Methodologies for Music Understanding and Generation in the Context of Trustworthy AI

Explore methodologies for music understanding and generation through the lens of trustworthy AI in this plenary conference talk delivered by Xavier Serra from Universitat Pompeu Fabra, Barcelona. Discover how the Music Technology Group approaches computational music challenges by combining powerful machine learning techniques with domain-aware, transparent, and ethically grounded methodologies that address the deeply human and culturally rich nature of musical expression. Learn about cutting-edge research at the intersection of music and artificial intelligence, including audio analysis, symbolic music processing, generative modeling, and multimodal representation learning. Examine how principles of trustworthy AI—including fairness, transparency, cultural awareness, and reproducibility—are integrated throughout the research pipeline to ensure socially responsible advancement in the field. Gain insights from recent projects that demonstrate the critical importance of interdisciplinary collaboration and open science practices in music AI research. Understand how technical choices in music AI can reflect broader societal values, particularly when working with creative, subjective, and culturally diverse musical domains. Engage with discussions on balancing innovation with accountability in music AI research, and explore how computational approaches can empower creators, educators, and users while preserving the cultural and social dimensions that make music fundamentally human.