- Solutions / Audio & Wireless Sensing
Intelligence from sound and signal. No camera required.
On-device keyword spotting, wake-word detection, audio denoising, and acoustic anomaly detection — running on-chip below the power budget of always-on embedded products. Plus, wireless sensing for presence and motion analytics when using a camera isn’t an option. No audio leaves the device. No image sensors required.
Three things your devices can do today.
Enable always-on voice interfaces with keyword spotting.
Deploy lightweight audio models for wake-word detection, keyword spotting, and voice-triggered interactions directly on embedded devices. Designed for low-power operation, these models continuously monitor audio streams and activate downstream processing only when relevant keywords are detected.
- Low-latency keyword detection
- Always-on listening
- Optimized for embedded deployment
Improve audio quality with real-time denoising.
Reduce background noise and improve signal clarity using real-time audio denoising models optimized for edge AI hardware. Ideal for voice interfaces, communication systems, industrial audio monitoring, and pre-processing pipelines that depend on clean audio input.
- Real-time noise suppression
- Edge AI optimized
- Ready for downstream audio pipelines
Detect and classify important audio events.
Identify speech activity, alarms, impacts, machinery anomalies, environmental sounds, and other acoustic events directly on-device. Audio classification models enable real-time monitoring and automated responses while keeping raw audio data local.
- Speech and environmental event detection
- Domain-specific classification
- On-device inference
Wireless sensing — in development.
Where It Applies
For environments where camera-based sensing isn’t feasible, isn’t permitted, or isn’t appropriate — patient care areas, private offices, residential spaces, and industrial zones where camera installation is physically constrained.Presence detection
Occupied vs. unoccupied state for HVAC, lighting, and energy management — no camera required.
Motion classification
Differentiate active occupancy from passive presence using wireless signal analytics.
Zone occupancy analytics
Track occupancy across zones in spaces where camera infrastructure isn't an option.
Audio never leaves the device.
Wireless captures no images.
Audio is the most personal sensor modality after video. Voice data is treated as biometric under many regulatory regimes and on-device inference means it doesn’t have to leave the chip. Wireless sensing captures no images and no identifiable signal content.