Feedback Active Noise Cancellation Using Single Sensor with Deep Learning

Constructive measures should be taken immediately to tackle urban noise pollution, which is an omnipresent but neglected threat to human health. Many attempts have been made by researchers during recent decades to alleviate this issue; however, due to the nature of linear filters, conventional active noise control (ANC) methods, for example, filtered-x least mean square (FxLMS) algorithm, are useful just for attenuating narrowband linear or tonal noises. To deal with environmentally complex ANC applications, we developed a new deep learning-based artificial intelligence algorithm, which is able to model the intrinsic nonlinear behavior of various noises and produce anti-noise, which destructively interferes with unwanted noise to neutralize it. The proposed algorithm as a feedback controller significantly outperformed the traditional feedback FxLMS method in terms of noise attenuation metric.