Artificial Intelligence and music creation: What is OpenAI's Jukebox?

12 May 2020 | 3:01 pm | Emma Jones
Originally Appeared In

OpenAI has begun using artificial intelligence to actually make music with their latest development, Jukebox.

The future is now people. Not only do we have pandemic-proof rave suits being designed, we also now might be on the precipice of having music released made with Artificial Intelligence thanks to the latest development from OpenAI. Aptly titled 'Jukebox', the new model is now able to generate genre-specific music. WILD.

According to OpenAI's website, Jukebox is "a neural net that generates music, including rudimentary singing, as raw audio in a variety of genres and artist styles." Using over 1.6million songs as their dataset, Jukebox is able to use a song provided as input, and generate a sample produced from scratch in specific genres as output. Basically, it can decode music and reassemble it as something else using only artificial intelligence.

I first came across this program on triple J's Something More on Sunday night. Hosted by AI-enthusiast Tim Shiel, he explained the whole concept explaining that machine-learning -the same AI that powers streaming algorithms- is being taken to another level allowing people to make music with it. As Shiel explained, "They've trained a neural network with 1.6million songs. They've set a computer up and had a listen to 1.6million songs, the raw audio, just the sound. It's processed it and churned on it and thought about it and thought about it for a long time."

"Now what they have as a result of that is this kind of bizarre and potentially maybe accurate model of what structures make up a piece of music. From pop, hip hop, classical, across all these genres covered in these 1.6million songs."

Don't miss a beat with our FREE daily newsletter

Shiel then went onto to play some of the 7000 examples publicly available to show just how it works. In the example, a song was inputted into Jukebox for twelve seconds, and the AI has continued the song. The song Shiel played was 'Uptown Funk', and the results were surprisingly on point for obviously a very early form of this technology. Words are indeed identifiable, and you can make out Bruno Mars' voice in the song. What's also interesting is the way the structure of the song continues — it's not one big noise, and has a noticeable flow that is discernible despite it not being the most coherent thing you've ever heard.

It's an exciting evolution in the world of AI and music, and one that opens up endless possibilities in regards to how music is created.

Listen to Tim Shiel on triple J (the Jukebox part kicks off around the 29 minute mark), and check out OpenAI's Jukebox HERE.

Words by Emma Jones

Image via inno.si

SEE ALSO