MUSE (Multilingual Semantic Embeddings)

Linking concepts in music-theoretical texts across languages

AI/ML
Data Development
Multilingual
Music Studies
Natural Language Processing
Text Analysis
music

Despite the enormous diversity of musical phenomena that exist across historical and cultural spaces, the majority of music-theoretical and scientific approaches in music studies have focused on interpretations of Western canonical source materials, neglecting a vast dataset of source documents from underrepresented languages and communities. To date, most global discourse on music theory remains untranslated, which limits the possibility of building equitable relationships among global music communities and privileges intellectual traditions occurring in European languages, particularly English.

During the collaboration with the CDH, we will focus on assessing the use of multilingual LLMs for tracking concepts between texts, based on a controlled vocabulary developed by the project PIs. The goal is to assess and develop the capacity to link concepts in music-theoretical texts across languages as part of a larger project to expand the “canon” of music theory.

Team

Co-PI

Jürgen Hackl
Anna Yu Wang

Technical Lead

Laure Thompson

Technical Project Manager

Project Advisor

Jeri Wieringa

Grants

2024–

Research Partnership