MIDI

The paper introduces a technique for online symbolic music alignment using reinforcement learning. The reinforcement learning agent, an attention-based neural network, is trained to iteratively predict the current score position…