Some firms continue to attempt to create computer-generated music that can compete with the work of professional musicians, despite the fact that this is not yet possible.
Bob Dylan was most likely referring to human musicians when he remarked, "Creativity has a lot to do with experience, observation, and imagination, and if any of these important aspects is missing, it fails."
Now, however, an increasing number of programmers are developing sophisticated songwriting tools that employ artificial intelligence (AI) to aid wannabe composers. Some computer programs can compose complete pieces of music from scratch, but the results may not be as engaging as Dylan's "Tangled Up in Blue" or "Hurricane."
Vibert Thio, a self-described "poetic technologist" from Taiwan, created the Lo-Fi Player, an interactive website for songwriting powered by machine learning, which went live earlier this year.
Thio, who designed the website while interning at Magenta, Google's research initiative researching AI as a tool for creators, said, "I want to make music easier to perform for practically anybody."
In order to make music, Thio's website employs machine learning, an application of artificial intelligence in which computers automatically learn from experience and adapt without human interaction. In the Lo-Fi Player's virtual room are concealed two AI tools. By merging various rhythms, the television, when activated, can make electronic beats. Or, as Thio puts it, "Imagine constructing a new face for a virtual sibling by combining your and your mother's faces, but using music." Radios conceal Melody RNN, an artificial intelligence technique for producing melodies. A new tune is immediately generated upon clicking it.
Thio's Lo-Fi Player is one of the most recent additions to an ever-expanding collection of music making applications. AI has been utilized in recent years to build choral harmonies based on Bach's music, compose hip-hop and jazz compositions, and sing in the voices of various celebrities. OpenAI, a nonprofit competitor to Google Magenta, introduced JukeBox in April, an algorithm that can generate complete songs in the style of 2Pac, Ella Fitzgerald, or Frank Sinatra.
Robert Laidlow, a British composer and researcher who has utilized artificial intelligence to make music for the BBC Philharmonic Orchestra, believes that appealing AI-generated music will not be viable for some time.
"I have yet to hear a piece of AI-generated music that is either stunningly beautiful or utterly unexpected," he says.
Laidlow creates music using a number of AI algorithms, each tailored to a certain task. For his current composition named "Alter," he employed MuseNet (from OpenAI) to generate new melodies and WaveNet (from DeepMind) to construct vocals that sounded human.
Laidlow considers AI to be more of a musical assistance than a replacement for human composers, as he uses it to create beats and simple melodies, among other things.
OpenAI, Sony, and other firms have released full-length songs composed solely from AI, but this is not the best use of the technology, according to Gus Xia, a computer scientist exploring AI and music at New York University Shanghai.
"The objective is not to compose a piece entirely from AI, from start, and then — bam! — you have a masterpiece," Xia explains, despite the fact that such a feat may be possible in the future.
In a few years, it may be technically feasible for AI to produce convincing music on its own, but Xia believes it would be uninteresting to listen to.
Laidlow concurs, noting that computers lack cultural context; hence, they are unable to present contemporary stories that elicit an emotional response from listeners. "People want to know why a work was written and to what it responds," he explains. And with AI, "I believe that will be a really challenging issue to solve."
In other words, even if AI were capable of producing a breathtaking piece of music, Laidlow asks, "What would it value?" Why would it compose music about a particular subject instead than another?"
The solution to these issues is currently, as Dylan would say, "blowing in the wind."