The piano prelude begins insistently, with a loud, rhythmic figure repeated immediately at a lower register. The music winds up and down the piano, accents increasing, until a high note hangs in the air. A slower, meditative passage begins, pleading. The end is abrupt; the music slows down and fades, too quickly, into nothingness. What was the composer trying to convey through this curious piece?
As it turns out, nothing at all. That’s because the composer is a computer program named Emily Howell, who released “her” first CD in May. Her creator, composer-programmer David Cope, has stirred up the music world and stunned audiences with his music-making software. But the independent-minded Cope doesn’t seem to take notice of his critics. “I follow where my nose takes me and where logic suggests I should go,” says the 68-year-old Cope, professor emeritus at the University of California at Santa Cruz. “I don’t look to the outside world for verification that what I’m doing is the right thing.”
Cope works in the field of algorithmic composition, the use of formal rules to create music. Algorithmic composition usually involves computer programs, which can create music by using chance procedures or programmed rules, by analyzing pieces and generating their own rules, or by “learning” to make good music through user feedback.
In 1980, Cope was suffering from composer’s block and turned to his computer for relief. He had learned to program in the 1970s while doing research for his book New Music Composition, but he still doubted that a computer could create worthwhile music. Instead, he figured that it could act as a foil—“I would say, ‘I wouldn’t do that, I would do this instead,’” explains Cope.
This endeavor evolved into Cope’s first major project, Experiments in Musical Intelligence (EMI, pronounced “Emmy”). He originally wanted a program that understood his own personal style and could produce musical passages when he had another composer’s block. Unable to pin down his own style, however, he opted to create a program that produced Bach chorales.
EMI makes music using three basic steps. First, after Cope inputs pieces by a particular composer, they are deconstructed. EMI separates the music into parts (for example, groupings of one beat long) and analyzes their pitch, rhythm, harmony, and other musical features, then records what kind of groupings follow them in the piece. Next, EMI searches the inputted music for “signatures,” short passages that recur in different works and characterize the composer’s style. Finally, EMI recombines these groupings and signatures into a new piece in the style of the original composer. From 1981 to 2003, EMI produced thousands of pieces modeled on 35 composers, including Bach, Chopin, and Cope himself.
During the 1990s, Cope embarked on a new project named Emily Howell. This time, he wanted to produce new music, not imitate the style of famous composers. To compose with Emily Howell, Cope inputs objects (such as notes or chords) into the program, which at first assigns random weights to whether one object will be followed by any of the other inputted objects. This means that the program initially produces gibberish. But Cope uses a “carrots and sticks” method to teach Emily Howell to produce output he likes; when he says “no” to an output, the weights that produced it decrease; when he says “yes,” they increase.
Emily Howell’s input consists solely of music from EMI, so Cope calls her a “second-generation” program. The style he is cultivating, however, is not classical but contemporary. “I’m often pleasantly surprised by its output,” says Cope. “It produces music in styles that I’m not familiar with, and if I keep saying yes, it slowly begins to acquire a personal style.”
“She—quote unquote—is not afraid to use dissonant or harsh chords,” he adds.
In the case of EMI, Cope listed himself “with EMI” as the composer. But he has chosen a different approach with Emily Howell, listing her as the composer of the album From Darkness, Light and simply noting on the back cover that she is a software program he created.
“I treat myself as Emily’s audience. Every ‘yes’ I put into the machine, I think of as applause for what she’s done, and every ‘no’ I think of as boos,” Cope explains. “In other words, I consider myself a sidekick.”
To justify this attitude, Cope compares Emily Howell to a real-life composer getting boos from an audience, and subsequently altering his musical style. When asked whether he should take more credit as the creator of the software, he retorts, “My mother and father created me. Should they be getting credit for what I do?”
"I am not sad. I am not happy. I am Emily."
Emily Howell can even be trained to output intelligible English, by inputting words as objects. On Cope’s website, he quotes her as saying: “Why not develop music in ways unknown? This only makes sense. I cannot understand the difference between my notes on paper and other notes on paper. If beauty is present, it is present. I hope I can continue to create notes and that these notes will have beauty for some others. I am not sad. I am not happy. I am Emily.”
Cope’s dream for Emily Howell is simply to produce music that he likes. After spending a lifetime around music—-sometimes listening nine hours a day, seven days a week—he is looking for new pieces to be moved and inspired by. “It’s very exciting for me to imagine that my program can create new [music] that can make me happy,” Cope says.
But many listeners aren’t so happy about Cope’s music. If Bach’s creative genius is reducible to a set of algorithms, they think, maybe he wasn’t a genius at all; maybe humans cannot be truly creative. If unthinking, unfeeling computer programs can produce music, maybe our profound emotional responses to music just don’t make sense.
But Cope doesn’t share his listeners’ fears. “I think that the public has been so involved with this Romantic notion of the gods touching the composers and breathing through them the musical secrets of the universe,” says Cope. “We’ve forgotten that these things can be thought through and understood.” Cope sees his work not as an attack on human creativity, but as a way to better understand it.
For example, Cope hopes to gain insight into the style of individual composers, which may shed light on how musical styles evolved over time. He was intrigued to find 116 uses of parallel fifths—strictly against the rules of Western counterpoint—in 317 of Bach’s chorales.
“Knowledge can never be a bad thing. It always seems strange to me that when I attempt to understand things, what I’m doing is perceived as dangerous or problematic,” says Cope.
