Musical Structure

Musical structure analysis aims to understand how musical elements are organized and how this organization impacts perception and creation. Current research focuses on developing computational methods, employing techniques like coupled hidden Markov models, graph representations, and transformer networks (including variations with fine- and coarse-grained attention), to automatically analyze and generate musical structures, often leveraging chord embeddings and symbolic music representations. These advancements improve music information retrieval tasks, such as segmentation and transcription, and inform the development of more sophisticated music generation systems capable of producing musically coherent and stylistically consistent compositions. The resulting insights contribute to both music theory and artificial intelligence, bridging the gap between human musical understanding and computational modeling.