Description

Sound event detection and extraction form the foundation of a growing range of automated systems aimed at improving quality of life. Recent advances in AI—particularly
neural network models—have rapidly expanded real-world applications, offering transformative possibilities across domains. At the core of this event lies a shared challenge:
enabling accurate, robust, and efficient sound interpretation in complex, real-life environments. From AI-aided home stethoscopes for non-invasive health diagnostics, to passive
acoustic monitoring for animal welfare, to urban soundscape analysis supporting public health—these applications all rely on reliable sound event detection. 

Achieving this goes beyond accuracy. Neural models must recognize meaningful patterns across diverse scenes, devices, and noise—learning to interpret sound in ways aligned with human perception and practical needs. This calls for context-aware systems, not just low-level signal processing. At the same time, growing AI capabilities require sustainability. Balancing model accuracy, size, and computational demands is crucial for scalable, low-impact technologies.

This event brings together experts at the crossroads of acoustics, AI, and domains like medicine, veterinary science, and urban health. Despite different applications, they face
common technical challenges. The goal is to create an open forum for exchanging ideas, methods, and questions—fostering collaborations that advance the field.

This event precedes the PhD defence of Tomasz Grzywalski on Advanced Techniques for Isolating Sounds of Interest in AI Listening.