semiconductors

Check Instagram Post

  • sales@struentsemi.com
  • +91 91760 15566

Products

Product Description

Struent Edge-AI Audio Coprocessor

The Struent Edge-AI Audio Coprocessor is an ultra-low-power, always-on next-generation smart audio intelligence chip designed to bring true sound awareness to any battery-powered device. Acting as a companion to Bluetooth or Wi-Fi SoCs, it adds advanced voice, safety, and wellness features without increasing host power consumption or system complexity.

Fully digital with PDM/I²S interfaces, it eliminates the need for analog front ends, making integration simple and cost-effective.

Its built-in neural processing unit (NPU) enables multi-model inference for wakewords, short phrases, safety alerts, and wellness cues — all within sub-milliwatt power envelopes.

For OEMs, it’s a plug-and-play silicon solution that transforms everyday products — from hearables and toys to doorbells, smart glasses, and home hubs — into intelligent, context-aware devices. For investors, it represents a massive opportunity: a scalable, drop-in technology powering the next generation of audio-aware consumer electronics.

Block Diagram

Block Diagram

Key Features

⚙️
Compute Architecture
Compact NPU with 128 INT8 MACs, multi-head inference for concurrent wake-word, safety, and wellness models.
Power Efficiency
80–120 µW (listen), 200–400 µW (inference burst); optimized for continuous operation.
👂
Always-on Intelligence
Simultaneous detection of wake-words, short phrases (3–5 words), and environmental/safety events.
❤️
Wellness Analytics
Detects cough, snore, breath-rate proxies, and stress tones — enabling basic wellness tracking on edge devices.
🛡️
Safety Monitoring
Identifies glass breaks, sirens, horns, and loud impacts for smart home, toy, and earbud safety features.
🔊
Digital Audio Front End (DAFE)
Integrated PDM/I²S input, decimation, noise suppression-lite, AGC, and framing — no Analog circuitry required.
🔌
Interfaces
I²C/SPI for control + IRQ signaling; optional I²S TX for denoised PCM output.
💾
Memory System
System SRAM, NPU scratchpad, I/D TCM.
📦
Integration Options
Internal NVM SKU for compact BOM; optional external QSPI NOR for larger model sets.
SDK & Ecosystem
Open SDK supporting ONNX/TFLite import; pre-trained model packs for wake-word, safety, and wellness detection.
🔐
Security
Secure boot, encrypted event transmission to host, privacy-first edge AI design.

Applications

Hearables

HEARABLES

HEARABLES

Always-on voice and safety detection at ultra-low power.

Hearables

HEARABLES

HEARABLES

Always-on voice and safety detection at ultra-low power.

Smart Toys

SMART TOYS

SMART TOYS

Enables voice interaction, safety sound detection, and health-aware cues for smarter, engaging toys.

Smart Toys

SMART TOYS

SMART TOYS

Enables voice interaction, safety sound detection, and health-aware cues for smarter, engaging toys.

Smart Glasses

SMART GLASSES & WEARABLES

SMART GLASSES & WEARABLES

Low-power voice and wellness detection for continuous monitoring.

Smart Glasses

SMART GLASSES & WEARABLES

SMART GLASSES & WEARABLES

Low-power voice and wellness detection for continuous monitoring.

Smart Hubs

SMART HUBS / ASSISTANTS

SMART HUBS / ASSISTANTS

Enables wakeword and safety listening while the main SoC sleeps.

Smart Hubs

SMART HUBS / ASSISTANTS

SMART HUBS / ASSISTANTS

Enables wakeword and safety listening while the main SoC sleeps.

Appliances

APPLIANCES / IOT SENSORS

APPLIANCES / IOT SENSORS

Brings edge voice and safety awareness to smart appliances and sensors.

Appliances

APPLIANCES / IOT SENSORS

APPLIANCES / IOT SENSORS

Brings edge voice and safety awareness to smart appliances and sensors.

Security

DOORBELLS & SECURITY CAMERAS

DOORBELLS & SECURITY CAMERAS

Local detection of baby cries and alarms—real-time alerts, zero cloud risk.

Security

DOORBELLS & SECURITY CAMERAS

DOORBELLS & SECURITY CAMERAS

Local detection of baby cries and alarms—real-time alerts, zero cloud risk.