Gregorian chants and baroque and classical operas will be transcribed and made available in searchable databases using AI-powered tools.
These 400,000 images come from the Musical Paleography Workshop of the Abbey of Solesmes. Once digitised, these images will be stored and publicly available in DIAMM (Digital Image Archive of Medieval Music, University of Oxford). The catalogue information (metadata) will be indexed in the openly accessible MMMO portal (Medieval Music Manuscripts Online), DIAMM, and Cantus Index.
Artificial Intelligence is used to automate mundane but time-consuming tasks. Optical Music Recognition will extract information from scanned manuscripts, including melody and lyrics, creating a digital version of the score.
Automatic Music Information Retrieval will be used to attempt to identify the digital scores to automate the process of cataloguing the 2 million musical works by accounting for parameters such as the type of liturgical book, the work’s position in the liturgical cycle, and melodic correlation among the chants already indexed in the Solesmes archive as well as those of external digital archives processed with the project’s same optical music recognition technology. The training process requires a substantial amount of training data. 127,000 chants will be manually indexed, and many will be manually transcribed into digital format to supply the necessary materials.
The resulting technologies will be made publicly available for other archival projects. Optical music recognition tools will also be helpful to music publishers interested in digitising their back catalogues.