Hey everybody, good afternoon. My name is Austin Mitchell and I am a computational designer for Harman International on their automotive industrial design team. And today I want to talk to you about some of our work specifically somewhere where you might not realize it actually uses computational design. It’s in our car speaker systems. So, today I’m going to go over a bit of our process, but I also want to talk to you about why I think that computational design provides a new paradigm for brand.

0:35 So, this is what our team tries to answer. What does premium sound look like? And not just how do we make something that’s that’s completely beautiful, but how does form actually impact acoustics and how we perceive sound. I personally came from an architecture background like many of you, I assume. So, my head goes to something like this. This is the Meerson Symphony Center in Dallas, Texas by architect IM Pay.

And, this was an important space for me because growing up, I played violin forever and I still do to this day. And music’s always been a huge part of my life. I’ve been trying to figure out how to blend design, music, acoustics, and architecture. So, this this space is where computational design clicked for me without really knowing it was computational design at the time. But the mass overhead that you see actually moves up and down to adjust the reverb time of the hall.

1:23 So I was at one concert there and during intermission it lifted to change the reverb of the space to feel like a cathedral and I I started to see this this impact between form and acoustics and also a clear connection between quantifiable data and artistic expression. So it’s our kind of vision to take this experience that we have in live performances and put this into your vehicle.

1:46 This is a program that I worked on called the Cadillac Celestic. It is Cadillac’s flagship EV that they came out with with with bespoke services and it uses an AKG sound system. So a quick note about Harmon itself. So we’re a connected technology company and a wholly owned subsidiary of Samsung, but we’re most known for iconic audio brands that you see on screen here. We have over 16 audio brands.

And if I were to give this presentation about a month ago, this list would be way shorter. We actually just finished an acquisition for Sound United. So we have even more brands as part of our portfolio that we design with. Each one has a distinct DNA as far as branding goes and the way we actually perceive sound. So how many of you own JBL Bluetooth speakers out of curiosity?

2:34 A lot of people. This is this has been my experience going everywhere. Even on the beach you see people carrying party boxes, Bluetooth speakers, and this is probably what we’re most known for. But we also have another range of products from our consumer lifestyle brand team. For automotive, we want to take the experiences that you have either with your JBL Bluetooth speaker or the brand CMF and translate that one to one into your vehicle.

So, I personally love the one on the far left that it’s one of the JBL Bluetooth speakers for Toyota Tacoma that actually docks into the, IP of your vehicle and becomes part of your surround sound system. But before I can discuss how we use computational design to support brand, I think we have to dissect brand DNA. Specifically for premium audio brands, we can distill this into signature sound, experiential technology, and design.

3:26 But I’ve personally found that computational design tends to thrive in these overlapping areas. This is where we’ve been able to provide new solutions and create new IP. So I want to break these down just a little bit, but signature sound is what we hear and how we perceive the differences between each audio system. So a big part of this is speaker architecture. What we see on screen here is a Bowers and Wilkins system with their tweeter on top technology.

3:54 Tweeter is a speaker that that produces very high frequency sound. And by doing this and transitioning it into a vehicle, we’re actually able to create a very centralized sound stage. So you actually perceive the sound as in coming like in front of you on a stage, right? I think so many of us are so used to headphones now that we actually perceive sound as coming in through the sides.

4:11 U but we want to create a concert experience. But if we start to dissect signature sound, it’s the frequency response that we hear. It’s a speaker architecture. Where are the speakers placed? We start to see the visuals for for the grills and actually the the speaker technology. Themselves. And then once we move on from from the doors and the the front of the cabin, we get into surround speakers.

So you find speakers in some headrests for the systems as well as overhead. So you actually start to create a wholly immersive environment. Then we have experiential technology. So what sound adjacent technologies extend our experience with sound? This is where our UIUX team and motion graphics teams totally thrive. It could be a sound responsive lighting system or an app that brings you into Abbey Road Studio. But this image is actually of our internal vehicle program called Auto1 where we put all of our latest connected technologies and we use parametric design methods for the light diffuser on the front.

5:17 And then we get to design itself. What are the brand materials? What’s the CMF driving this? So color, material, finish u and what are the physical signatures of a brand. This is an image of our Revel CMF brand definition and it uses a form by sound matra and with this it has a more scientific feel. So we want that to be conveyed to the customers. Right? If if we’re taking something like sound that you only hear, how do we convey that through physical parts?

