Piano Performance
Research on piano performance is increasingly leveraging artificial intelligence to analyze, generate, and even teach musical performance. Current efforts focus on developing deep learning models, particularly transformers and convolutional neural networks, to address tasks such as score transcription from audio or MIDI, generation of expressive piano performances from scores or text descriptions, and automated assessment of performance quality and musical shape. These advancements have implications for music education, robotics, and music information retrieval, offering tools for personalized instruction, robot musicianship, and improved music analysis.
Papers
Are we describing the same sound? An analysis of word embedding spaces of expressive piano performance
Silvan David Peter, Shreyan Chowdhury, Carlos Eduardo Cancino-Chacón, Gerhard Widmer
Sounding Out Reconstruction Error-Based Evaluation of Generative Models of Expressive Performance
Silvan David Peter, Carlos Eduardo Cancino-Chacón, Emmanouil Karystinaios, Gerhard Widmer