Adaptive Ambiance Mode For Noise Cancelling Headphones

This paper introduces a novel and robust model for ambiance mode in noise-canceling headphones. Prior to this study, the focus in this domain has primarily been on Active Noise Cancellation as a comprehensive solution to eliminate unwanted sounds. However, the ambient mode has traditionally been regarded as a conduit through which most surrounding noises pass through. Although a few techniques, such as Transparent mode, Voice pass, and sound control, exist for ambient modes, these approaches mainly aim to modify the level of noise cancellation and not the type of noise that a user must hear. To address these limitations, the present study proposes an Adaptive Ambiance Mode that leverages deep learning to classify audio signals and, based on the context, turns the Active Noise Cancellation on or off for a particular interval of time. In this regard, the paper categorizes ambient modes into three categories: Street Ambiance, Workspace Ambiance, and General Ambiance. A neural network is employed to classify the sound signals into three groups, yielding an accuracy of 93%. The Active Noise Cancellation component is implemented using the Least-Mean-Squared algorithm, which is highly effective, achieving a Karl Pearson's coefficient of correlation of 96.51%.