muScribe: Automated tranScription of muSic

muScribe aims to transform music transcription by utilizing advanced AI to convert audio recordings into detailed music scores, making music more accessible to the public. Conducted at the RITMO Centre for Interdisciplinary Studies in Rhythm, Time, and Motion at the University of Oslo, the project explored both B2B targeting music archives and B2C aimed at a larger range of users, offering new solutions to digitize performance recordings through cutting-edge deep learning techniques and innovative research. By combining machine learning with symbolic AI, the project addresses key market needs for cultural institutions, music publishers, and copyright organizations, automating transcription to reduce costs while enhancing precision and availability of music scores. This technology not only improves music rights management, facilitating better royalty distribution through linking audio performances with their corresponding scores, but also enhances music discovery and allows for more detailed manipulation of music representations. Furthermore, it aims to empower underrepresented music cultures and preserve fragile cultural heritages, including traditional Nordic folk and contemporary classical music, while making music more accessible to the hearing-impaired and advancing music therapy for mental health.

We target a B2B model for cultural institutions like archives, libraries, opera houses, and conservatories needing to transcribe extensive music archives. We also target a B2C model for a broader market segment. Current manual transcription is slow, costly, and imprecise. Music publishers need our technology to expedite score production and avoid using copyrighted scores. Our transcription technology can help protect copyrights for performances registered with Gramo, facilitating transcription and comparison of original and potentially infringing performances. Our technology can help generate videos displaying music scores, potentially attracting large audiences. It also aids creative sectors in producing computerized music representations, as already used in TV series with professional endorsements.