⚡ electrical engineering

Perceived Annoyance in Multi-source Electric Vehicle AVAS Environments

This paper investigates how the perceived annoyance of electric vehicle AVAS sounds changes in realistic multi-source traffic scenarios compared to isolated single-source conditions through binaural listening tests with spatially arranged, overlapping vehicle pass-by sounds.

Original authors: Berkay Kullukcu, Jonas Krautwurm, Serkan Atamer, Ercan Altinsoy

Published 2026-07-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Berkay Kullukcu, Jonas Krautwurm, Serkan Atamer, Ercan Altinsoy

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you're walking down a city street, and suddenly, a quiet electric car zooms past. It makes a little "beep-boop" sound (called an AVAS) so you know it's there. Now, imagine that same street, but instead of one car, you have two, three, or even a whole parade of them zooming by at the same time. Does the noise just add up like stacking two cups of water to get a bigger cup? Or does the brain react differently when the sounds mix?

That's exactly what researchers at TU Dresden wanted to find out. They set up a virtual traffic jam in a quiet room to see how annoying it is when multiple electric vehicles pass by at once.

The Big Surprise: It's Not Just a Math Problem
For a long time, scientists thought about car sounds like a simple recipe: if one car is annoying, and you add another, the total annoyance is just the sum of the two. But this study suggests that's not how our brains work.

Think of it like a party. If one person is telling a funny joke, it's fun. If two people tell jokes at the exact same time, it might be hilarious. But if they start talking over each other, or if their voices create a weird, jarring rhythm, it can become a headache. The researchers found that when two electric vehicles pass by together, the annoyance doesn't just add up; it often spikes higher than you'd expect.

In their experiment, they played sounds for 10 volunteers (ages 20 to 37-ish) using high-quality headphones. They tested single cars and then pairs of cars.

Single Car Annoyance: When just one car passed, the average annoyance rating was about 46.4 on a scale of 0 to 100.
Two Cars Together: When two cars passed at the same time, the annoyance jumped to 52.9.
The "Perfect Storm": The most annoying combination was a specific mix of sounds (labeled S1 and S3), which hit 58.0.

The team calculated that, on average, adding a second car increased the annoyance by about 6.54 points. This wasn't a tiny, random fluctuation; the math showed it was a real, significant difference.

What About the Volume?
You might think, "Well, maybe two cars are just louder, so of course they are more annoying!" The researchers checked this carefully. They measured the sound pressure levels (the loudness) and found they were almost identical for all the tests, hovering around 72.91 dB (with a tiny wiggle room of 0.77 dB).

Here's the twist: The loudness didn't explain the extra annoyance. Even though the volume was the same, the feeling of annoyance went up. It's like how two people whispering in your ear can be more annoying than one person shouting, even if the total noise level is the same. The brain gets confused trying to figure out where the sounds are coming from, and that confusion makes the scene feel more irritating.

Timing Matters (But Maybe Not How You Think)
The researchers also played with timing. They made one car pass, and then made the second car pass 1 or 2 seconds later.

Simultaneous (Same Time): Annoyance went up significantly.
Delayed (1 or 2 seconds later): The annoyance still went up a bit (about 6.31 points), but the results were a bit more all over the place. Some people hated it, others didn't notice much difference. The team couldn't say for sure that the delayed timing was significantly more annoying than a single car once they did a strict check of the numbers.

What About Direction?
They also wondered if it mattered which car was closer to you. If Car A is in the near lane and Car B is in the far lane, does it feel different than if they swap places? The answer? No. Swapping the lanes didn't change the annoyance rating at all. The brain didn't care who was closer; it just cared that there were two sources of sound mixing together.

The "Pulse" of the Noise
So, what makes the two-car scene so annoying? The researchers looked at "psychoacoustic" features—how our brains process sound. They found a strong link between annoyance and something called Impulsiveness. Think of this as how "punchy" or "jumpy" the sound feels. The more "impulsive" the sound felt, the more annoying it was. It's not just about how loud it is, but how the sound hits your ears.

The Bottom Line
This study suggests that we can't just design electric car sounds in isolation. A sound that is perfectly fine for one car might become a major headache when that car is part of a busy street scene with other vehicles. The "scene" matters more than the individual source.

However, the researchers are careful to note that this was a small test with only 10 people and a limited set of sounds. They didn't test intersections, different speeds, or real-world traffic chaos. They suggest that future work needs to test more cars, more scenarios, and different background noises to really understand the full picture. But for now, the evidence points to a clear idea: in the world of electric vehicle sounds, 1 + 1 is not always 2; sometimes, it's a headache.

