Imagine your favourite song. Now imagine a machine cranking out 50 similar tunes in a matter of days. Musicians are already experimenting with artificial intelligence (AI) to create songs that actually aren’t that bad.
This is nothing new
This may come as a surprise, but musicians have been experimenting with computer-generated music as far back as the ‘50s. Composer Lejaren Hiller used a computer to help create “Illiac Suite” in 1957, according to the New York Times.
Now, musicians can take advantage of deep learning, which involves “neural networks” loosely modelled after the human brain. These networks can sort through vast amounts of data to find patterns and essentially learn concepts.
A work of art
Google built a neural network that learned the concept of a cat and then used that to accurately pick out kittens in YouTube videos. Similarly, Sony’s Flow Machine analysed classical masterpieces to understand the concept of music and used that to create new songs.