5:48 So we take a step back and see the the entire picture. We create a holistic design experience with sound. So, how do we use computational design? For us, it’s going beyond just using it for one-off design themes or motifs. It’s complete infrastructure for a design design team of 12. And here’s why. We have 16 iconic audio brands, as I mentioned previously, and we work with over 50 OEM partners.

6:16 OEM stands for original equipment manu manufacturer. So that’s the Toyota or Hondas of the world. And if we have one brand that’s in one of those vehicles, we might have three different system levels. So that’ll determine the number of speakers where the speakers go, how they’re packaged. It could be anywhere from 10 speakers up to 36 like in the Cadillac Escalade with the AKG Studio reference system.

6:41 That’s 36 touch points that we have to design for and account for in the vehicle. Usually this is through speaker grill design. If we average and say 500 perforations exist per grill, that means simple math says that we have between 5,000 and 18,000 unique perforations per speaker system per vehicle. If we have two different brand levels on one vehicle, we can easily double that number. So it goes like I said beyond just a single script or single design and it becomes infrastructure for a team.

7:15 But computational design really acts as a hub in our organization. And I don’t mean that this is something where we take all these constraints into account and automatically spit out a design but it lets us navigate these constraints and inputs. On the right side this is just how we typically work. We we have industrial designers, we have mechanical engineers and acoustic engineers. We typically try to find a balance between all three.

7:38 So it becomes an optimization problem between these three groups. I found that computational design within our organization acts as a communication tool between these three. We have designers generating actual quantifiable data that we can pass to engineering teams and validate products. So encoding brand through parametric behavior. When you see our speaker grill designs I want you to think that it’s unmistakably that brand. It’s expressed through how geometry behaves under constraints.

8:08 And I think our Bowowers and Wilin system for Aston Martin is a great example of this. This is the Aston Martin DV12. And this is the first collaboration we did with Aston Martin. In this way, grills become a way for us to express brand and it becomes a pallet for our design team. You can see how drastically the surface vary the surfaces start to vary across the vehicle.

8:38 We have to account for different constraints, different surface curvatures and we want to maintain a similar design vocabulary across the cabin. So once we’ve developed this pattern u I especially love this picture because it starts to look like a evolutionary evolutionary transformation diagram. This is the second vehicle that we did. It was the Astonart DBX77. And this is exactly what I want to see. I want to see a pattern that starts to behave and express brand, express motion, and it’s consistent across a bunch of different designs.

9:17 But this all starts with a very human element. It starts with expressing story. With the AKG system. I’m not sure how many of you are familiar with AKG, but it was a studio born in the studio or sorry, it was a brand born in the studio. It’s famous for its microphone designs. It started in the 1940s and they created the world’s first large diaphragm condenser microphone, which I know that’s a lot of words in one sentence there, but their microphones were used by the Beatles.

9:47 They were used by Michael Jackson. They were used by Frank Sinatra. And we want to convey that through pattern. We want to convey the story through pattern. So once we’ve defined this at the element level, we can start to expand and put that into a brand guideline for how the behavior should act to express that brand. So I want to walk you a quick through a quick pipeline for the Cadillac Celesteic just to show how we use these tools.

So our entire ecosystem is built around Rhino 3D and Grasshopper right now with custom C# nodes and and Python scripts. One interesting variable that we have is that a lot of the surfaces that we receive are locked in place. Meaning we sometimes design a surface data that goes in the cabin, but a lot of times we receive these surfaces and they’re in varying levels of finished. You know it could be a bre that that is disjointed and and is bad data but we have to apply our designs regardless of what of what we receive and we know that it’s going to be about a 2year to threeyear process that we’re going back and forth between design teams and manufacturers to generate these.

10:57 Then we start to look at the actual system constraints themselves. So every speaker has different performance requirements and different orientations. We have to take this into account along with any experiential elements like lighting when we start creating patterns. We actually want to understand the critical minimums. When we start to look at radius that can be applied to metal sheets, right? This helps make scripts robuster and our design team is less likely to break them.

