Artificial Intelligence Music is Going to Change the World

Artificial Intelligence Music is Going to Change the World

The 2019 Japanese science fiction TV show Kamen Rider Zero-One introduces cutting-edge technology, artificial intelligence, which many people predicted would be the next crucial technological invention to change everyone’s lives. In this show, artificial intelligence is made into robots that look like real human beings. People use these AIs for many jobs, such as doctors, astronauts, teachers, and even actors (Kamen Rider). As in this show, artificial intelligence is already used in a wide range of fields in real life. Among all kinds of artificial intelligence, the most disrupting ones are those used in the music industry. They could both change music production dramatically and make the relationship between music and ordinary people closer in daily life.

Computer-based and digital music has been developing since the last century. When digital recording first appeared in 1996, more and more digital music technology was invented. With the introduction of music composing programs such as Cubase and LogicPro, composers can now compose songs without actual instruments. They could simply install these music programs with plugins of virtual instruments and write songs just with computers. In 2004, the Japanese company Yamaha developed a voice synthesizer software called Vocaloid, which many people describe as “virtual singers.” With Vocaloid, it is no longer necessary for music producers to hire a vocalist for their songs. They can purchase the voice databases and use these synthesized voices inside Vocaloid to make singing voices for their songs. It makes music production much more accessible, especially for amateur music producers. Vocaloid’s success gives many talented producers more chances to express their creativity, thus changing the music industry significantly (Boxall).

However, the evolution of music technology never stops. Vocaloid marks the beginning of an era in which computers can sing under programming, but now, a more radical revolution is happening. In the 2019 Japanese anime Carole & Tuesday, the story is set in a future world where people’s lives are mainly dependent on computers. Even music is written mainly by artificial intelligence (Motamayor). That is, in the future, computers can learn and compose music with codes that humans develop.

Although music artificial intelligence is not as popular as other artificial intelligence, the history of artificial intelligence in music begins much earlier than most expected. Earlier in the 1950s, two programmers from the University of Illinois, Lejaren Hiller and Leonard Issacson created an artificial intelligence called ILLIAC I, producing a song titled Illiac Suite for String Quartet. This is the first song ever that is composed entirely by artificial intelligence. The machine first generates a cantus firmus, the existing melody in polyphonic music, then generates three other voices according to the four-voice writing principle and sets the song’s rhythm, dynamics, and play instruction. This song might seem to be a masterpiece, but it was a failure, and many music fans questioned this song. Nonetheless, it is the first experiment of using only computers to compose music in human history, so ILLIAC I is still considered an important development of music artificial intelligence (Kharkovyna).

In the 1970s, music artificial intelligence gained more popularity than even pop singers wanted to help write songs. One of the most influential figures in the music industry, David Bowie, is the first pop singer who experimented with music artificial intelligence. Bowie and Ty Roberts developed a Mac application called Verbasizer to help him write lyrics. Verbasizer generates lyrics for songs from a significant source of different literary material, such as newspaper or diary. However, Verbasizer can only be used for writing lyrics but not songs, so it cannot fulfill people’s demand of generating a complete piece with both music and lyrics as good as human’s work (Kharkovyna).

Then, here comes the milestone in the history of music artificial intelligence—Experiments in Musical Intelligence (EMI)—programmed by a professor at the University of California and the composer David Cope. EMI works through analyzing existing music pieces and then generates its music. The result was shocking that many people would not realize the music produced by EMI is written entirely by a computer. Therefore, EMI no doubt brings us closer to the goal of automatizing music production (Kharkovyna).

Today, programmers have developed more advanced music artificial intelligence in various forms. For example, Continuator is an algorithm that can learn from and play with a human musician. Whenever a musician stops his performance, Continuator could pick up from where the musician has leftovers and create his song. In 2012, David Cope designed a new intelligence called Emily Howell, reflecting his idea that not all songs are unique because composers arrange the melodies and harmonies they learn from other works with their styles to create new songs. Recently, artists are already using artificial intelligence to write their music, as Taryn Southern produced an album composed by the program Amper, which can generate music with the genre and mood that the user’s input. However, Amper could only create basic random patterns of a song, and Southern had to do the rest of the work by herself. Music artificial intelligence are not mature enough to completely compose music on their own, but in the future, these virtual songwriters can gradually take on the jobs of actual composers as the developers keep refining their codes (Kharkovyna).

Music artificial intelligence can significantly change the music industry like an earthquake because it affects composers, producers, engineers, and the whole business. Artificial intelligence can learn patterns and styles of specific genres by analyzing the data and can help professional musicians create better works. These algorithms can generate random patterns that provide musicians with hints while writing songs. They can also help amateur artists who do not have enough professional knowledge make their pieces. Approximately twenty to thirty percent of the popular songs that people listen to today are partially written with the assistance of artificial intelligence. As mentioned in the previous paragraphs, David Bowie is also a fan of music intelligence and claimed that “I use it as a basis for improvisation, which might be enough to send me off to writing a song. It’s kind of like a technical dream in its way, and it will give me access to areas that I wouldn’t be thinking about otherwise.” (Beentjes)

Besides composing, artificial intelligence can also play a significant role in post-production, also known as audio mastering, which is adjusting the sounds of humans or instruments after recording. LANDR is artificial intelligence expertise in mastering (Marr). Similarly, plugins that music producers use during mastering can also be assisted by artificial intelligence. For example, iZotope, a company known for making plugins, uses artificial intelligence for its plugins for effects such as eliminating noises and reverberation. The algorithm works by loading the whole sound file into the graphical user interface so that mastering engineers can change the results at once instead of real-time processing (Beentjes).

