What should electric vehicles sound like?
Electric vehicles are really quiet, and more dangerous to pedestrians as a result. Here I take a human factors approach to understand what electric vehicles should sound like.
A few years ago now, transport regulators around the world started to ponder the dangers posed by quiet, electric vehicles on unsuspecting pedestrians. At the time the risk was low - the uptake of electric vehicles was small.
But I think it's fair to say the electric vehicles (EVs) of today, made by the likes of Tesla, have massively increased in popularity, desirability and numbers. EVs are now creeping into the mainstream, and creeping around our roads, bringing upcoming American regulations to the fore.
In response, design agency UsTwo have recently published some findings from their investigation into what EVs should sound like, in a way that might improve pedestrian safety AND reduce noise pollution AND allow for brand expression - phew!
Intrigued, I revisited some of my own thoughts on this subject from a few years back and fleshed them out. I did this to join the debate, and hopefully to show the benefits of taking a human factors approach - like the one I would use in my day to day work. So I've structured my thoughts around the model used to describe Situational Awareness, which is a core concept used in many human factors projects.
Situational Awareness is:
“The perception of elements in the environment within a volume of time and space, the comprehension of their meaning, and the projection of their status in the near future.” Dr. Mica Endsley
Situational Awareness (SA) is also a really important part of effective decision making, especially in complex and dynamic situations. So using this model should primarily help to identify what an EV should sound like to help keep pedestrians safe.
SA is broken down into three levels, which I'll describe in turn, and the insights they bring for identifying user needs.
This is the most basic level. At Level 1, pedestrians are detecting sounds, among other cues, to start building up a mental picture of all the individual elements in their surroundings. For example, pedestrians are using sound to identify vehicles, other pedestrians, as well as any environmental factors such as weather, and locating them.
This highlights three important sound characteristics, or user needs, for the sound of EVs; Detectable, Recognisable and Locatable.
Obviously, to have any beneficial impact on safety at all, pedestrians must be able to hear EVs in their vicinity (hence the regulations).
While that's a simple notion, how loud is loud enough? How loud is too loud?
The sounds must be discernible under a wide range of background conditions, from quiet rural locations to busy urban environments, as well as under adverse environmental conditions when noise from wind and rain can potentially mask the noise of approaching traffic.
You could consider adding ambient noise level sensors to cars, and varying the EV sound level accordingly, but it's important to remember that the regulations aren't about making EVs heard above EVERYTHING ELSE. It's not a competition, but it’s important to make sure that they are detectable at all.
We are all taught at an early age that cars are dangerous, and to listen out for them when crossing the road.
In this spirit, I think it's important that EVs sound, at least a bit, like the other vehicles sound today.
That's not to say I think EVs should sound artificially mechanical. Rather, I think the space for innovation here is more about branding. Many people appreciate and enjoy the sound made by engines - Ferraris, Harley Davidsons and so on, and the sound made by these engines expresses, among other things, the engineering excellence embodied within it. Likewise, I think the sound of EVs should speak to the technology used, and the values that EVs represent, while retaining something of the familiar thrum of traditional engines.
I can imagine that this seems rather conservative, but let me try another way of putting this; if something looks like a car, and (importantly for this discussion) sounds like a car, then by using familiar cues pedestrians will quickly and instinctively recognise the something as a car - and behave accordingly. This sort of heuristic driven behaviour is sometimes known as 'System 1' thinking, which is fast and subconscious (unlike 'System 2', which is slow and effortful).
So it follows that certain sounds should also be avoided, because they won't easily be recognised as vehicles. In fact, the equivalent Japanese commission looking into this issue specifies a range of sounds that electric vehicles should not make, such as insects, waves or the wind.
I would go further, and research the soundscape, to identify any other sounds that pedestrians encounter, and avoid using them. The bleeps of pedestrian crossings, the wails of emergency vehicle sirens and jingle jangle of ice cream vans spring to mind.
As well as needing to recognise the sound as a vehicle, pedestrians need to judge where the vehicle is, relative to them.
One technique used to help drivers better locate the direction of approaching emergency vehicles is a short burst of white noise made in between wails of the siren;
A long-standing problem for emergency services has been traffic being unable to determine the direction a siren is approaching from, and different tones have been developed on some electronic sirens to help combat this, such as the use of white or pink noise in between more conventional siren noises, which helps people to pinpoint their origin - Wikipedia
Perhaps there is an opportunity to use a similar principle in the sounds made by EVs, though it would be important to test how well this would scale-up to a situation with multiple cars. I think it’s also important to consider what impact this might have on which way the speakers face.
