Creative Artificial Intelligence in Music: Genius Innovation for Sync and More Fascinating Use Cases

29 August 2017
By Sergey Bludov, SVP Media & Entertainment

Creative-Artificial-Intelligence-in-Music

Now that we’re past the first songs composed by artificial intelligence, do you think using AI is the logical next step in optimizing the generation of music for the film, television and games industries?

“I’ve always been fascinated by the concept that we could automate, or intelligently do, what humans think is only theirs to do. We always look at creativity as the last bastion of humanity. Could you press a button and write a symphony?”  — Siavash Mahdavi, CEO of AI Music, a British tech startup exploring the potential intersections between artificial intelligence and music.

Many AI companies are now deeply entrenched in research for a variety of applications of this technology in the music industry, and remarkable achievements have already been made. Increasingly tight budgets in the film, television, video and games industries have led people to seek out methods for reducing costs without sacrificing the quality of the music being used. Now, a wide range of AI firms are advancing their efforts toward developing technology that can become the creative force behind the generation of the actual tracks to be utilized in these mediums.

This sector is expanding rapidly, as an increasingly large number of companies race to get their software to the next level, including British-based Jukedeck and AI Music, San Francisco’s Humtap, Berlin-based Melodrive, and Groov.AI in Mountain View, California. Even Google has an AI music research project underway called Magenta, and Sony’s Computer Science Laboratories (CSL) in Paris is conducting a similar project called Flow Machines.

Ramifications on professional musicians

All of these companies and researchers are attempting to answer the same complex and sometimes controversial question: can machines use AI technologies such as neural networks to analyze existing human-made music and subsequently learn how to produce their own compositions? And while this is certainly a massively complex question to answer on its own, it also poses the additional question of what ramifications this technology has on professional musicians, many of whom make all or part of their income through composing or licensing music to be used in television shows, films, and video games. In fact, some musicians are going so far as to propose a restriction on using their previous recordings and compositions for AI and machine learning purposes. At the same time, other artists realize the creative power of AI and incorporate it into their work.

Valuable and non-threatening uses

Although a range of questions and concerns exist within the aforementioned applications of AI in music, other valuable and non-threatening uses are already working well.

One of these applications is the creation of playlists based on moods. While music data specialists at Gracenote have been classifying music by this type of criteria for years, the employees have not actually listened to every one of the 100 million songs in the company’s database. Instead, Gracenote has used machine learning and artificial intelligence to teach computers how to detect the emotions in music, thereby allowing machines to accurately classify tracks as dreamy, sad, sultry, or other feelings.

Another fascinating current integration comes from Google researcher Douglas Eck, who instigated the company’s Project Magenta. One of the concepts that the research team is working on involves teaching a neural network how to learn the musical characteristics of an instrument by analyzing hundreds of notes. The machine subsequently creates a mathematical representation that identifies that particular instrument. This results in the ability to simply move a button across a screen to combine instruments in order to create entirely new virtual instruments, such as one that is 62 percent trumpet and 38 percent saxophone, or any other combination of sounds.

Treat AI not as an enemy — but as a collaborator

The music industry is on the cusp of radical changes to the manner in which songs are composed for a vast range of uses. From film and television background music to video games and possibly even hit pop songs, AI is at the forefront of this new era of machine-driven music composition. The key to fully utilizing its potential is treating AI not as an enemy — but as a collaborator. Let it complement the artist’s work. Even as it automates some tasks, humans will still be at the helm of the creative process. At least for the foreseeable future.


Add Comment

Name Mail Website Comment