Indoor Localization has been one of the most extensively researched topics for the past couple of years with a recent surge in a specific area of Device-free localization in wireless environments. Particularly FM-radio based technologies are being been preferred over WiFi-based technologies due to better penetration indoors and free availability. The major challenges for obtaining a consistent and highly accurate indoor FM based system are susceptibility to human presence, multipath fading and environmental changes. Our research works around these limitations and utilizes the environment itself to establish stronger fingerprints and thus creating a robust localization system. This novel thesis also investigates the feasibility of using neural networks to solve the problem of accuracy degradation when using a single passive receiver across multiple ambient FM radio stations. The system achieves high fidelity and temporal stability to the tunes of 95% by utilizing pattern recognition techniques for the multiple channel spectra.