As part of its Horizon Research and Innovation Program, the European Commission has allocated three million euros to the Repertorium project, which aims to develop artificial intelligence (AI) tools to help preserve the ancient musical patrimony by digitizing hundreds of thousands of medieval musical manuscripts. Most of them concern sacred music.
The Project
Repertorium is the acronym for Researching and Encouraging the Promulgation of European Repertory through Technologies Operating on Records Interrelated Utilizing Machines.
The project operates in two primary spheres, musicology and sound processing. In musicology, it combines artificial intelligence (AI) and deep learning (DL) solutions to perform optical music recognition (OMR) and music information retrieval (MIR) across multiple musical datasets.
The project will provide a technological platform for preserving European musical heritage through databases of medieval and classical works. Over 400,000 ancient manuscript images will be scanned and digitized from the Solesmes archive and made available for musicological study through Oxford’s DIAMM portal.
More than two million musical works will be correlated to check the concordance among the Solesmes dataset and external datasets such as the Cantus Index network. Approximately 4,000 novel chants are expected to be discovered and prioritized for manual indexing.
As for sound processing, it will enable the development of cutting-edge technologies for the separation of sound sources from instruments in real time, specifically adapted to classical music. Repertorium will create solutions for sound field reconstruction that will work in real time.
An Initiative That Continues Past Efforts
As Solène Tadié explains in the National Catholic Register, this initiative is a continuation of Neumz, the Gregorian chant recording project developed between 2019 and 2022. Repertorium, which began in January 2023 for a duration of three years, will provide online access to the entire traditional Latin liturgy.
At the intersection of preserving cultural heritage and promoting technological innovation, Repertorium will serve the Catholic tradition of Gregorian chant on several levels. After the digitization of the ancient manuscripts, will come the indexing and cataloging of the approximately two million pieces contained in these archives. This is where AI comes into play.
According to John Anderson, founder of Neumz, “It takes an expert about three minutes to identify a song and catalog it.” One of Repertorium’s coordinators told NCR: “It would take an expert 500 years to index everything, so AI will help us.”
The Spanish universities of Alicante and Jaén are responsible for developing an optical recognition and musical information retrieval algorithm that will analyze the entire archive and compare it to other external databases such as the Cantus network Index.
Based on nearly 8,000 hours of audio files provided by the Neumz application and its developer Odratek, the AI will be gradually trained to listen to music and follow the music score. Likewise, a total of 127,000 pieces will first be indexed manually, in order to help the AI develop musical information search functions.
The AI system designed by Repertorium, capable of reading music scores, tracking audio, and recognizing the sounds of individual instruments, will be extended to other types of music, such as classical music, offering immersive experiences where it will be possible to follow the performance in the form of an “acoustic hologram.”
The work provided by the music information retrieval technology should make it possible to identify and catalog more than 90% of the Solesmes archives within three years, according to Mr. Anderson. Experts will thus be able to save time and concentrate on unidentified works.
“As songs are cataloged, new works are constantly being discovered,” continues Mr. Anderson. Project coordinators estimate that around 4,000 new pieces are expected to be discovered. They will be made public during concerts in Europe.
“We plan to organize traditional concerts where the public will discover Gregorian pieces, some of which have not been heard for 1,000 years,” promised Mr. Anderson.
Gregorian Treasures Made Accessible to All
One of the most anticipated aspects of the project is the recording and online broadcast of the entire Tridentine liturgy. This option is made possible by the St. Madeleine du Barroux Abbey, which opened its doors to the Neumz team as part of the Repertorium project.
For a little over a year, some 2,200 hours of Gregorian chants sung during the Mass and the Divine Office in Tridentine forms, according to the ancient handwritten tradition, will be recorded and made available to Neumz subscribers, with translations of Latin texts into 5 languages and scores in synchronized square notation.
Mr. Anderson's ambition since the conception of Neumz, which led to Repertorium, is to reintroduce Gregorian chant on a large scale into churches, particularly in the face of the growing ignorance of this tradition, including among priests. By facilitating access to all the chant classified by Office, he hopes to popularize them.