Meet Shimon - a four-armed robot that composes and performs its own Marimba music using artificial intelligence. 

While it’s played music for years, the developers at Georgia Tech have implemented deep learning technology into the robot. This means it can study huge datasets of songs from well-known musicians and create its own original composition.

And even though it’s fair to say that robots pose a serious threat of going all Skynet on us and enslaving mankind, turns out their music taste is pretty chill.

Under the original programming, Shimon (created by Gil Weinberg, director of Georgia Tech’s Centre for Music Technology) was able to improvise music alongside live musicians in October 2016 - using what was called an “interestingness” algorithm to make sure it wasn’t just copying what the rest of the band was doing.

Now, though, Ph.D. student Mason Bretan has added a AI, Shimon can autonomously generate a song with full understanding of melodic and harmonic structure.

What happened to make this possible? Well, the robot studied a database of almost 5,000 songs using deep learning - covering a wide array of artists such as Beethoven, Lady Gaga, Miles Davis and The Beatles. Alongside this, Shimon also listened to two million musical licks, motifs and riffs. 

After this, Bretan gives the robot a “seed” of music, consisting of four bars, and then it’s all on Shimon to compose.

“Once Shimon learns the four measures we provide, it creates its own sequence of concepts and composes its own piece,” said Bretan in a press release. “Shimon’s compositions represent how music sounds and looks when a robot uses deep neural networks to learn everything it knows about music from millions of human-made segments.”

Have a listen to two of it’s compositions below!