We propose a novel approach to multicopter localization, using sound landmarks and one embedded microphone. This approach can benefit to multicopter localization in that it requires less computational power and smaller payloads than image-based approaches. However, the high ego-noise of multicopters is a serious threat for sound-based algorithms. We simulated a 2D localization method based on a Kalman Filter using measurements of acceleration and sound landmarks' intensity. A random walk model is used to update the multicopter's position with the Kalman Filter; the calculated estimation is then corrected using noisy measurements from the embedded microphone and accelerometer. Simulation results show that the proposed algorithm can successfully track the multicopter's motion in a noisy environment. We confirmed the effectiveness of our proposed algorithm by comparing its performance and robustness to a time/phase based algorithm.