Jeong-Woo Choi

Professor Jeong-Woo Choi is a faculty member in the Department of Electrical Engineering at Korea Advanced Institute of Science and Technology (KAIST), specializing in cutting-edge audio artificial intelligence research. His work bridges electrical engineering and machine learning with applications in industrial monitoring and immersive media technologies.

His primary research focuses on Audio Signal Processing and Sound Source Separation, developing next-generation AI systems capable of isolating and classifying complex soundscapes. Key areas include spatial semantic segmentation of acoustic scenes, multichannel signal analysis, and transformer-based neural architectures for abnormal sound detection in drones, factory pipes, and border surveillance systems. His team pioneers techniques mimicking human auditory processing through waveform analysis and directional cues.

Professor Choi's team secured first place in the 2025 IEEE DCASE Challenge against 86 global competitors, achieving unprecedented 11 dB performance in sound separation metrics. This breakthrough enables advanced applications in AR/VR spatial audio editing and industrial anomaly detection.

His scientific recognition includes:

• First place in 'Spatial Semantic Segmentation of Sound Scenes' at IEEE DCASE Challenge 2025

He actively mentors graduate researchers including PhD candidate Kwon Young-hoo and Master's student Kim Do-hwan. His lab operates under significant funding from South Korea's National Research Foundation (Mid-career Researcher Support Project), Ministry of Education (STEAM Research Project), and Defense Acquisition Program Administration (Future Defense Research Center).

The research team maintains KAIST's leadership in audio AI through their specialized laboratory within the Electrical Engineering department, recently developing the world's highest-performance sound separation model combining Transformer and Mamba architectures.