Abstract:

This talk presents a unified framework for Intelligent Urban Noise Mitigation, emphasizing the integration of AI-driven acoustic sensing with advanced Active Noise Control (ANC) strategies. Conventional ANC systems, while effective in confined environments and controlled conditions, face limitations in dynamic, spatially complex urban settings due to challenges in acoustic path modeling, slow adaptation, and lack of context awareness. To overcome these limitations, the speaker introduces an AI-enhanced front-end that performs real-time noise classification, source localization, and environment interpretation. This “AI listens, learns, and cancels” paradigm allows ANC systems to respond intelligently rather than applying uniform noise suppression.

A key contribution discussed is the Selective Fixed-Filter Active Noise Control (SFANC) approach, where a lightweight neural network identifies the dominant noise characteristics and rapidly switches among a library of pre-trained control filters, achieving fast and stable noise reduction with low computational overhead. The talk further explores Generative Fixed-Filter ANC (GFANC), which employs generative AI models to synthesize new control filters on demand, enabling adaptability beyond a finite filter bank.

Finally, the talk highlights a complementary direction: Acoustic Soundscape Augmentation, which shifts focus from purely reducing sound pressure levels to enhancing perceived acoustic comfort. By selecting and spatializing pleasant natural sound maskers through human-response-trained models, this system improves the experiential quality of urban environments. Together, these approaches illustrate a pathway toward scalable, context-aware, and perceptually aligned urban noise mitigation strategies. 

[1] B Lam, WS Gan, DY Shi, M Nishimura, S Elliott, “Ten questions concerning active noise control in the built environment,” Elsevier Building and Environment Vol 200, Aug 2021, 107928, https://doi.org/10.1016/j.buildenv.2021.1079281] 

[2] ZD Luo, DY Shi, WS Gan, “A Hybrid SFANC-FxNLMS Algorithm for Active Noise Control based on Deep learning,” IEEE Signal Processing Letters, Vol 29, 2022, pp1102-1106.

[3] ZD Luo, DY Shi, WS Gan, "Delayless Generative Fixed-filter Active Noise Control based on Deep Learning and Bayesian Filter," IEEE Transaction on Audio, Speech, and Language Processing (2023)