C4DM Seminar: The importance of multimodal data in the computational understanding of music
QMUL, School of Electronic Engineering and Computer Science
Centre for Digital Music Seminar Series
Seminar by: Anna-Maria Christodoulou
Date/time: Thursday, 21st May 2026, 3 pm
Location: G2, Engineering Building, Mile End Campus, Queen Mary University of London
Title: The importance of multimodal data in the computational understanding of music
Abstract:
Music is multimodal, perceived not only through hearing but also through vision, touch, and movement. Humans naturally integrate these sensory streams to interpret a music performance. Similarly, in order to achieve a high-level computational music understanding, multimodal data is essential. This approach is put into practice through a five-category framework for music context, which guides computational systems to analyze instrumentation, performance dynamics, spatial arrangements, emotional expression, and causal relationships (the "why" behind performative choices). The presentation will detail the construction and use of large-scale datasets for Music Question–Answering (MQA) and ways in which we can benchmark SOTA models to examine the extent in which current technologies use auditory-visual perception to understand a music performance.
Bio:
Anna-Maria Christodoulou is a PhD researcher at RITMO, UiO and a machine learning data engineer at Acon Digital, working at the intersection of music and technology. Her research examines computational music understanding through multimodal datasets. As a dedicated advocate for reliable, ethical, and accessible Music Information Retrieval (MIR) projects, her research advocates for open and FAIR datasets. In her free time, she works on research and community projects that analyze the effect of music in health and well-being. Beyond her individual endeavors, she is part of communities exploring the connections between music, AI, and technology, and is a mentor in teams supporting women in tech and music informatics.