11:27 And I’m showing 2D kind of baseline parametric behaviors on this page, right? I think this is something we we can all probably do in our sleep is make 2D patterns. Where it gets interesting is we have to translate these 2D patterns to 3D and go back and forth because a lot of metal parts that we create are actually created in 2D ultimately. And then we can start to look at actual acoustic behaviors.

11:58 So this is where we go beyond styling and tune perforations for acoustic performance using C car C car scripts based on combination of performance requirements and organizational knowledge. We’re able to provide live acoustic feedback to our designers. So these are heruristics. They’re not full simulations in this in this scenario. We do this because we want the computational load to be extraordinarily light and to develop really fast feedback.

Also, we touched on this, the tension between designer and and outcomes. When I first started working on these scripts and and working on perforation patterns for speaker grills, I think it’s our inclination to optimize everything. I I wanted to have the best sound performance, the best mechanical performance, but that didn’t always equate to the most beautiful speaker grills. It didn’t equate to the most beautiful cabinet interiors.

12:46 So, we need to create loose tools that empower our design team to be creative. What was the term that was used previously? I think it was it was satisfice. It’s good enough, right? We want it to be good enough and and look beautiful. So then once we have the base pattern of the story, we can start to actually create sophistication. We can add layers. In this scenario for the Cadillac, we used the 2D plus process.

13:19 So it’s a chemical etching process. And we were able to, like I said, take that base pattern and start adding complexity to this. And then with the final outcomes, we’re generating 2D and 3D exports. We’re able to check patterns for distortion and see what mapping method would be best applied to that program. This in in this scenario is a pretty flat grill, which is great. When we start to add more curvature into account, we have to look at different meth methods on a case-by case basis.

13:50 And then it comes to prototyping and validation. So, what I mentioned earlier is as long as we have acoustic heruristics where we’re able to tell if a speaker grill might perform well, then we’re able to quickly 3D print that prototype and test it in our lab and that usually takes 1 to two days. Whereas, even if we did have all the data for a simulation, which in a lot of scenarios we don’t, I think this would be faster and take less computational time.

14:18 And hopefully over time we start to see our scripts get smarter and smarter based on the feedback that we gain. And I would love to go back to that evolutionary diagram that I had the the showing all the speaker grills for the Aston Martin DB77 and see how that evolves over time too based on that feedback. So with a computational design pipeline we can let RAN speak across materials and not fight them.

14:41 So we author our scripts in a way that we can transition between materials and adjust for manufacturing constraints. Like a plastic grill for the Toyota Tacoma in the top left is significantly more complicated because we have to take into account draft angles and actually produce 3D geometry that’s manufacturable in the end. So we’re able to quickly adjust parameters and produce the data that manufacturers need. It’s not about, you know, having one down perfectly and and doing that over and over again.

15:11 It’s the fact that as a project changes over time, we can adapt to those constraints. To take it a step further or past speaker grill design, we also look at actual brand texturing. For Revel, as I mentioned previously, it has a form by sound mantra. I can’t imagine another way of doing this without computational tools. And then we also start to look at acoustic surfacing. This is the frontier we’re working on and and it’s this balance between acoustic performance and mechanical engineering.

15:50 We have always changing environments. This is Bang and Olson’s ALT lens technology. So in this case, we’re actually morphing sound in a specific way and that envelope has to adapt and we have to make sure that we’re maintaining that performance of the lens itself. So this is where I rely on the experts at Harmon. We’re super fortunate to work with some of the world’s best acoustic engineering engineers and I find that computational design is this linkage between our worlds right with design and performance.

16:29 So I’ll leave you with a few notes. I truly believe computational design is the connective fabric that drives innovation in organizations. There’s a lot that we’re working on and and a lot of exciting things that I can’t talk about right now, but I it all relies on computational design as creating new solutions and IP heristics empower creativity during the design process. We want our tools to actually empower design and play, not stifle it.

16:59 Huristics give us computationally light ways of providing feedback during the design process. And then finally, our work is no longer a downstream service. As computational designers, we’re now the authors of brand and physical space. We’re the ones creating geometry, performance metrics, manufacturing. We’re combining this all into one single source of truth. And as you change that system, you ultimately change the product and you change the brand. To see the full recording of this and previous presentations, as well as information about future CDF events, visit CDFAM.com.