My involvement in automotive sound design began back during my college years while I was working in the Active Noise Control department – HALOsonic – of Harman International. I was chosen to lead a newly developing project due to my intimate experience with synthesizers. They had been an absolute passion in my life for many years prior to this. I initially became interested in synthesizers because of my interest in electronics and artists such as Aphex Twin, Nine Inch Nails and Boards of Canada. I’ve always found the range of emotions that a synthesizer can cover to be an endless journey. My first analog synthesizer was a Juno 106 which my mother got for me at a yard sale. I’ll never forget the joy of plugging it in and hearing it for the first time.
Surfing on Sine Waves
An inspiration who I give much credit to is Alessandro Cortini. His music was always incredible to me and seemed like the closest thing to mind reading. I was able to connect with it so deeply, as if it were a soundtrack to my own life. One day, I read a sticker placed on the sleeve of his Forse 1 LP that said that the entire album was made with only with one synth – the Buchla Music Easel. I then found out that all his records are created using only one synth and that was his defined approach. His theory is that by limiting your options you effectively become more expressive with less. This was the pivotal moment that made me realize that synthesizers are deeply structured machines worthy of more exploration than just hitting a pre-set, (thank the DX7 for that). He was the first real “student” of synthesis that I had come across taking me even further down that glorious rabbit hole.
Alessandro Cortini Forse 1 LP.
Intro to Automotive
In 2014, I was working for Harman International in their Active Noise Control department and was exposed to the various ways sound design was involved in automotive engineering. One area that was already very active was interior sound generation. That year I was tasked with leading a project at Harman for a new technology called AVAS (Acoustic Vehicle Alert System). AVAS enables near-silent hybrid and electric vehicles to emit an external sound which is controlled by the vehicle’s speed over a range of 0-30 KPH to better aid pedestrians in avoiding dangers. There had just been a new legislature proposal (FMVSS-141) which declared the legal minimum requirements of sound which these quiet vehicles needed to meet. Many people initially scoffed at the idea, but in the research, it shows that a large amount of injuries and fatalities can be avoided by implementing this law. It is especially necessary for those who are visually impaired or elderly. Throughout that year and many years to come, I was involved with different OEM’s on researching this technology and what would be needed to implement it on a production vehicle.
There is some major synchronicity in the fact that I am working with Audiokinetic in automotive applications today. When I first started out my research and development plan for AVAS, I realized that the video game industry has done this very thing for decades. My research then took an obvious shift into what had been done in game studios for the huge amount of racing games that are out there. I took note of the different sound designers involved, their approaches and the technology used. I then had the idea to reach out to a long-time inspiration of mine in music and synthesis, Richard Devine. When I reached out to him he was extremely humble, helpful and receptive to my ideas and pleasantly informed me that he had just completed some of the very same work for an automotive OEM. We shared ideas and approaches, and it was the best foot forward into developing this new technology that bridges the gap between sound design and automotive engineering. This is when I first learned of Wwise and the greatly diverse field of interactive sound design. It was exciting and eye-opening.
History and Advancement of Car Sounds
The history of sound design in automotive is deep, but in a not-so-obvious way. The acoustics of a car is engineered in more ways that most people realize. Mufflers are tuned as band-pass filters which attenuate and amplify desired frequencies. The engine and intake have a natural acoustic signature which is then engineered to fit certain spectral patterns and sound pressures. In high quality vehicles the interior of the cabin is quite isolated from both wind and road noise. In many modern vehicles, certain frequencies are played back inside the cabin to give back some of the feedback from the engine that is missing due to the noise isolation. Many automotive OEMs use this same technique to make a four-cylinder engine sound like a six, eight- or twelve-cylinder engine. In some cars, when switching from ECO to Sport mode, sounds or tones are played back in the cabin to bring the “sport” vibe back and enhance that mode of driving. Many “purists” scoff at the idea of synthetically generated sounds, and so some manufacturers use a physical source in the engine and create a channel to the cabin to transfer the desired sound waves. This has been done for some time with the Chevy Camaro, by way of a resonant tube from the air intake, through the firewall, and into the cabin. Another technique that has been used is placing a microphone inside the engine bay and piping the sound to the loudspeakers, similar to what some researchers are currently doing with electric motors. One area that I can see coming up is the exploration of using electromagnetic transducers (such as the LOM Elektrosluch) and capturing the electromagnetic field surrounding these motors. It seems like a good way to capture speed and power information while keeping the “purists” happy with a natural source, although more research needs to be done on extracting relevant sound information and conveying it to the driver and/or pedestrians.
