A deep learning system that generates osu! game beatmaps by training models on audio FFT data and rhythm features. The rhythm model classifies note types (circles, sliders, spinners) from music spectrograms and timing inputs, while the momentum model predicts spatial positioning and movement patterns for placement decisions.