And the award goes to…Rock and Roll Bot 101111. While that may not be something we are accustomed to hearing, it may be soon thanks to OpenAI.
OpenAI is a research organization focused on artificial Intelligence (AI). It was founded in 2015 by Sam Altman and Elon Musk. Musk, in particular, has long been critical of AI, warning that it could bring about the downfall of the human race. OpenAI was founded to promote safe, responsible AI research and development.
OpenAI’s latest announcement is “Jukebox, a neural net that generates music, including rudimentary singing, as raw audio in a variety of genres and artist styles.” Providing Jukebox with an artist, style and lyrics gives the AI the raw material it needs to produce a unique music sample from scratch.
“We chose to work on music because we want to continue to push the boundaries of generative models,” reads OpenAI’s announcement. “Our previous work on MuseNet explored synthesizing music based on large amounts of MIDI data. Now in raw audio, our models must learn to tackle high diversity as well as very long range structure, and the raw audio domain is particularly unforgiving of errors in short, medium, or long term timing.”
The accompanying paper goes further, saying “we show that our models can produce songs from highly diverse genres of music like rock, hip-hop, and jazz. They can capture melody, rhythm, long-range composition, and timbres for a wide variety of instruments, as well as the styles and voices of singers to be produced with the mu- sic. We can also generate novel completions of existing songs. Our approach allows the option to influence the generation process: by swapping the top prior with a conditional prior, we can condition on lyrics to tell the singer what to sing, or on midi to control the composition. We release our model weights and training and sampling code at https://github.com/openai/jukebox.”
OpenAI is hiring and invites any interested in contributing to apply. In the meantime, Jukebox is another significant step forward for AI.