Machine listening for robust, intelligent, and responsible sound understanding.

My research area is centred on audio and music intelligence, with a broader goal of enabling machines to understand, model, generate, and reason about sound in real-world contexts. The work spans machine listening, acoustic scene and event analysis, audio-language understanding, symbolic music and music generation, expressive performance modelling, speech and voice processing, and healthcare audio. Across these topics, the core motivation is to build computational systems that can extract meaningful structure from complex sound, connect audio with human-centred applications, and support both scientific understanding and practical intelligent systems.

This research area is further strengthened by three connected directions. Robust, trustworthy, and efficient AI addresses the reliability of audio and music systems under domain shift, limited data, computational constraints, privacy risks, and adversarial conditions. Multi-modal and cross-domain perception extends audio intelligence by integrating sound with language, vision, wireless signals, radar, skeleton data, and other sensing modalities. Peripheral review and interdisciplinary work broadens the impact of the research by connecting audio AI, generative AI, signal analysis, education, healthcare, engineering, and applied systems.

Together, these four directions form a coherent research portfolio: audio and music intelligence provides the foundation, robustness and trustworthiness support deployment, multimodal perception expands the scope, and interdisciplinary work connects the research to wider academic and societal contexts.