"Every limitation we place on the potential of machines is a limitation we indirectly place on ourselves."
According to Cope, composers have always operated within certain rules and been influenced by their predecessors. Classical Western tonal music follows many general rules, including the division of music into phrases and sections, and the use of major and minor scales. In chapter 5 of his book Computer Models of Musical Creativity,Cope traces similarities between music by Beethoven and Mozart, Mahler and Handel, and Stravinsky and Rimsky-Korsakov.
In general, Cope sees creativity as an active process, a type of nonlinear thinking that requires ignoring assumptions and making connections between disparate ideas; other composers’ works simply provide the raw material. “Creativity does not originate from a vacuum,” he explains.
And Cope himself has not become disillusioned about the great composers. In fact, he admires their work more and more because of the difficulties he experienced in trying to imitate their styles with EMI. “It made me appreciate what humans were capable of,” says Cope.
This is a theme that runs through much of Cope’s writing. In defending his work, he emphasizes the skill, imagination, and inspiration it takes to program software like EMI. He views his work as a triumph for humans, not the dethroning of human creative genius by the lifeless computer. “Every limitation we place on the potential of machines is a limitation we indirectly place on ourselves,” Cope writes. “I see no reason why computer-created music cannot move us to tears, find roots in our cultures...as much as any music composed in more traditional ways. As heretical to some as these thoughts may be, I believe them profoundly.”
Cope has faced his share of struggles in the past three decades. In 1987, he presented his first paper on EMI at the International Computer Music Conference in Illinois, and played some examples for the audience. Strangely, no one applauded or asked questions, as was customary. Cope spent the rest of his trip in his hotel room writing a defense of his work.
No one applauded.
Later that year, he presented the same paper at another conference, this time demonstrating the progression of EMI’s work. The audience laughed at EMI’s early attempts, but became silent as Cope played its later work. Most of the ten attendees, many of whom were his friends, left without speaking to him.
While trying to find a record company for EMI, Cope received over 100 rejections before Centaur agreed to record the music. He recalls receiving two letters on the same day—from one record company explaining that it only produces contemporary music, and from another saying it doesn’t produce contemporary music.
Cope also says that many people closed their minds and ears to EMI’s music, listening for mistakes or simply claiming that they couldn’t relate to the composer. In 1989, someone reviewed one of EMI’s Bach-style pieces without hearing it—and claimed he never wanted to.
There is some irony to this: there are those who are indignant over this electronic work—which has many affinities with classical compositions—while easily embracing Cope's firebrand avant-garde works from an earlier era; works often characterized by energetic, angular lines fusing into walls of varied dissonance which seem to then expand outward on several different tonal and timbral planes at once (E.g. Glasswork, 1978 ).
But Cope has never troubled much about the public’s response, choosing instead to use his own judgment. “I felt strongly that the work was worth something,” says Cope, who only reads about five percent of the reviews of his music. “I don’t worry about what people think of me.”
"I follow my own calling."
Cope recounts a story from childhood to help explain his individualist streak. When he was 13 or 14, he told his classmates that he was part American-Indian. Soon after, a boy followed him home from school carrying a large hunting knife. The young Cope screamed loudly and scared off his would-be attacker. The next day, Cope visited a school for American Indians in Phoenix and asked to join a basketball game. But the students said - politely - that he looked too much like an Anglo.
“I remember thinking, ‘I’m not a white person and I’m not an Indian. What am I?’” Cope recalls. “Even at that young age, it somehow freed me from following cultural stereotypes. I was me, and it felt very good.”
“Realizing that I didn’t belong has probably fostered my lifelong ability to not be concerned about those who attack me for being independent,” says Cope. “I follow my own calling.”
While Cope keeps plowing forward, the future of algorithmic composition is less certain. But the public reaction to Cope’s work has changed over the past 25 years. More students are becoming interested, and this summer will be the eighth year of the Workshop in Algorithmic Computer Music at UC Santa Cruz.
As Cope proclaims boldly on his website: “The works have delighted, angered, provoked, and terrified those who have heard them. I do not believe that the composers and audiences of the future will have the same reactions. Ultimately, the computer is just a tool with which we extend our minds.”
And Cope is not alone in his project. Various 20th-century composers have used algorithms to generate music. For example, experimental composer John Cage wrote music based on chance; the British electronic music group Autechre experiments with some algorithmic processes; and Lucas Arts video games feature a program called iMuse that adapts music to the action in real time. The experimentation, across a wide variety of musical styles and goals, has had the aggregate effect of pushing the technology forward at a brisk pace.
Computers are also becoming tools for composers in a broader sense. New software is constantly released to aid composers with their work and to analyze music. Digital instruments are created that produce music based on tactile cues—such as sliding a hand down a tube—and music can even be produced from brain waves.
In the future, Cope will keep working with Emily Howell and trying to create pieces that he loves. No matter how listeners ultimately judge his work, it encourages them to question the meaning of music and creativity, and the role of computers in their lives. Their answers will determine whether his music becomes the music of the future.
Emily Howell’s first album, From Darkness, Light, is now available for download on iTunes.
Kira Newman has studied writing, music, and philosophy and prior to graduation in 2010, completed a program in Paris in writing. She previously interned at The New Individualist and is now an assistant editor at the American Enterprise Institute.