Beici Liang: Protecting music copyrights with MIR and AI

Music Information Retrieval (MIR) and AI-driven techniques have played a crucial role in protecting music copyrights in the era of user-generated content (UGC). Traditional peak-based audio fingerprinting methods have been effective in identifying exact copies of music recordings but struggle with detecting transformed versions, such as remixes, covers, or altered pitch and tempo. Recent advancements in deep learning, particularly representation learning approaches, have significantly improved the robustness and accuracy of music recognition systems.

This talk will explore the evolution of music copyright protection techniques, from conventional peak-based fingerprinting to modern embedding-based approaches. We will discuss how deep neural networks learn compact and invariant audio representations, enabling accurate music identification despite signal distortions and transformations. Key topics will include feature extraction, similarity search, and the trade-offs between precision and recall in copyright detection. Additionally, we will highlight real-world deployment challenges, such as scalability, database design, and ethical considerations in automated copyright enforcement.

Beici Liang is a Signal Processing Engineer at Epidemic Sound, specializing in digital rights management with advanced audio technologies. She holds a PhD in Media and Arts Technology from Queen Mary University of London and a BEng in Integrated Circuit Design from Tianjin University. Before her current role, she worked as a Backend and Cloud Engineer at multiple audio startups and as a Research Engineer at Tencent Music Entertainment, gaining extensive experience in large-scale music recognition and recommendation systems. Her work bridges the gap between academic research and real-world industry applications.

Beyond her professional career, Beici is passionate about making music technology more accessible through her popular science writings and active involvement in the MIR community. In 2023, she served as one of the Scientific Program Chairs for the International Society for Music Information Retrieval (ISMIR) Conference.