This study is focused on advancing the application of structure-borne sound sensors in the manufacturing of micro gears. A robust approach is developed through the incorporation of machine learning techniques to address concept drift, especially resulting from tool wear during the hobbing process. A detailed long-term study is undertaken to refine the predictive models, ensuring their accuracy in the face of a dynamic manufacturing environment. By tackling the challenges associated with concept drift, the capabilities of acoustic emission sensing are sought to be fully exploited. The aim is to enhance the reliability and precision of quality predictions, establishing a new benchmark in the field and significantly contributing to the optimization of micro gear production processes.