Project description
Using AI to preserve classical music
The EU-funded REPERTORIUM project is developing AI tools to automatically digitise and catalogue neumatic and classical manuscripts, and preserve and make accessible European musical heritage through sophisticated linked data models and APIs. Moreover, it is creating state-of-the-art (AI and algorithmic) sound processing technologies (e.g. sound source separation, sound field reconstruction, audio-to-score alignment) that will enable real-time low-cost metaverse-ready immersive audio streaming. Its outcomes will revolutionise music scholarship, enhance streaming revenues and empower musicians, including offering on-demand "minus-one" filters for practicing music students. These technologies will form the foundation of a general musical AI that fully unleashes the powers of machine learning upon the domain of European classical music heritage, advancing us towards a human-centred digital world.
Objective
Music, as one of the most preeminent European artforms that has impacted worldwide cultural heritage, has an intrinsic value enriching our lives. However, music manuscripts frequently remain private, unshown, or unexploited because they are only available as printed or handwritten in local archives.
REPERTORIUM aims to: 1) to provide a technological platform for curating databases of mediaeval and classical European art-music works, linked to other relevant existing databases around the world and fed by automated manuscript digitisation and music information retrieval techniques based on Artificial Intelligence (AI); and, 2) leveraging the above technology to create state-of-the-art audio recording and instrument separation technologies (AI-based, stochastic signal processing, and ambisonics spatial audio) targeted at music education institutions (conservatories), professionals (musicians and orchestras) and the public (streaming services).
Combining a novel digitisation tool that leverages AI and Deep Learning solutions to perform Optical Music Recognition and Music Information Retrieval across multiple music datasets opens valuable solutions to problems affecting music businesses while efficiently preserving and rendering accessible European musical heritage. Thus, it is possible to provide cost-effective solutions for immersive streaming and virtual reality experiences by leveraging Sound Source Separation and Spatial Audio technologies.
The consortium includes musicologists (ICCMU, MMMO, UOXF), a musical organisation (AHECG), an orchestra (LNP), and a company focused on early music (ODRATEK). Its members have been previously awarded funding by the EC for RIA projects (TUNI, POLIMI, ICCMU, UOXF), UJA has experience in coordinating H2020 projects. It is composed of a balanced combination of research participants and industrial / commercial partners, from 8 European countries (4 universities, 2 RTOs, 2 NGOs, 1 orchestra and 3 companies in the music sector).
Fields of science
CORDIS classifies projects with EuroSciVoc, a multilingual taxonomy of fields of science, through a semi-automatic process based on NLP techniques.
CORDIS classifies projects with EuroSciVoc, a multilingual taxonomy of fields of science, through a semi-automatic process based on NLP techniques.
Programme(s)
Funding Scheme
HORIZON-RIA - HORIZON Research and Innovation ActionsCoordinator
23071 Jaen
Spain