Maritime surveillance and safety rely heavily on situational awareness, often hindered by limitations in traditional visual and radar-based systems under challenging environmental conditions. This research proposes a self-adaptive system based on organic computing principles to enhance perception in autonomous ships through audio classification and localization. By leveraging advanced machine learning algorithms, the system autonomously detects, classifies, and localizes sound sources, such as vessel signals and environmental sounds, even if characterized by high noise and dynamic conditions. Integrating this adaptive layer into existing frameworks for perception models offers a robust approach to improving situational awareness, demonstrating how organic computing principles can address complex, evolving challenges in maritime environments.