About
AI-MUSE is built by two companies, in Venice and in Larnaca, with research and heritage partners in Italy, Cyprus and Türkiye, around one conviction: a machine reading of an archive is worth something only if a person who knows the material has checked it.
The partners
-
ELAN42
E42 S.r.l., Venice. Lead partner: it builds the software, the reading pipeline, the training runs and the editor.
-
Elite Tutors
Private Institute C.F. Elite Tutors Ltd, Larnaca. An Erasmus+ accredited educational institute that has already brought AI into EU-funded education projects. In AI-MUSE it checks the output in Greek and the other languages, with reviewers independent of those who produce it, and coordinates the Cypriot pilot.
Associated partners
-
CREAaps
Centro Ricerche EtnoAntropologiche, Florence, a research organisation in the social sciences and humanities. It designs the metadata schema, the taxonomy and the codebook, and the expert review of meaning, context, bias and omissions on which every approval rests.
-
IVESER
Istituto veneziano per la storia della Resistenza e della società contemporanea, Venice. Italian pilot: the clandestine press of 1943-45 and the newspaper «Giustizia e Libertà».
-
Pierides Museum – Bank of Cyprus Cultural Foundation
Larnaca, with the Pierides Foundation. Cypriot pilot: the Foundation's archive and the Pierides family archive, late 19th and 20th century, in Greek, English, French and Ottoman Turkish.
-
Larnaka Archives – Phivos Stavrides
Larnaca. Sessions and reserve material for the Cypriot pilot.
-
NOAH Culture and Arts
İstanbul. Turkish pilot: handwritten material of its own choosing, approved by its own curators.
-
ENOVA Training Consultancy
Türkiye. Commercial and market development for Türkiye.
With the support of VeniSIA, Venice's deep-tech venture builder, which has signed a letter of interest for bringing AI-MUSE to market.
Compute
AI-MUSE has been granted compute resources through the IT4LIA AI Factory (EuroHPC), project EHPC-AIF-2026PG01-1068. Processing runs on Leonardo Booster at CINECA, Bologna.
Built on open work
We stand on tools other people made and published: Kraken for line recognition, olmOCR-2 from the Allen Institute for AI and LightOnOCR from LightOn for general reading, copisteria from KobiMusic for music, musicdiff for measuring it, and the datasets listed in the case studies, whose authors transcribed thousands of lines by hand.
Contact
For a test on your own material, or anything else: info@ai-muse.eu. The application is at app.ai-muse.eu; accounts are given to the institutions we work with.
Image credits
- Screenshots of the AI-MUSE application, E42, 6 November 2026.
- «Giustizia e Libertà», 1945-1946: scans courtesy of IVESER, Venice.
- Greek, Latin and Arabic manuscript pages: the datasets listed in the case studies, CC BY 4.0 and Apache-2.0; BULAC images in the public domain.
- Mozart, K.155, Breitkopf & Härtel 1881: IMSLP, public domain.
- The illuminated logo is provisional.
