A Pilot Study on the Acoustic Signal Processing at a Small Aerial Drone

This contribution deals with the acoustic signal processing at an unmanned aerial vehicle (UAV, aerial drone) to gear up for intuitive user interfaces including speech commands. The sound and speech signal processing at flying drones is complex because of the rotor and related noises. The signal processing has to consider various sound sources, such as noise which disturbs the analysis or “useful signals” like speech commands and other external signals that might characterize environmental objects. The article describes simple measurement scenarios to directly record acoustic signals including noise at a flying or an affixed drone with external microphones in different positions, based on indoor and outdoor experiments, to analyze the influences of microphone position, flight maneuver and the environment in the spectrogram or power density spectrum. The blade passing frequency (BPF), the motor rotation speed, and the associated harmonics pose the dominating effects in the drone-recorded signals, which are post-filtered to increase the signal-to-noise ratio (SNR) for different use cases, e.g., to feed a speech recognizer with word commands. The chapter concludes with a draft conception for an improved acoustic analysis by integrating a microphone array.