Acoustic-Driven Computational Design: Premium Branded Audio In The Automotive Industry
Interview with Austin Mitchell – Harman International
At HARMAN International’s automotive division, computational design is central to translating brand identity into manufacturable, acoustically optimized components for the world’s leading automotive audio systems.
In his presentation at CDFAM NYC, Austin Mitchell, Senior Computational Designer at HARMAN International will share how his team uses computational design workflows to generate brand-specific speaker grille geometries across 50+ OEMs, balancing aesthetics, performance, and production at scale.
The following interview explores how computation acts as connective tissue across disciplines, enabling real-time feedback, rapid iteration, and design that truly resonates.
Could you start by introducing Harman International and what you will be presenting at CDFAM?
Harman International is a global leader in connected technologies and premium audio experiences, shaping the future of mobility and consumer electronics.
While many may first think of Harman Kardon, that’s just one of the iconic brands in our portfolio. With brands like JBL, Harman Kardon, and AKG, we combine cutting-edge engineering with human-centered design to deliver immersive sound, intuitive interfaces, and seamless connectivity across automotive, consumer, and professional markets. As a subsidiary of Samsung Electronics, Harman empowers designers to craft experiences that resonate—both literally and emotionally.
At CDFAM, I’ll be sharing how computational design is central to our automotive industrial design teams’ daily workflows—how it defines and strengthens brand identity, and how it acts as a connective tissue across disciplines within our organization.
How does computational design integrate with acoustic engineering in your workflow to produce brand-specific yet manufacturable audio components for vehicles?
In our workflow, computational design acts as a bridge between industrial design intent, acoustic performance requirements, and manufacturability. My work on the design team falls into two main categories: production work and acoustic-driven computational design development.
For active vehicle production, we generate brand-specific parametric models of speaker grilles and audio components that respond to speaker type, location, and orientation.
We use computationally lightweight heuristics—based on historical acoustic data and internal design guidelines—to provide live feedback on a component’s predicted acoustic performance during the design phase.
While this may sound straightforward, our team designs components for over 12 audio brands in Harman’s portfolio, serving more than 50 OEM clients globally. Premium automotive systems can include up to 36 speakers, each representing a unique industrial design touchpoint and requiring a custom grille.
If each grille contains around 500 perforations, that’s roughly 18,000 designed perforations per vehicle—an ideal challenge for computational design.
While production work prioritizes industrial design with acoustic-feedback, my efforts on acoustic-driven computational design focus on optimizing components to achieve specific acoustic outcomes and exploring novel digital fabrication processes. More to come on this in the future…
What software platforms and computational methods are most commonly used in your acoustic-driven design process?
Our automotive industrial design team primarily works within a Rhino/Grasshopper pipeline, using Python and C# for specific use cases and interoperability.
Each of our twelve industrial designers is proficient in parametric Grasshopper workflows, and we’ve built a robust library of both brand-specific and brand-agnostic scripts.
While our acoustic engineering counterparts rely on COMSOL for full acoustic simulations, our team has developed acoustic heuristics based on organizational knowledge.
These allow us to provide live feedback during the design process, helping assess whether a given solution aligns with a brand’s acoustic requirements. In many cases, real-world prototyping is faster and more practical for us to execute. Computational design methods enable rapid iteration—we can prototype and test designs within a day to gather real-world acoustic feedback.
Can you describe the data flow from acoustic performance requirements through computational modeling to final manufacturable geometry?
It’s impossible to talk about starting from acoustic performance requirements without first addressing audio brand DNA.
As consumers, we all perceive sound differently—this is the realm of psychoacoustics – and have different sonic preferences. Each brand’s signature sound is shaped by unique speaker architectures, including layout, speaker quantity, and technology.
For high-performance systems like the AKG Studio Reference found in Cadillacs, our acoustic engineering team sets more stringent design requirements, such as grille acoustic transparency.
As a tier-1 automotive supplier working with roughly 50+ OEM’s (original equipment manufacturers, like Toyota or Cadillac), consistent data inputs are akin to the wild west. That’s why we rely heavily on computational design to stay agile and responsive to constant change.
In speaker grille design, we ideally begin with finalized speaker locations and defined manufacturing requirements. We receive BREP surfacing data from OEMs, apply our branded pattern scripts, and adjust for mechanical, acoustic, and manufacturing constraints before delivering CAD data aligned with the manufacturing process. The geometric behavior of these patterns must be robust enough to adapt to shifting surface data, speaker placements, material types, and manufacturer preferences. For example, an injection-molded grille with draft angles is a very different deliverable than a laser-cut stainless steel sheet—and our scripts are built to handle both.
How do you balance the need for brand differentiation with scalability across multiple automotive platforms?
Every brand and automotive partnership requires thoughtful attention. Our goal is to faithfully represent the lifestyle and identity of each consumer audio brand, translate its design language into the vehicle environment, and simultaneously align with the OEM’s design direction.
While brand expression varies from vehicle to vehicle and brand to brand, we build our geometric logic in Grasshopper to respond to consistent constraints—such as material thickness, minimum radii, and manufacturing tolerances. This approach allows any member of our team to deploy these parametric scripts to deliver designs that are both visually aligned with brand identity and functionally optimized for acoustic performance.
What do you hope to share with, and learn from, the CDFAM community regarding computational design in the context of acoustics and industrial design?
I hope to highlight the critical role computational design plays in empowering and defining brand identity—beyond just aesthetics.
Computational approaches are already making a significant impact in consumer-focused industries like footwear, and I’d love to see these methodologies expand into other niche fields to drive meaningful, high-impact outcomes.
I also want to share how my experience at Harman demonstrates that computational design can act as a connective tissue between disciplines, fostering collaboration and innovation.
My background is in design—not acoustics—so I rely on the collective intelligence of the experts around me to synthesize new methods of creation. Through this collaborative approach, we’re pioneering acoustic-driven processes and generating designs that truly perform.
I’m excited to engage in meaningful conversations with the CDFAM community, learn how others are applying computational design in industrial contexts, and explore the novel design methods being developed across the field.
Seats are filling fast, and in true Harman fashion, we’re nearly at full volume.
Don’t miss your chance to hear how computation is redefining acoustic design at scale. Join Austin and the rest of the CDFAM community in NYC for two days of boundary-pushing talks, real-world applications, and cross-disciplinary insight.
Register now before the last few spots are gone.
