.’” Critical Arts, vol. 32, no. 3, 2018, pp. 77–94. Crossref, https://doi.org/10.1080/02560046.2018.1447594.

Povilioniene, Rima. “Definition Problem of Algorithmic Music Composition. Re-Evaluation of the Concepts and Technological Approach.” Musicology & Cultural Science, vol. 16, no. 2, 2017, pp. 3–20. Ebscohost, dist.lib.usu.edu/login?url=https://search.ebscohost.com/login.aspx?direct=true&db=asn&AN=129574633&site=ehost-live.

Prisco, R. de, et al. “EvoComposer: An Evolutionary Algorithm for 4-Voice Music Compositions.” Evolutionary Computation, vol. 28, no. 3, 2020, pp. 489–530. Crossref, https://doi.org/10.1162/evco_a_00265.

Wang, Yu. “Music Composition and Emotion Recognition Using Big Data Technology and Neural Network Algorithm.” Computational Intelligence and Neuroscience, edited by Bai Yuan Ding, vol. 2021, 2021, pp. 1–11. Crossref, https://doi.org/10.1155/2021/5398922.