LOM Elektrosluch 3+ and Priezor Antenna
Now with the advancement of electric vehicles, there is a new blank canvas to create sounds. Electric engines do make some sound, but nothing quite as loud or as easily identifiable as a combustion engine. There is much research out there looking into the acoustic signatures and properties of electric engines and how they might naturally be useful for conveying vehicle behaviour as the combustion engine does. They are extremely quiet in comparison to the level of road & wind noise present inside a vehicle, and so some sort of transduction and amplification method is usually employed in these techniques with the addition of a dynamic filtering.
The near silence of these new electric vehicles creates problems and opportunities. The first problem is that it is dangerous to have a massive object travelling at high speeds in virtual silence. The second problem is that the feedback sound from the engine is now missing, taking with it much of the feel and excitement of driving. These problems are also opportunities to create brand-centric sounds that aid the identity of the vehicle and the right user-experience since brand identity is a major influencer in automotive design. A big concern for many OEMs and researchers is the question of whether or not this new law for external sound would cause harmful noise pollution and annoyance. The answer to this is that the specifications set out in the law is a result of years of research on the ambient noise levels on roadways and their effects. The required sound levels for pedestrian warning sounds (AVAS) are very close to the average ambient noise levels of our roadways and will not contribute noise pollution problems. The law will only affect those who are within dangerous distances from vehicles in an attempt to save lives and avoid injuries.
Wwise in Action
Wwise has been everything I’ve ever dreamt of for automotive sound design. It has been an excellent toolbox for creating complex sounds and coupling their parameters to the real-time vehicle physics. The ability to combine samples, sine waves, complex waveshape oscillators, granular synthesis, effects, and control almost every parameter over the entire vehicle drive cycle, allows for a huge range of sound design options. One does not need to spend much or any time at all in a DAW to create their vision anymore, it can all be done in Wwise. It also facilitates the multitude of different design strategies used by automotive engineers and designers.
Soundcaster simulation session for AVAS and Internal Sound (ESG) tuning.
The flexibility of creating a simulation timeline over the range of vehicle parameters lets you test out how the designs will work in any scenario, which is quite large in an automotive context. Using the Soundcaster as you would normally with simulating game behaviours and parameter interactions is exactly what is needed in this context, since more often than not you will have to perform much of the designs on the bench without a working vehicle. User experience development is very active in the early prototype phases of the automotive design cycle. Below is an image of the newly developed simulator tool for Wwise automotive which enables the playback of real captured CAN data from a vehicle’s CAN bus in order to tune with accurate data. This is another crucial step bringing the designers even closer to a real vehicle.
AVAS CAN playback.
Interactive (manual control) capabilities.
Moving from bench designs to an actual vehicle is then very easy and consists of just making sure the CAN parameter ranges (speed, rpm, pedal position) are the same and smoothing is applied to the CAN data, if necessary, to arrive at the same responses that were attained in the Soundcaster simulation. There’s always some tweaking to be done, but nothing too extraordinary. The limitations of much of the previous and current design strategies used for AVAS and interior were very much due to the linear nature of the DAW tools used. Having an interactive and simulated environment like Wwise allows sound designers to completely break free from those limitations.
Speed simulation timeline for AVAS acceleration to deceleration.
