Page 52 - EE Times Europe Magazine - June 2025

52 EE|Times EUROPE

Voice Technology: Answering Europe’s Call for Safer Automotive Controls

algorithms, “acoustic zones” can capture commands, and driver patterns. These sys- a consistent and intuitive experience. A good
speech from any location while attenuating tems can make intelligent predictions about rule of thumb is that single-touch interaction
unwanted noise. A key consideration, how- likely commands, improving recognition rates is optimal for straightforward functions, but
ever, is the increased system cost associated even under challenging acoustic conditions. when tasks become more complex and require
with these multi-microphone arrays. This represents a significant leap beyond multiple swipes or taps, voice UI becomes a
An innovative AI-powered 3D acoustic simple voice control, delivering the natural more efficient method. Voice systems must
analysis solution, using a single overhead conversational experience you’ve always recognize when commands refer to on-screen
microphone array, promises to lower the wanted in your car. elements, while visual interfaces should indi-
vehicle’s component costs while maintain- cate voice command capabilities.
ing robust speech enhancement and voice Edge AI processing
recognition accuracy for up to six “acoustic Edge computing augments cloud-based TECHNICAL IMPLEMENTATION
zones”—a necessity for automotive appli- LLMs in voice UIs by handling local que- CHALLENGES
cations. A recent study by Head Acoustics ries, reducing latency and bandwidth costs, Several engineering challenges remain for
demonstrated that this advanced system and enhancing system reliability when the automotive voice systems:
sustains consistent speech recognition rate cellular network is unavailable. This shift is • Integration with vehicle acoustic
performance even at high speeds (120 km/h), particularly advantageous for safety-critical design. Manufacturers should consider
a condition under which conventional, single- automotive applications, where minimizing using advanced voice AI software systems
microphone array systems typically falter. command latency is crucial for preventing to avoid the cost/performance conundrum
potential hazards. associated with traditional approaches,
Neural noise suppression which require multiple microphones
Deep-learning–based noise suppression THE POTENTIAL OF VOICE UI IN VEHICLES throughout the cabin.
models can differentiate between speech and With these technical challenges addressed, • Multilingual performance. European
various types of vehicle noise with unprece- voice technology could transform the markets require robust performance
dented accuracy. These systems are trained in-vehicle experience into a truly intuitive across multiple languages and accents.
on extensive datasets of in-vehicle audio to interface that enhances both convenience • Processing efficiency. Embedded
identify and remove noise components while and safety. Drivers could issue natural systems must balance performance with
preserving speech integrity. commands—such as “I’m feeling cold” or power and space constraints.
Unlike traditional statistical models, “Find me the quickest route home, avoid- • Safety considerations. By implement-
neural network approaches can adapt to ing highways”—and engage in contextual ing advanced voice AI systems that can
the non-stationary noise profiles typical conversations. The system would recognize accurately identify and process who is
in automotive environments. This enables individual users, maintain personalized pro- speaking, driver safety is enhanced, as
higher speech recognition accuracy even as files, and proactively offer assistance based drivers and passengers no longer need
driving conditions change. Recent testing has on learned patterns and current conditions, to fumble with touchscreens, buttons, or
shown that next-generation, AI-driven speech from adjusting climate settings when rain is controls.
enhancement technologies can significantly detected to postponing notifications during While Euro NCAP is leading this regulatory
outperform traditional noise reduction solu- complex traffic situations. shift, similar safety considerations will likely
tions across multiple challenging scenarios. Advanced voice systems would dramatically influence standards globally. Manufacturers
reduce driver distraction by handling tasks developing for worldwide markets must con-
Speaker identification and separation that previously required multiple touch- sider how the European requirements might
Beyond creating virtual listening zones screen interactions. Complex vehicle settings forecast broader regulatory trends.
with 3D spatial processing, advanced could be adjusted with simple commands,
machine-learning models can identify indi- such as “Switch to sport mode.” The system CONCLUSION
vidual speakers within the cabin. This enables could also serve as a central interface for the Despite its historical limitations in automo-
a two-tiered personalization system, offering driver’s virtual assistant, allowing tasks such tive contexts, voice technology has reached
location- and identity-based infotainment as reading important emails, checking home a technological inflection point where it
settings for each passenger. Furthermore, it security systems, or placing food orders for can serve as a genuine safety enhancement.
enhances security for driver-specific functions pickup along a route. Through careful integration with physical
such as navigation updates and provides a controls and the thoughtful implementation
welcoming experience by loading person- THE PATH FORWARD: MULTIMODAL of advanced acoustic processing, voice AI can
alized infotainment profiles for passengers. INTERACTION DESIGN help create vehicle interfaces that are both
Today, this can be done with a single micro- The future of automotive UI likely lies in safer and more capable than those using tra-
phone array. Previously, such capabilities thoughtfully designed multimodal interfaces ditional buttons or touchscreens alone.
were available only if the automaker added that combine physical controls, voice inter- The coming years will likely witness
multiple microphones around the cabin, action, and visual displays. This approach significant innovation at the intersection
which increased design complexity and hard- acknowledges both human factors: research of regulatory requirements, safety consid-
ware costs. and evolving user expectations. erations, and technological capabilities,
Critical safety functions require immediate, potentially transforming how drivers interact
Context-aware processing reliable access through physical controls. Sec- with increasingly complex vehicle systems. ■
Thanks to today’s small language models, you ondary functions benefit from voice control’s
could interact with your car using your nat- hands-free advantage, while browsing com- Dani Cherkassky is the CEO of Kardome,
ural voice. No more struggling with awkward plex information may still use visual displays a voice AI company offering speech signal
commands or feeling restricted. The model when the vehicle is stationary. processing that works in any acoustic
is purpose-built for the specific functions of Integrating these modalities requires care- environment by clustering speech based on
your vehicle, tracking vehicle state, recent ful human-machine interface design to create location rather than direction.

JUNE 2025 | www.eetimes.eu