Sensory
Sensory was founded in 1994, years before voice assistants or cloud AI existed, with a clear goal: Allow people to interact with products the way we interact with each other. Early on, the team focused on neural-network-based speech recognition at a time when most systems relied on brittle rule-based approaches.
Hundreds of leading brands have licensed Sensory’s solutions, spanning voice, speech, sound, and biometrics, all designed to work locally on devices with limited resources. Sensory maintains a foundational portfolio of approximately 70 issued patents covering AI fundamentals like wake words, speaker verification, face and voice biometrics, and hybrid approaches to voice I/O. When you deploy Sensory technology, you gain both market‑leading on‑device AI and technologies protected by our proven patent portfolio.
Sensory’s innovations have become the backbone of always-listening, always-ready intelligence that respects user privacy and removes cloud computing fees.
Partner Solutions
AI-Enabled Voice Control
Sensory used its VoiceHub tool to quickly build a compact wake word and large vocabulary speech model tailored for appliance control, running entirely on the Alif Ensemble E3 SoC powered by Arm Cortex-M55 and Ethos-U55. The TrulyNatural speech recognition and NLU engine was optimized for efficient on-device performance, ensuring complete data privacy without cloud reliance. Workloads were carefully partitioned, with speech recognition accelerated on the Ethos-U55 and text processing handled by the Cortex-M55, while additional optimizations such as Helium acceleration, memory efficiency improvements, and INT8 quantization reduced power consumption and model size. Together, these enhancements enable accurate, customizable voice interaction within the tight performance and memory constraints of consumer devices.