Note Onset

Note onset detection, the identification of musical note beginnings in audio recordings, is a crucial task in music information retrieval (MIR) with applications in automatic music transcription and analysis. Current research focuses on improving accuracy, particularly for challenging cases like soft onsets in string instruments and singing, employing advanced neural network architectures such as convolutional and transformer networks, often incorporating data augmentation techniques to enhance model robustness. These advancements are leading to more accurate and efficient algorithms for various instruments, enabling real-time applications and deeper understanding of musical performance nuances, including playing techniques.