There are three approaches that OEMs can take on creating external or internal engine sounds. They either want a traditional type of sound that is similar to a combustion engine, a sound that is completely new since these vehicles are not combustion engines at all, or some combination of the two. With Wwise, all of these approaches can easily be achieved to meet whatever vision a car maker has. One joke that has been made often in the industry is “putting combustion engine sounds on an electric vehicle is like putting horse sounds on the first combustion engine cars.” While I do agree that it may be out of place and strange, the sound from a combustion engine is very iconic, informative and easily identifiable due to being present within our society for so long. Retaining some of the elements of combustion engines might prove beneficial for the purpose of alerting pedestrians of an approaching vehicle.
East Coast vs. West Coast
As per the Fourier theory, any complex waveform can be broken down into the summation of many individual sine waves. Engine sounds are cyclical and can easily be reproduced using sine waves, but the problem is when you need to get to a very complex waveform you may need an incredibly large number of sine waves. This is even harder with the fact that these vehicles are non-linear and highly dynamic. It is a better approach then to use complex wavetables or sample playback to get the desired waveform. Blending the two approaches, additive and subtractive yields effective results. One is able to use subtractive approaches to create a broadband signal that either replicates a known vehicle sound or some unique but identifiable spectrum to aid in detection. An additive sine wave approach can then be added to the broadband signal, bringing in specific engine-orders associated with the various engines (6-cylinder, 8-cylinder, etc). The benefit of having separate signals as opposed to summing them all into one signal would be the complete control over the gain of individual components of the sound. One could attain a perfect brand sound for the vehicle with the broadband signal, and if there are certain legally required signals then the additive signals can address them.
Some examples of the possible scenarios for external vehicle sounds are Engine On, Idle, Drive Enabled, Acceleration and Deceleration, Park Enabled, Engine Off. There are also the two different parameters relevant to the drive, which are speed and pedal position. The National Highway Traffic Safety Administration (NHSTA) legal requirements do not include pitch shifting whatsoever due to the main function of the law being to aid in detection and the research revealed that volume changes are more important than pitch changes in identifying approaching vehicles. The NHSTA does add that implementing pitch shifting does still bring some helpful information for alerting pedestrians and they encourage system designers to incorporate this into their designs if possible. The real world research shows that volume changes and the frequency span of the sounds are the most important in preventing fatalities and injuries, hence only those are present in the legal requirement.
Simulated vehicle events.
Applying these techniques on a real vehicle takes a good amount of effort and consideration. Designers need to study and take into account the various frequency responses of the vehicle such as the electrical hardware, the speakers, the physical propagation effects of the vehicle structure, and its environment. All of these points will factor into the resulting sound characteristics inherent within a particular vehicle. A common approach is to take transfer functions or impulse responses of the various points and monitor them with convolution processing. Taking impulse responses and using them in the Wwise convolution plugin is one way of achieving this. This is of course assuming that the system is linear time-invariant, which of course it’s not but it is certainly a beneficial technique to get the sound in the ballpark. Nothing will replace performing measurements and tuning on a physical vehicle system. I will add though, that there are a number of newly developed (yet very expensive) tools on the market that allow for an accurate simulation of vehicle pass-by measurements which are used to test a vehicle’s adherence to the legal requirements.
When a vehicle’s external sound design is approved by the OEM in both aesthetic and technical aspects, the next step is to have real pass-by measurements taken on a track. The NHSTA requirement only reaches up to 30 KPH because above that speed the road and wind noise will be louder than the vehicle itself which is enough to aid in pedestrian detection. Most OEMs will have their own test track or have access to one. The NHTSA also allows for public access to a MATLAB script for taking three consecutive pass-by measurements and quickly estimating whether there was a successful attempt. This helps OEMs ensure they have a successful design implementation before spending the time and money on getting an official test taken at the Transportation Research Center in Ohio.
It has been quite an exciting journey to become familiar with the ways in which sound design is evolving within automotive design. I am very interested to see just how these technologies and techniques adapt and mature as they are put into production. The vehicle landscape is changing rapidly with the advancement of electric and hybrid vehicles and so are the sounds we hear on streets and highways. It’s a bright future for transportation, pedestrian experience, and sound designers with skill sets that are now totally applicable to automotive engineering.