Drone audition aims to carry out sound source localisation and enhancement from audio signals captured by on-board microphones. However, a central challenge is that the considerable amount of noise from the drone's propellers (ego-noise) severely impacts the localisation and enhancement performance. In this chapter, a new mechanical configuration is proposed for signal capture that positions an 8-microphone array away from the drone to reduce ego-noise, by hanging it from below with a fixed-length cable. To reduce its influence to the drone's manoeuvrability, the array can be driven by a winch to pull it back to the drone if needed. The viability of this configuration is established by using SRP-PHAT for sound source localisation, and MVDR for speech enhancement, with both artificially mixed data and real-life captured experimental data. To this effect, this new mechanical configuration provides a wide array of potential benefits for drone audition.
F. Proulx, L. Oyuki Rojas-Perez, C. Rascon, J. Martinez-Carranza, F. Grondin. Drone Audition Using a Suspended Microphone Array. Handbook of Intelligent Robots. CRC Press, 2026. 325-339.