From Music Think Tank
The music industry has often been an early adopter of new technology. Musicians are keen to experiment with equipment that can usher their creative vision into new directions. Still, there have always been those who have seen the use of gadgets as in some way cheating, or uncouth. From the backlash Bob Dylan received after using an electric guitar at the Newport Folk Festival, to the current industry reliance on Auto-Tune. Perhaps the discussion shouldn’t focus on the fact that musicians are using technology, but how they are using it in order to support their creative visions.
Artificial Intelligence has begun to transform multiple industries, from improving diagnostic medicine, to helping provide a more personalized curriculum for students. For those working in the music industry, AI has the capacity to make creative pursuits more rich and diverse. The use of this technology isn’t limited to musicians, but has also found a place in studio engineering, production, even distribution. We’ll take a look at how different aspects of the sector are affected by machine learning, and how creative users are exploring its potential.
As early as the 1980s, composer David Cope was using a relatively primitive form of AI to create classical music in the styles of Bach and Beethoven. Programming with an early analytical engine, Cope’s computer created entirely new compositions which still held recognizable “fingerprints” of long dead composers. Cope’s project was most effective when he intervened in the predictable computer algorithm, and developed a randomness generator to break compositional rules. This rather reflects how machine thinking is being used in business today. Contrary to popular fears, AI isn’t removing the human aspect of work. Instead, we find that the technology is providing human minds with an additional creative tool.
In understanding the value of AI in songwriting, it’s helpful to know the bones of how it works. A program doesn’t simply pluck a new song out of thin air. It is reliant upon deep learning, processing huge amounts of data in the form of our rich history of human-created songs, before being able to create something new. This is not unlike how human composers operate. They’ve been fed data: music compositions, theory, chord structure; which they use as foundations upon which to build a new song.
While AI has the capacity to crank out complete songs, the musicians embracing it are generally using it as an inspirational tool. Programs like IBM’s Watson Beat consume and process a volume of musical data that a songwriter could not, and provide a base song for composers to edit and communicate through their unique human lens. This approach represents an interesting shift for the new music industry, in that composers are having to consider whether the programmers of these tools are also due songwriting credits and royalties.
Recording in the Studio
Technology has often offered musicians new options when recording in the studio — from Frank Zappa’s experimentation with 5-track recording in the 1960s, to that first use of Auto-Tune on Cher’s “Believe” in 1998. Whether in a garage or a huge and expensive studio, computers are a mainstay of today’s recording environments. AI is starting to provide producers and engineers new tools to make studio sessions more efficient and creative spaces.
Great studio engineering requires expertise acquired from years of experience, working with different producers and bands, feeling around for the right sound to match the song. However, AI is beginning to find a place at the desk — virtual mixing assistants are now available, using machine learning to balance and EQ audio tracks. These virtual assistants take care of some of the more mundane, repetitive aspects of the production process, freeing engineers and producers to concentrate on the creative aspects of recording.
The future of many industries is being shaped by AI, and we are certainly seeing this shift in studio mastering. While many musicians still prefer the human touch in this process, it can be prohibitively expensive for many smaller, independent artists. AI mastering services — such as Landr — are providing fast, more affordable mastering options. These programs analyze recordings using vast musical genre and style databases as reference, process them based on the analysis, and masters the tracks instantly.
AI technology is not only present in the writing and recording aspects of the music industry. Touring musicians are faced with many challenges, two of which are the costs and logistics of taking personnel around the globe. AI is finding a place in the live environment as a useful member of the touring ensemble. AI live musicians are more than just creators of electronic music, they’re agile performers.
Pop musician Holly Herndon, who also happens to have a PhD from Stanford’s Center for Computer Research in Music and Acoustics, has recently been touring with a piece of AI software called Spawn. The AI has assisted in composition, and has toured to promote Herndon’s album. Herndon asserts that Spawn doesn’t dictate the shape of a song, but can make decisions about how the song is performed — even recording, processing, and using sounds made by the audience.
The approach to live AI is not usually one of replacing musicians, but of augmenting a touring ensemble’s creative output with machine learning. Ash Koosha recently created an AI singer, called Yona, who collaborates with humans both in recording and during live shows. One of a growing population of auxumans (auxiliary humans), Yona has performed at the Rewire Festival in the Netherlands, her image a CGI-created avatar. While there is an element of autonomy to AI performers, the creators of these machine musicians are always adamant that the goal is not to create a machine free of human interaction.
An element of suspicion surrounding AI sometimes feels very natural; between popular culture and hard science, we’re fully aware of the most extreme dangers a thinking machine can present. However, the music industry has taken an informed, curious approach to its AI adoption. Rather than seeking to replace human talent, it is considered an exciting collaborative tool, and a key to greater creativity.