Joint Estimation of Primary and Secondary Paths for Personalized Hearable Applications

Hearable devices are increasingly expected to deliver advanced features such as active noise control (ANC), transparency, and personalized equalization. Accurate estimation of the primary and secondary paths, determined by the user’s ear–device coupling, is essential for personalized performance. Existing approaches rely on offline calibration, auxiliary noise injection, or data-driven mappings, but these are impractical for commercial hearables due to user discomfort, limited adaptability, or computational complexity. Joint estimation is further complicated because the input of one path interferes with identifying the other. We propose a practical framework that simultaneously estimates both paths under realistic conditions, without auxiliary excitation or offline training. Experiments with commercial earbuds on a head simulator demonstrate that the proposed method achieves accurate path estimation and more consistent ANC performance across multiple re-insertions than a nominal fixed filter, showing its potential to enable personalized hearable applications.