Why Always-On Voice Has Been So Hard for Wearables
Always-on voice detection promises hands-free convenience, but in wearables it has long been constrained by power budgets. Fitness trackers, smart rings, and earbuds are tiny, with equally tiny batteries. Keeping microphones and processors continuously listening for a wake word can quickly drain those batteries, forcing users to recharge more often and limiting real-world adoption of wearable voice control. Many device makers have compromised by offering partial voice functionality or offloading heavy processing to the cloud, which introduces latency and privacy concerns. To make always-on voice practical, wearables need an ultra-low-power chip that can run AI models locally, at milliwatt levels, without inflating device size. This is the gap EMASS is targeting with its ECS-DoT AI system-on-chip, designed to deliver continuous, on-device intelligence while staying within the tight power and space envelopes of next-generation wearables.
Inside ECS-DoT: An Ultra-Low-Power AI System-on-Chip
EMASS’s ECS-DoT is positioned as a breakthrough AI system-on-chip that brings milliwatt-class processing to edge devices. Built on a RISC-V architecture and leveraging non-volatile memory technologies, the chip is optimized for highly compressed AI models and power-constrained environments. EMASS claims the ECS-DoT can deliver 10–100× lower energy consumption and up to 3× faster inference for common AI tasks compared with existing solutions. This combination is critical for always-on voice detection, where the chip must continuously analyze sensor data while preserving battery life. Beyond audio, ECS-DoT supports integrated sensor fusion, enabling real-time processing of audio, vision, and motion signals in compact devices. The flexible SDK, supporting multiple model types, is intended to shorten the path from proof-of-concept to production designs, making it easier for manufacturers to differentiate their wearables with persistent, low-latency AI capabilities.
From Smart Glasses to Smart Rings: New Voice-First Wearables
At Sensors Converge 2026, EMASS is showcasing how ECS-DoT can power always-on voice features in smart glasses, including a hearable application that uses bone-conduction voice detection. By embedding an Inertial Measurement Unit within the glasses frame, the system senses subtle jaw vibrations along the temple arm, allowing ECS-DoT to perform voice activity detection and keyword spotting without relying on always-on microphones. This approach can reduce power consumption and improve privacy, since audio does not have to be continuously captured and streamed. The same ultra-low-power design principles can extend to fitness trackers, earbuds, and smart rings, enabling wearable voice control that doesn’t demand daily charging. By processing AI workloads on-device at the milliwatt level, ECS-DoT makes it more realistic for manufacturers to design voice-first wearables that stay responsive throughout the day while still fitting into compact enclosures and existing battery capacities.
Industry Signal: Edge AI Ready for Next-Generation Voice Wearables
ECS-DoT’s appearance at Sensors Converge 2026, and its recognition as a finalist in the Best AI & Edge Computing Solution category, underscore a broader shift in the sensor and semiconductor landscape. The market is moving beyond simple data logging toward embodied intelligence, where devices act autonomously using on-device AI. EMASS is aligning itself with this trend by demonstrating live applications that show smarter, longer-lasting products without changing batteries or enclosures. Voice detection in smart glasses is one showcase, but EMASS also highlights gains in drones and industrial predictive maintenance, hinting at a diverse ecosystem of ultra-low-power edge AI applications. For wearable makers, the message is clear: the technology stack needed for reliable, always-on voice detection is maturing. As more AI system-on-chip platforms like ECS-DoT reach production, next-generation voice wearables could finally combine convenience, responsiveness, and battery life in a single design.