October 1, 2026 – As the physical AI landscape matures, the bottleneck for high-fidelity spatial awareness has long been the "front end"—the ability to capture high-resolution audio data with precision, reliability, and ease of integration. Today, Norwegian deep-tech pioneer sensiBel addressed this challenge head-on with the launch of the NOVA32™, a sophisticated acoustic-detection hardware platform designed to accelerate the development of complex beamforming applications.
By integrating 32 of its proprietary optical MEMS microphones into a single, plug-and-play USB interface, sensiBel is effectively removing the hardware integration barriers that have historically slowed down developers in the drone detection, industrial maintenance, and acoustic imaging sectors.
Main Facts: A New Paradigm for Acoustic Data Acquisition
At its core, the NOVA32 is more than just a microphone array; it is a turnkey solution for developers working at the intersection of acoustics and artificial intelligence. The hardware platform aggregates 32 SBM100B optical MEMS microphones arranged in a precision spiral array, designed to provide high-resolution spatial capture.
The key innovation lies in the platform’s ability to bypass the traditional, cumbersome integration process. Previously, engineers building large-scale acoustic cameras or sensor arrays had to navigate the complexities of designing custom PCBs, integrating field-programmable gate arrays (FPGAs), and managing multiple analog-to-digital converters (ADCs). The NOVA32 eliminates these hurdles by utilizing the XMOS XCORE.AI processor to synchronize all 32 channels and output them as a single, 24-bit USB audio stream.
Key Technical Specifications:
- Microphone Array: 32-channel SBM100B optical MEMS.
- Connectivity: USB plug-and-play, compatible with Windows, MacOS, and Linux.
- AI Integration: Native support for NVIDIA Jetson™ and other edge AI development platforms.
- Dynamic Range: 132dB.
- Self-Noise: 14dBA equivalent input noise (EIN).
- Signal-to-Noise Ratio (SNR): 80dBA.
- Acoustic Overload Point (AOP): 146dB.
Chronology of Development: From Optical MEMS to Market-Ready Hardware
The launch of NOVA32 is the culmination of years of R&D at sensiBel’s Norwegian headquarters. While the company has long been recognized as a leader in optical MEMS technology, the transition from component supplier to platform provider marks a significant strategic pivot.
- Foundation: sensiBel’s early years were defined by the perfection of its optical MEMS sensors, which utilize light rather than traditional capacitive structures to detect diaphragm movement. This physical breakthrough enabled the industry-leading noise floor and dynamic range that characterize the SBM100B.
- Prototyping Phase (2024-2025): Recognizing that its high-performance microphones were often "bottlenecked" by subpar secondary electronics, sensiBel began working on a reference design that would allow customers to utilize the full potential of the sensors.
- Platform Integration (Early 2026): The development team focused on the "plug-and-play" aspect, realizing that to scale adoption in the AI community, the hardware needed to speak the language of common development kits like NVIDIA Jetson.
- The Launch (October 2026): With the release of NOVA32, sensiBel transitions from a component manufacturer to an essential enabler of the "Physical AI" ecosystem, providing the hardware "ears" for the next generation of autonomous and industrial machines.
Supporting Data: Why Physics Matters in Acoustic AI
In the world of AI, the adage "garbage in, garbage out" is particularly relevant. When an AI model is tasked with identifying a drone’s sound signature in a noisy urban environment or detecting the microscopic bearing wear of a factory turbine, the quality of the input data is the primary limiting factor.
Traditional capacitive MEMS microphones, while ubiquitous, are limited by the physical constraints of their structure. In contrast, sensiBel’s optical MEMS technology—which uses light to measure diaphragm displacement—is fundamentally more sensitive and stable.
According to internal benchmarks provided by the company, the SBM100B sensors within the NOVA32 provide two-and-a-half times the microphone-array performance over the current industry-standard capacitive MEMS sensors. With an EIN of just 14dBA and an AOP of 146dB, these microphones can capture both the most delicate whispers and the most thunderous industrial sounds without distortion or noise floor interference. This high-fidelity capture ensures that AI algorithms have a "cleaner" data set to work with, leading to faster training times and more accurate real-world inference.
Official Responses: The Vision for Edge AI
The leadership at sensiBel is clear about the platform’s role in the broader technological landscape.
"NOVA32 is the acoustic-detection platform for which the industry has been waiting," said Kieran Harney, CEO of sensiBel. "Providing a single stream of 32 synchronized MEMS microphones from sensiBel, NOVA32 lets product developers get up and running in record time—so they can focus on what they do best: creating edge AI firmware for their applications on commercial platforms such as NVIDIA Jetson development kits."
Tom Hazlett, Vice President of Sales and Marketing at sensiBel, emphasized the competitive advantage provided by the underlying physics. "The physics of the optical MEMS microphones in NOVA32 tells the story for us," Hazlett stated. "By moving from traditional capacitive structures to our optical approach, we are providing developers with the precision they need to solve problems that were previously thought impossible to handle at the edge."
Implications: The Future of Acoustic Intelligence
The implications of the NOVA32 for various sectors are significant. By lowering the entry barrier for high-fidelity acoustic arrays, sensiBel is poised to accelerate innovation in three key areas:
1. Advanced Drone Detection
As the proliferation of drones continues, security and defense sectors are demanding better detection systems. Acoustic arrays are vital for tracking drones that might be visually obscured or flying in "stealth" modes. The NOVA32’s high SNR and sensitivity allow for more precise beamforming, enabling the system to pinpoint the location of a drone in a 3D environment with higher accuracy than ever before.
2. Industrial Predictive Maintenance
In smart factories, acoustic monitoring is the "canary in the coal mine." Bearing wear, gas leaks, and electrical arcing all have distinct sound profiles. The NOVA32, when integrated with industrial AI platforms, allows for the deployment of massive sensor networks that can "listen" to a facility 24/7, catching failures before they become catastrophic.
3. Environmental Noise Mapping and Acoustic Cameras
Urban planning and environmental protection agencies are increasingly using acoustic cameras to map noise pollution. With the NOVA32’s 32-channel synchronization, researchers can build portable, high-resolution acoustic cameras that provide a visual overlay of sound intensity, helping cities design better sound barriers and urban layouts.
Conclusion: A New Standard for the Edge
The launch of the NOVA32 is a milestone for sensiBel, but it also represents a larger shift in how we approach edge computing. As we move toward a world filled with "physically intelligent" devices, the hardware that bridges the gap between the physical world and the digital processor must become more capable and easier to implement.
By delivering a complete, high-performance acoustic acquisition package, sensiBel has effectively "democratized" the high-end microphone array. For developers, this means the focus can shift away from the difficult, time-consuming tasks of signal synchronization and hardware design, and toward the true goal: building more intelligent, more aware, and more capable AI systems.
As we look toward the remainder of 2026 and beyond, the NOVA32 is set to become an essential tool in the developer’s toolkit, proving that when it comes to artificial intelligence, the quality of your ears is just as important as the strength of your brain.
