Repertorium: Paving the Way for a Human-Centred AI Future
AI for European Music
Repertorium represents a significant step towards creating a general musical artificial intelligence with machine equivalents for eyes, ears, memory, intellect, and imagination. Doing so is an essential first step in exploring the AI-human values alignment problem by making European cultural heritage available to machine learning solutions.
Repertorium builds our future from our past by preparing the ground for an AI to understand and conform to human intentions, preferences, and goals as embodied in centuries of European civilisation’s traditions and artefacts, in addition to our instructions. Repertorium will serve as a model in shaping our new human-centred digital world.
Repertorium represents a significant step towards creating a general musical artificial intelligence with machine equivalents for eyes, ears, memory, intellect, and imagination. Doing so is an essential first step in exploring the AI-human values alignment problem by making European cultural heritage available to machine learning solutions.
Repertorium builds our future from our past by preparing the ground for an AI to understand and conform to human intentions, preferences, and goals as embodied in centuries of European civilisation’s traditions and artefacts, in addition to our instructions. Repertorium will serve as a model in shaping our new human-centred digital world.
Repertorium promotes European culture, heritage, and values through tools that offer access to and expand the knowledge and appeal of mediaeval and classical music among the public and academics, providing new ways to search for and interact with musical works databases.
Archive music with OMR and MIR
Developing AI-based tools to automate the digitalisation and cataloguing of historical mediaeval and classical music manuscripts through optical music recognition (OMR) and music information retrieval (MIR) tools.
18th Century Opera Scores in MEI format
Creating machine-readable versions of a selection of 100 18th-century aria scores corresponding to musical settings of the 27 dramas by Pietro Metastasio (1698-1782), written by different composers, will be produced in MEI format.
This result will facilitate the work of musicologists interested in 18th-century operatic music and attract the interest of performers and music publishers, which could rescue this almost-forgotten genre. This outcome preserves and promotes these opera scores.
Musical Archive Goes Digital
Publishing digital images and machine-readable versions of the publicly inaccessible musical archive of the Atelier de Paléographie Musicale of the Abbey of Solesmes, France, publicly available via DIAMM.
Chant Indexing
Manually indexing 127,000 Gregorian chants from Solesmes Archive and publishing results to MMMO, DIAMM, and Cantus Index.
Tridentine chants on Neumz
Recording the complete Gregorian chants of the Tridentine Rite, publishing to Neumz.
Metastasio's Manuscripts Digitized
Digitising and manually indexing 100 Metastasio operatic manuscripts and publishing them through DIDONE.
Revived Opera Treasures Unveiled
Two public concerts of restored mediaeval and classical opera scores recovered through Repertorium.
Linked Data Models
A Linked Data model will enable the sharing of metadata among musicological databases. This gateway will connect Cantus Index, DIAMM, MMMO, Neumz, and others.
A2S Tool
An A2S (audio-to-score) tool will automatically align Tridentine Rite recordings to their digital scores for synchronised playback.
The tool will be trained and tested leveraging the over 6500 hours of recordings currently in the Neumz archive and further refined while recording the c.2000 hours of Tridentine chants.
Minus-One Tools
Music students will be able to “take the place” of an instrument in any standard stereo recording or solo their instrument to hear the line independently.
MIR
MIR techniques will assist in an analysis of Solesmes archives, filtering potentially novel chants for priority manual review. It is estimated that this project will bring to light approximately 4,000 novel Gregorian chants previously unknown to the musicological community.
Novel compositions will be assigned new Cantus IDs and catalogued with metadata and chant text in the Cantus database. Develop a set of AI-based Music Information Retrieval (MIR) tools that avoid duplicates when cataloguing musical works and automate annotation procedures.
Immersive 3D audio
The tool will enhance the streaming concert experience to favour the streaming business in the classical music market and grow new audiences.