Sound Quality Directivity Characterization of a Hovering Drone Using Microphone Array Measurements

Abstract

Unmanned aerial vehicles are increasingly considered for urban operations, yet their noise emissions remain a key limitation due to their directional and perceptually complex character. This paper presents a three-dimensional acoustic and psychoacoustic characterization of a quadcopter drone measured under controlled free-field conditions in an anechoic chamber. A 112-channel phased microphone array was used to record the drone sound radiation for 19 azimuth angles and 11 vertical positions (polar angles). The overall sound pressure level (OSPL) and its A-weighted counterpart (OASPL) were employed to quantify the acoustic directivity and conventional frequency-domain beamforming was used to generate two-dimensional source maps at selected one-third-octave bands. Moreover, sound quality metrics, together with a psychoacoustic annoyance model, were evaluated to examine the directional dependence of perceived sound quality. The results show that energy-based metrics exhibit a dipole-like directivity pattern, characterized by a quasi-axisymmetric bi-lobed dependence with respect to the polar angles. The lowest and most nearly-isotropic radiation occurs close to the rotor plane (i.e., polar angles close to zero), and higher levels occurring as the polar angle is steered away from the drone's horizontal plane. The beamforming source maps reveal compact sources associated with the rotors at higher frequencies. Regarding perception, loudness and psychoacoustic annoyance broadly follow the OSPL and OASPL trends. In contrast, sharpness shows a more diffuse angular dependence. Tonality, roughness, and fluctuation strength show localized maxima, indicating that the loudest directions are not necessarily the most perceptually salient or annoying. Furthermore, the perceptual minima do not coincide with the minimum OSPL and OASPL directions. These findings confirm that purely energy-based metrics are insufficient to fully characterize the perceived nature of drone noise and contribute to a more complete characterization of drone noise for source identification, auralization, and mitigation-oriented design.