ImAFUSA Pioneers Novel Way to Evaluate Drone Noise

With the ImAFUSA Project concluding on 28/2/2026, the following post was created in coordination with the Delft University of Technology to provide an overview of their work and results in the project.

To better understand the impact Innovative Air Mobility will have on our soundscapes, the ImAFUSA project has developed pioneering methods to measure and model how the sound created by drones and new forms of aircraft will impact people on the ground.

In this interview, we speak with Mirjam Snellen, Professor of acoustic data analysis and imaging for aviation noise at Delft University of Technology (TU Delft). As a key partner in ImAFUSA, her team has introduced a novel, beamforming-based experimental methodology designed specifically to characterize drone noise in outdoor conditions. 

Read on to learn how this technology provides the high-fidelity data essential for psychoacoustic studies and the perception-driven design of future, quieter drones.

The 112 microphone array used by TUDelft during the ImAFUSA soundwalk experiments in Egaleo Park in Athens in the summer of 2024

What exactly did you develop in this work?

Mirjam Snellen: In this project, we introduced a novel beamforming-based experimental methodology specifically designed for drone noise characterisation in outdoor conditions. The proposed technique combines a state-of-the-art phased microphone array with acoustic imaging techniques (also known as beamforming) in both the frequency and time domains. 

This innovative approach enables the isolation of the drone’s acoustic emissions under realistic operating conditions and reconstructs a quasi–free-field sound signal, bridging the gap between field and laboratory measurements.

Your results are described as having an “unrivalled quality of estimates.” What makes them stand out?

Mirjam Snellen: The quality comes from the combination of several factors: the use of advanced beamforming algorithms, a high signal-to-noise ratio, and a dense 112-microphone array. Together, these elements produce extremely clean and spatially- and time-resolved acoustic estimates.

This allowed us to accurately identify the blade-passing frequency (BPF) of the drone’s propellers (related to its rotational speed) and its harmonics, as well as to achieve robust source localisation for frequencies above 1 kHz (corresponding to a Helmholtz number based on the array diameter larger than 3). 

Achieving this level of detail in outdoor drone measurements is truly exceptional, especially in noisy urban conditions like in the experiments conducted in ImAFUSA.

Did your work also reveal anything about how the ground affects drone noise?

Mirjam Snellen: Yes, very clearly. By comparing the raw single-microphone recordings with the beamformed signals, we were able to directly observe and quantify the effect of ground reflections on the measured sound field.

We went one step further by applying the Delany–Bazley ground impedance model, which allowed us to estimate the effective surface flow resistivity of the experimental site. This provided valuable new insights into how terrain and surface properties influence drone noise propagation (and its perception) in realistic outdoor environments.

Why are these measurements particularly well suited for perception studies?

Mirjam Snellen: In addition to conventional noise metrics, our measurements allowed a comprehensive psychoacoustic characterisation of the drone noise. We evaluated state-of-the-art perceptual metrics such as loudness, sharpness, tonality, roughness, and fluctuation strength.

Because beamforming significantly improves signal fidelity and reduces masking from background noise, the resulting dataset is highly suitable for listening tests, annoyance modelling, and perception-influenced design of future drones. In fact, the data directly support the link between physical acoustic properties and how humans perceive drone noise.

You also used SQAT, the open-source Sound Quality Analysis Toolbox, in your analysis. How important was it?

Mirjam Snellen: SQAT played a central role in this work. All sound quality and psychoacoustic metrics were computed using this tool. It proved to be robust, consistent, and very effective for analysing complex, time-varying drone noise signals.

Its use was essential for translating the acoustic measurements into meaningful perceptual information, and we were very satisfied with its overall performance. It also guided the development of the ImAFUSA noise tools.