Do you ever wonder what it would be like if your all-time favourite artist covered your most played song? Usually, this wouldn’t be conceivable unless it was by complete coincidence, but thanks to the artificial intelligence firm OpenAI launching their own Jukebox, a neural network that generates music from existing tracks, the collaboration of your dreams is now possible thanks to the power of AI.
According to CJ Carr, one half of the DADABOTS team who were responsible for this player - alongside his co-producer Zack Zukowski in an interview with Futurism, he explains that this works by a process called neural synthesis.
“It’s reading a sequence of audio at 44.1k or 16k sample rate, and then it’s trying to predict what the next sample is going to be. This net is deciding a fraction of a millisecond to come up with the next sample. This is why it’s called neural synthesis. It’s not copying and pasting audio from the training data; it’s learning to synthesize.”
“What can that language tell us about the underlying rules and components of music, and how can we use these as building blocks themselves?” muses Zukowski. “They’re much higher-level than chords — maybe they’re genre-related. We really don’t know. It would be really cool to do that analysis and see what happens by using just a subset of the language.” It’s as though a robot is writing its own audio fan-fiction.
John Williams x Napalm Death
The samples that are available on the Jukebox include everything from Diana Ross to Snoop Dogg, and you can create a hybrid of the considered melodies of a John Williams soundtrack to the sheer velocity of Napalm Death in a single click.
But the editorial team at Futurism weren’t happy, instead requesting a version of Britney Spear’s 2003 banger Toxic in the style of the late Frank Sinatra, a feat that would usually not be possible without a necromancer getting involved. Have a listen below. It’s pretty trippy.