Level 2 SA: COMPREHENSION
Comprehension is the second level of SA. At level 2, pedestrians start to synthesise all the individual bits of information together, to complete the picture of what's happening in their surroundings.
For example, pedestrians will need to know if certain vehicles are getting closer or further away. This helps to decide which elements are important and need to be monitored.
This highlights a fourth characteristic, or user need, for the sound of EVs; Scalable.
I think that pedestrians understand that cars sound louder as they approach, and sound quieter as they travel further away. It's simple, intuitive and scales well.
This is important because it steers me away from the idea of using other, more complex, sound based cues such as changing pitch or repeating frequency, to indicate that a vehicle is approaching. Plus, there are some strange things that happen with sound when vehicles move around you, such as that weird perceived change in pitch you hear when an ambulance goes past (the Doppler Effect).
The idea of using a repeating sound at increasing frequency in response to the proximity of a pedestrian seems to carry the potential risk of confusion; is that sound coming from 1 car that's close to me, or multiple-cars slightly further away, or cars close to other pedestrians near me?
So it's also really important to consider how effective these sort of sounds would be, when scaled-up to multiple cars and multiple pedestrians.
I think the most practical and safe thing to do is keep the sound simple, and consider scale.
Projection is the the third level of SA, and is this where the pedestrian's judgements about the vehicle location, direction and speed are translated into predictions about what will happen next. Their prediction has an impact on the decisions they make as they try to achieve their goals - for example, safely crossing a road.
Pedestrians need to make projections about the path of a vehicle, the change in speed, and so on. This highlights the fifth characteristic, or user need, for the sound of EVs; Predictable.
To identify sounds that can help pedestrians predict what a vehicle will do next, it's a good idea to listen to what conventional combustion engines cars do now.
This made me think of the little audible cues that I might use to try and predict what the cars around me are going to do next, such as when;
The driver uses the accelerator to set the revs above the biting point in preparation to set off (the sound pitch gets higher).
The driver lifts off the accelerator, the revs drop, indicating that a car could be slowing down (the pitch lowers).
I think the opportunity is to emulate these sort of audible cues, for EVs.
I also wonder if it's strange to use sounds that change based on a principle of several separate levels, rather than sounds that change gradually because cars don't move in a discreet way - it's continuous.
I think the situational awareness model is a useful framework for structuring the thought process, as well as generating ideas for what sounds might work and what sounds to avoid. However, it's fair to acknowledge here, that by itself situational awareness cannot guarantee safe outcomes, though it does provide the foundations for good decision making.
I also found it useful to think about ‘decision making’ rather than ‘risk perception’, because this reinforced the need to design for ‘System 1’ style thinking which is fast and intuitive.
So to sum up what might constitute some basic Guidance Notes (pun absolutely intended), I think EV sounds should be;
Detectable - loud enough to be heard above the background ambient noise.
Recognisable - a bit like other vehicles sound today.
Locatable - help pedestrians judge vehicle location in relation to them.
Scalable - carefully consider how the sound might scale for scenarios with multiple cars and multiple pedestrians.
Predictable - use audible cues to help pedestrians predict what the vehicle will do next.
The last four user needs are essentially covered more broadly by just one principle; conform with pedestrian expectations.
Together, these essentially form a hypothesis about what would make a good sound for EVs. It’s based on theory, but needs to be tested to understand the reality. So the next steps I’d take would be to test the theory using a range of scenarios and different conditions. I’d want to research into pedestrian’s expectations for EV sounds, gather feedback on potential solutions and prioritise the most important factors. I loved the virtual reality assessment used in the ustwo study - you should really check this out in their post!
So, come on, what specific sounds should EVs make?!
Well, I don't think we can expect car manufacturers to collectively agree on a single sound for EVs. Branding is far too important for that to happen.
And I'm not sure we even want that sort of homogeneity anyway. Collectively, the current soundscape of engine noises, are a kind of mess. My hunch is that the mess of existing sounds might actually help pedestrians, even if just in tiny little ways. Say, picking out the difference between the rickety old banger that's lurching towards them on one side of the road from the Prius gliding by on the other and from the booming Ferrari following it. Maybe we should keep something of that?
Sounds like an electric car