Technical Summary: Perceived Annoyance in Multi-Source Electric Vehicle AVAS Environments

Problem Statement
The increasing prevalence of electric vehicles (EVs) in urban environments has altered city soundscapes, particularly at low speeds where propulsion noise is significantly reduced. While this reduction benefits public health, it raises safety concerns for visually impaired pedestrians who rely on auditory cues for vehicle detection. Current evaluation strategies for Acoustic Vehicle Alerting Systems (AVAS) predominantly follow a source-centric approach, assessing individual vehicle sounds in isolation. However, real-world traffic scenarios frequently involve the simultaneous presence of multiple vehicles. The paper argues that the combined presence of multiple AVAS sources may alter perception—specifically regarding annoyance—in ways that cannot be predicted by simply summing the energy of individual sources or by analyzing isolated signals. Existing environmental noise models often address long-term exposure, whereas AVAS perception involves short-duration events with strong contextual cues (motion, spatial location, visibility). Furthermore, it remains unclear whether psychoacoustic models derived from single sounds can accurately account for interaction effects when multiple AVAS signals coexist in an auditory scene.

Methodology
The study employed a controlled binaural listening test to investigate annoyance perception in multi-source scenarios.

Stimuli: The experiment utilized three distinct sound sources: a synthesized sound (S1) and binaural pass-by recordings of two real electric vehicles, a Ford Kuga (S2) and a Ford E-Transit (S3). All sounds were recorded or modeled as constant velocity pass-bys at 20 km/h.
Scenario Construction: Using the virtual acoustics renderer TASCAR, the researchers created a two-lane road environment with opposing traffic directions. Tire-road noise (recorded from a Cupra Born at 20 km/h) was added to all stimuli to simulate a realistic traffic soundscape.
Conditions: Participants were exposed to 13 stimuli: three single-vehicle pass-bys and ten two-vehicle scenes. The two-vehicle scenes varied by spatial arrangement and temporal offset, including simultaneous crossing and delays of +1 second and +2 seconds.
Procedure: Ten participants (5 male, 5 female; mean age 30.7) listened to the stimuli via headphones (Head LabO2) in a quiet room. All AVAS files were maximum SPL equalized. After each stimulus, participants rated perceived annoyance on a continuous 0–100 scale. Each stimulus was presented twice, and ratings were averaged per participant.
Analysis: Statistical analysis included paired t-tests, Wilcoxon signed-rank tests, and Holm correction for multiple contrasts. Psychoacoustic parameters, including the Zwicker loudness model and descriptors like sharpness, roughness, and impulsiveness, were calculated to correlate with annoyance ratings.

Key Results

Increased Annoyance in Multi-Source Scenes: Two-vehicle scenes resulted in significantly higher annoyance ratings compared to single-vehicle stimuli. The mean annoyance for single vehicles was approximately 46.4 points, while simultaneous two-vehicle scenes averaged 52.9 points.
Statistical Significance: A paired t-test confirmed a significant difference between single and simultaneous two-vehicle conditions (t(9)=3.14, p=0.012), which remained significant after Holm correction (p=0.036). The mean within-subject increase was 6.54 points.
Temporal Offset: While delayed (time-offset) two-vehicle scenes also showed a mean increase in annoyance (6.31 points), the difference from single-vehicle stimuli was not statistically significant (p=0.139) due to larger between-subject variability.
Psychoacoustic Correlates: The study found that "Impulsiveness" (derived from a Hearing Model) was the strongest psychoacoustic correlate with mean annoyance (Spearman ρ = 0.823, p = 0.0006).
Non-Additivity and Direction: Annoyance could not be explained by A-weighted sound pressure levels ( $L_{Aeq}$ ), which showed no significant correlation with annoyance (Spearman ρ = -0.154). Furthermore, swapping the approach direction (near vs. far lane) did not significantly alter annoyance ratings, indicating that the specific acoustic dominance of the near-lane vehicle did not drive the perceived annoyance in this setup.

Significance and Claims
The paper claims that its primary contribution is validating a "scene-based" view of AVAS evaluation over the traditional isolated source approach. The results demonstrate that the annoyance of a multi-source AVAS scene is not merely the sum of its constituent parts; rather, the interaction of sources creates a non-additive effect that increases perceived annoyance.

The authors posit that this phenomenon may be explained by auditory scene organization and information masking, where the brain's attempt to organize multiple moving sound sources leads to confusion and increased annoyance. This aligns with soundscape theory, which prioritizes the perception of the entire scene over the analysis of individual sources.

The study concludes that current single-source testing methodologies are insufficient for predicting real-world annoyance in traffic environments. It suggests that future AVAS design and optimization should integrate perceptual tests and psychoacoustic parameters, specifically impulsiveness, to better model multi-source interactions. The authors acknowledge limitations, including the small sample size (N=10), the limited set of stimuli, and the fixed speed, noting that future work must expand the stimulus set to include more vehicle types, background sounds, and complex intersection scenarios with more than two vehicles.

More like this