An ultra low power wake-up detector based on frequency analysis is presented in this paper. Focused on environmental or military Internet of Things (IoT) applications, it aims at detecting in real time the presence of specific animal species or drones for generating alerts and for triggering power consuming tasks such as high frequency signal recording only when needed. This wake-up detector continuously monitors the presence of specific frequencies in an analog signal, with a good frequency selectivity and a high frequency detection capability. It is based on an ultra-low analog frequency to voltage converter using a current-mirror, analog timers and comparators. Dedicated to long term stealth environmental or military surveys, a strong emphasis has been put in power consumption reduction in order to limit size and weight of the system. This power consumption has been reduced to 34μW, leading to a full year of autonomy including the microphone when powered by 3 coin cell CR2032 batteries.