11:40
Session 12: Human Response to UAS and UAM Noise 2
Chair: Roalt Aalmoes
11:40
20 mins
|
Effects of night-time drone noise on sleep and annoyance
Susanne Bartels, Julia V. Lippold, Marcelo Sanchez Hernandez, Benjamin Aretz, Daniel Aeschbach
Abstract: Compared to conventional traffic noise, noise from drone manoeuvres has a high potential for disturbance and annoyance as recently found in listening experiments. However, the physiological effects of drone noise, particularly in relation to sleep, have not yet been investigated. As there are hardly any restrictions on the use of drones at night, we investigated the effects of drone noise on sleep and annoyance in a laboratory study with 37 subjects aged 20 to 65 years (24 female, 13 male).
In a crossover design, we exposed the subjects in a counterbalanced order during five consecutive nights to a) conventional aircraft noise, b) drone noise from over/passing flights, c) drone noise from take-offs, landings and hovering, d) road traffic noise and e) a control condition (no noise). During each noise night, 84 noises with maximum sound pressure levels of 35 to 60 dB(A) were presented via loud speakers. Sleep was measured via polysomnography. In addition, participants were asked about their perceived sleep quality as well as annoyance and disturbance due to the past night’s noise exposure. The study was double-blind and was announced as an investigation of the effects of environmental noise on sleep and annoyance. Neither the nature nor the timing of the noise events played during the night were known to the participants and experimenters.
Univariate analyses showed higher rates of awakening during nights with drone scenarios compared to the quiet control condition (p < 0.001) and the other noise nights. These findings were confirmed in linear mixed-effects models with random intercepts to account for inter-individual differences in baseline awakening rates. Self-assessed sleep quality was lower in nights with drone noise, while reported nocturnal annoyance and disturbance from noise was increased compared to the other modes of transport and the control condition. The current results highlight the increased potential for psychological and physiological adverse effects from drone noise and the need for early regulation of drone night-time operations to protect the public.
|
12:00
20 mins
|
Comparing a general sound quality metric for aircraft noise using a pre-trained neural network model to traditional psychoacoustics
Thiago Lobato, Tim Kamper-Schley, Marc C. Green, Max W. Ellis, Antonio J. Torija
Abstract: When dealing with sound perception, psychoacoustic is often the go-to solution to produce accurate and interpretable results. However, its meaning is very context-dependent, so that a general psychoacoustics-based sound quality metric for different sound types is challenging. This paper investigates a data-driven alternative to obtain sound-quality ratings of various types of aircraft noise using a pre-trained neural network model. As baseline traditional psychoacoustic approaches are used. The idea of using a pre-trained acoustic model is that it should be able to identify the context a sound is usually present and thus provide better predictions. All approaches are trained on diverse listening-test data from various datasets including diverse drones, airplanes and helicopters. Our psychoacoustic baselines use mainly psychoacoustic descriptors from the ECMA 418-2 standard as features to a regression models with different degrees of non-linearity, namely: linear models, KANNs, and Tree-based models (XGBoost). Those represent also different levels of interpretability from which an accuracy-interpretability trade-off is identified. Our results indicate that pre-trained acoustic representations can provide a more general solution for aircraft-noise sound-quality prediction, while non-linear regression on psychoacoustic features improves performance relative to linear baselines but may remain constrained by the lack of flexibility/contextual cues. These findings support the use of efficient pre-trained models as a practical route to general sound-quality metrics for aircraft-noise assessment. Additionally, if interpretability is desired, a hybrid approach of identifying annoying sounds with a neural network and then performing a fine-grained analysis with psychoacoustic parameters is also a promising approach
|
12:20
20 mins
|
Urban Air Mobility Noise Remote Test Response by Geographic Area and Comparison to In-Person Test
Siddhartha Krishnamurthy
Abstract: This paper discusses results of a remotely administered psychoacoustic test that investigated annoyance to Urban Air Mobility (UAM) vehicle noise. One of the main test results was that test participants residing in areas of low and high noise soundscapes reacted differently to the UAM vehicle noise. Potential reasons for this result are presented via examination of responses to post-test questions. The paper also compares the remote test responses with those from an in-person test that compared conventional lightweight helicopter noise responses to that of noise from a variety of UAM vehicles. A portion of the remote test participant responses produced a span of mean annoyance responses to UAM vehicle noise that was different from that produced by the in-person test participants. Post-test responses are also examined to provide potential reasons for this difference.
|
|