NPMC Tech builds AI tools for music score processing. Our system automates measure detection and numbering, paired with a correction workflow that keeps the user in control. Now in controlled beta with select partners.
Built at the intersection of machine learning, computer vision, and hands-on music domain knowledge
Deep learning models trained to identify structural patterns in music scores. Models are refined iteratively using partner data and controlled feedback loops.
Image processing pipelines that extract structural information from musical scores. Our systems handle variations in scan quality, notation density, and diverse score formats.
Grounded in music theory, notation conventions, and how scores are used in rehearsal and education settings. The product reflects direct experience with the domain.
Focused tools for music score processing, currently in controlled beta
The system identifies measure boundaries in scanned music scores and applies numbering automatically, reducing hours of manual annotation work.
Pages are processed in approximately 1.3 seconds on average, making batch processing of full scores practical. Speed varies with score complexity and scan quality.
After automated detection, users can review and correct results through a guided interface. The system assists; the user has final say.
Multiple specialized models work together on different aspects of score analysis, each trained for a specific detection task.
Currently in use with select institutional partners across rehearsal and education environments.
Built for web deployment from the start, with a processing pipeline that can handle concurrent users and batch uploads.
Score processing that understands the domain it serves, built for the people who work with music every day.
Whether you're an institution exploring score processing tools or a researcher interested in the underlying technology, we'd like to hear from you.
Interested in beta access, partnership opportunities, or learning more about the project — send us a message.