Research on Piano Harmony Automatic Orchestration System Based on Deep Learning

The research of piano harmony auto orchestrator system based on deep learning is to create a method to automatically create music scores from the original data of piano music. The original data includes not only notes, but also their timing, dynamic and other parameters. By using deep learning techniques, we can extract these parameters from raw data, and then use them to automatically generate scores. The core idea of the research on piano harmony automatic orchestrator system based on deep learning is that we first convert all notes into vectors (i.e., numbers). The system is an automatic arrangement system based on deep learning, which can automatically play piano melody. The research method is to train neural networks on various piano melodies, and then let them play in order. We used David Cope. ( https://www.pianocollections.com/ ) 6000+melody datasets in Piano Collections of. This dataset contains short clips and long clips with multiple themes, which makes it an ideal choice for training models because we can include many different types of tunes in the training set.