Piano Data

Piano data research focuses on developing robust and efficient algorithms for tasks such as piano transcription (converting audio to musical notation), performance difficulty estimation, and score-to-audio synthesis. Current research employs neural networks, including convolutional recurrent networks, transformers, and autoregressive models, often incorporating techniques like transfer learning and self-supervised contrastive learning to improve accuracy and efficiency, particularly for real-time applications. These advancements have implications for music education (assessing performance difficulty), music information retrieval (linking audio and sheet music), and music technology (generating realistic piano performances from scores).