Song From PI: A Musically Plausible Network for Pop Music Generation
This addresses the problem of generating musically plausible pop music for creators and listeners, though it appears incremental as it builds on existing neural network approaches.
The authors tackled pop music generation by developing a hierarchical Recurrent Neural Network that encodes prior knowledge about composition, with bottom layers generating melody and higher levels producing drums and chords. They conducted human studies showing strong preference over a recent Google method, and demonstrated applications like neural dancing, karaoke, and story singing.
We present a novel framework for generating pop music. Our model is a hierarchical Recurrent Neural Network, where the layers and the structure of the hierarchy encode our prior knowledge about how pop music is composed. In particular, the bottom layers generate the melody, while the higher levels produce the drums and chords. We conduct several human studies that show strong preference of our generated music over that produced by the recent method by Google. We additionally show two applications of our framework: neural dancing and karaoke, as well as neural story singing.