Legato 2: Multimodal recognition of musical scores

Artificial intelligence has transformed the way we interpret complex documents, and music is no exception. The Legato 2 system represents a significant advance in optical score recognition by processing musical notation sequentially, line by line, mimicking human reading. This innovation allows extracting not only rhythmic and melodic figures, but also textual elements such as titles and annotations, improving the semantic understanding of dense scores. Underlying this technology are vision language models and autoregressive processes that have boosted performance compared to previous approaches. In a business context, the ability to transform images into structured data opens opportunities for tailored applications in the publishing industry, music education, and digital archives, where AI for business becomes a strategic ally. Implementing such solutions requires not only sophisticated models but also a robust infrastructure, such as cloud services aws and azure that offer scalability and security. At Q2BSTUDIO, we develop custom software that integrates artificial intelligence, cybersecurity, business intelligence services with tools like Power BI, and even AI agents to automate workflows. The vision of Legato 2 reminds us that the future of technical and artistic documentation lies in the fusion of computer vision, language processing, and modular architectures, areas where expert guidance makes the difference.