Music-producing artificial intelligence can also impact the music business and profits that producers make from the songs. According to Deloitte’s most recent State of AI in The Enterprise Report, eighty-two percent of the companies that adopt artificial intelligence in their business are making more money than they spent since artificial intelligence decreases the costs and increases the profit in return. As computers automate the process of music production, producers only need as half as human power and less money on each project but still, make as much money as music produced by human artists (Shiloh). Therefore, artificial intelligence is beneficial to music producers because the producers would be able to make more money than using human composers, thus eventually changing the overall music business model.

On the other hand, music artificial intelligence can also positively impact ordinary people besides professionals who work in the music industry. It is no doubt that music, like other forms of art, has a relatively higher entrance than other subjects for people to learn. To become a musician, one needs to learn many different things, such as music theory, pitch ear training, and playing instruments, and sometimes having a good sense of pitch and rhythm can also be important for people to learn music (Musical U Team). In addition, it is also very costly for people to become a musician since they have to purchase many musical instruments and devices, especially for those who are going to compose music with computers. However, with music artificial intelligence, these problems can be solved as people do not need a lot of professional knowledge to write songs. People input existing music sources, making the computers learn the music’s pattern, chords, rhythm, and style. Then, the computers will generate music based on what they learn from analyzing these data. Amper, briefly mentioned in the previous paragraphs, is an example of music artificial intelligence that is very easy for people to use. The users are not required to have professional knowledge about music. Instead, they only need to choose the style and mood of the music they want to create, and the program will do the rest of the work on its own. Taryn Southern, the artist who published an album with the assistance of Amper, claims that she does not know a lot about music theory, which frustrated her a lot. She could play with a single chord for an entire song, but she does not know how to move on by changing into a different chord. Amper gives her an idea of composing songs that she cannot do alone (Deahl). Therefore, music artificial intelligence like Amper makes people who are not professionally educated with music but still want to become musicians create their pieces easily, incorporating music deeply into everyone’s lives.

However, as music artificial intelligence becomes more and more advanced, many moral issues arise, and people have to face them. Would artificial intelligence commit plagiarism? Who should the copyrights belong to? Will it completely replace human musicians in the future? Since music artificial intelligence works through learning existing music pieces, people might ask if it is cheating because computers just regenerate music created by other artists. There are still some ways to solve this problem. For instance, the FlowComposer team limits the proportion of existing music generated in the songs when programming artificial intelligence (Maude). However, it is still a tricky question since the standards for identifying whether a piece of music involves plagiarism or not is very vague, so it is very challenging for programmers to tell artificial intelligence to avoid plagiarism.

Music artificial intelligence also brings up legal issues such as copyright and whether the credit should go to the users or the computers. If a computer creates a song in the style of an existing artist, does that original artist share the copyright of this generated song? The answer is no since artificial intelligence simply makes a song that sounds like this artist, but this artist actually does not participate in this composition process. However, producers should ask first if it is legal to let artificial intelligence analyze and learn from music with copyright. Matt Aimonetti, CTO of Splice, claims that there is no need to be concerned about this issue because it would be tough for artists to identify that artificial intelligence uses their works as the computer only mimics their sound (Deahl). For the authorship issues, the current copyright law does not precisely offer a solution to these problems (Deahl). Therefore, a complete legal system is needed as artificial intelligence technology is developing.

Then we come to the most crucial question that threatens musicians’ identities: what does it mean to be a musician if computers can replace his job? To answer this question, Robin Reumers, a plugin and business developer at Leapwing Audio, claims that “the one thing about AI / Machine learning is that it is very good at being predictable, but music often is not. So fortunate for all of us, the creatives won’t be out of a job any time soon.” (Beentjes) Artifimprove their work qualityicial intelligence is supposed to be a tool that helps musicians improve the quality of their work and encourage more creativity. It provides a new view of music for people. However, one of the World Economic Forum’s reports found that artificial intelligence will create 133 million new roles but replace 75 million jobs by 2022 (Shiloh). Artificial intelligence is not advanced enough to create worry about whether music entirely by itself, but many musicians start to worry about if these computers would take their jobs away. According to Frithjof Bergmann, a philosophy professor at the University of Michigan, humans identify with the source of their actions (Bergmann). The musicians’ action is doing jobs related to music, and thus they identify themselves as musicians. When computers entirely replace their jobs, musicians would no longer make music, so their identities are threatened. In the anime Carole & Tuesday, Carole and Tuesday are two human musicians seeking success in this future world where artificial intelligence writes songs on their own instead of human beings (Motamayor). What should musicians do if the world in Carole & Tuesday comes true in the future?

It is no doubt that music artificial intelligence is a great invention that would benefit many people, both professionals and amateurs. People who work in the music industry can make better songs and more money, and people who do not have a formal music background will also be able to create music with the help of artificial intelligence. However, computers still bring many ethical problems, such as plagiarism and copyright issues. The rise of music artificial intelligence also leaves musicians a question of how they should identify themselves. People have to solve these problems before they fully implement computers into music production.