Why a Medical Label Makes Drone Noise Less Annoying: New Listening Experiment

The hum of a delivery drone overhead may soon become as familiar as the rumble of a passing van, but whether people welcome that sound or resent it could depend less on decibels than on what they believe the drone is doing. A new listening experiment built around the United Kingdom’s first national medical drone network has found that simply telling participants a drone is carrying vital medical supplies significantly reduces how annoying its sound feels, an effect that proved stronger than any acoustic factor tested.

The study, published in Frontiers in Acoustics, was conducted as part of Project CAELUS, an initiative developing a drone distribution network to transport medicines, blood, and diagnostic materials across Scotland, particularly to remote communities. Noise has been identified by the European Union Aviation Safety Agency as the second most significant barrier to urban air mobility adoption after safety, yet empirical data on how people respond to drones in a medical delivery context has remained scarce. The research team, led by engineers at Arup with AGS Airports, set out to fill that gap with one of the largest drone noise perception experiments conducted to date.

More than 700 participants took part in the online listening tests, with 425 rating drone overflights and 278 rating take-off operations. Each participant heard a series of 45-second audio simulations of the project’s electric vertical take-off and landing drone, a lift-and-cruise type aircraft with a cruising speed of 55 knots and a maximum take-off weight of 17 kilograms. The stimuli were presented against three different ambient soundscapes representing remote rural, rural village, and urban environments, and at three different distances between listener and drone.

Creating realistic drone audio for the experiment required considerable acoustic engineering. Field measurements were carried out at the Westcott Innovation Centre in southern England in August 2022, with the drone’s position tracked by GPS and its sound captured by an array of calibrated microphones. Because background noise contaminated the overflight recordings, the team could not simply play them back. Instead, they built a sophisticated auralisation pipeline: the recorded signals were de-Dopplerised in the time domain, decomposed into blade-pass frequency harmonics, motor tones, and broadband airframe noise, and then resynthesised as clean virtual microphone signals.

Those synthesised signals were then reconstructed at different listener positions by applying frequency-dependent directivity, Doppler shift, geometric spreading, atmospheric absorption, and amplitude modulation from air turbulence. The team even modelled the small random variations in rotor speed that give real drone sound its characteristic texture, deriving the statistical properties of those fluctuations from the measured recordings. The final audio was encoded in third-order ambisonics, mixed with authentic soundscape recordings, and rendered binaurally over headphones. Expert listeners confirmed the result sounded convincingly real, and spectral comparisons showed the synthesis faithfully reproduced the original drone’s acoustic signature.

Participants rated their annoyance on a seven-point scale using the standardised question from ISO/TS 15666: how bothered, annoyed, or disturbed were they by the sound. Half the participants received context before the experiment, entering through a portal explaining that the drones would deliver healthcare services across Scotland. The other half took the test with no information about the drone’s purpose, with all CAELUS branding stripped away. Statistical analysis used aligned rank transform ANOVAs, a non-parametric method suited to the ordinal rating data, with effect sizes estimated using Cliff’s Delta.

All three factors produced statistically significant effects, with p-values below 0.01. Annoyance decreased with increasing distance from the drone, whether that distance was expressed as altitude for overflights at 60, 90, and 120 metres or as proximity to the take-off area at 30, 60, and 120 metres. Annoyance was consistently highest in the quietest, remote rural soundscape, followed by the village and then the urban environment. But the contextual framing effect was the largest of all: knowing the drone was on a medical mission substantially lowered annoyance compared with hearing the same sounds with no explanation.

The interaction effects added nuance. Annoyance fell more gradually with distance in the rural soundscape than in the village or urban settings, suggesting that background noise in busier environments masks the drone and reduces its noticeability at greater distances. In the remote rural setting, the drone remained clearly audible across all altitudes tested. A supplementary analysis using the difference between drone and ambient sound levels as a predictor found that annoyance increased roughly linearly up to a level difference of about 15 decibels before plateauing, and that this single acoustic measure, together with operation type and context, explained 38 percent of the variance in annoyance ratings.

The findings align with a growing body of evidence that drone noise is perceived as more annoying than road traffic or conventional aviation noise at equivalent exposure levels, and that take-off and landing operations tend to be judged more annoying than flyovers. They also echo earlier work showing drone noise stands out more in quiet environments. What makes this study distinctive is the strength of the context effect. Previous research on contextual priming had produced mixed results, with at least one study finding no significant effect of medical framing. The authors suggest the strong effect here may stem from the depth of information provided, since participants in the context condition accessed a full project portal rather than a single sentence of instruction.

The practical implications reach well beyond the laboratory. The researchers argue that community engagement, transparent communication about what drones are doing, and even visual cues such as the universally recognised red cross displayed on medical delivery drones could meaningfully improve public acceptance of drone networks. Route planning also matters: avoiding flights over quiet rural communities, or scheduling operations when ambient noise is higher, could reduce the intrusion of drone sound. As Scotland prepares to fly blood and medicines between hospitals under regulated airspace, the study offers a data-driven reminder that winning hearts and ears may be as much about storytelling as sound engineering.

Subject of Research: Human annoyance response to eVTOL medical delivery drone sound

Article Title: Human response to eVTOL drone sound: an online listening experiment exploring the effects of operational and contextual factors

Article References: Woodcock, J., Thomas, A., Hiller, D., Maldonado, A. L. P., McLeod, L., Sharp, C., & Smith, F. (2025). Human response to eVTOL drone sound: an online listening experiment exploring the effects of operational and contextual factors. Frontiers in Acoustics, 3, Article 1624669. https://doi.org/10.3389/facou.2025.1624669

Image Credits: AI Generated

DOI: 10.3389/facou.2025.1624669

Keywords: drone noise, eVTOL, noise annoyance, soundscape, Project CAELUS, medical delivery, psychoacoustics, auralisation, advanced air mobility, public acceptance, listening experiment, non-acoustic factors