Ever wish you could whistle a tune and have a computer build a whole song out of that idea? It might just happen. Facebook AI Research has developed an AI that can convert music in one style or genre into virtually any other. Instead of simply trying to repeat notes or style-specific traits, the approach uses unsupervised training to teach a neural network how to create similar noises all on its own. Facebook's system even prevents the AI from simply memorizing the audio signal by purposefully distorting the input.
The results are uncannily good, as you can hear below, although they clearly depend on the music involved. A Metallica riff sounds odd as a classical piece (S&M notwithstanding), but you could easily convert a Mozart symphony into one of JS Bach's harpsichord-heavy performances. And yes, whistling works -- the scientists produced a decent, if slightly strange-sounding, orchestral take on John Williams' Indiana Jones score through that method.
This work is clearly appealing to musicians who want to experiment with genre-defying covers and remixes, or who just want to turn an idea into a song when they don't have conventional instruments on hand. It could help music novices, too. Rather than teach yourself an instrument or music editing software, you might only need your mouth and a knack for a good melody.