When he was 8 years old, Matt Reeves started making 8-millimeter movies inspired by his love for the original "Planet of the Apes." "I'd have my friends put on gorilla masks and run around shooting these little sci-fi films," he recalls. "As a kid, I was captivated by these images of horses with apes on them." Decades later, Reeves, perched on a sofa in his tidy Hollywood office, has taken his fascination with primate cinema to a whole new level as the auteur behind the 2014 performance-capture blockbuster "Dawn of the Planet of the Apes" and this summer's "War for the Planet of the Apes."
We propose a novel approach for the generation of polyphonic music based on LSTMs. We generate music in two steps. First, a chord LSTM predicts a chord progression based on a chord embedding. A second LSTM then generates polyphonic music from the predicted chord progression. The generated music sounds pleasing and harmonic, with only few dissonant notes. It has clear long-term structure that is similar to what a musician would play during a jam session. We show that our approach is sensible from a music theory perspective by evaluating the learned chord embeddings. Surprisingly, our simple model managed to extract the circle of fifths, an important tool in music theory, from the dataset.