Our Smart MEMS Microphone Series integrates a high-performance MEMS sensor with an ultra-low-power AI chip. This “Sense-Store-Compute” architecture enables edge-based Voice Activity Detection (VAD) and Keyword Spotting (KWS) at unprecedented power levels.

Core Advantages

• Integrated Neural Processor (NPU): Supports ultra-low-power voice wake-up and real-time VAD.

• Dual-Core AI Engine: Includes a low-power BNN (LP_NPU) for always-on anomaly detection and a deep CNN (NPU) for complex keyword recognition, supporting up to 30 keywords.

• Smart Power Management: Features an “Always-on” mode that only triggers the main processor when human speech is detected, significantly extending battery life.

• Studio-Quality Audio: High Signal-to-Noise Ratio (SNR) of 65dBA and an Acoustic Overload Point (AOP) of 127dBSPL for clear pickup in any environment.

Intelligent Working Modes

• VAD Mode (~70μA): Ultra-low-power standby for voice detection.

• KWS Mode (~160μA): Continuous inference for real-time keyword matching.

• Hybrid Wake-up: VAD-triggered KWS for the ultimate balance between performance and power consumption.

Technical Specifications

• Package Size: 3.5 × 2.65 × 1.0 mm

• Interface: Standard I2C Slave (up to 400Kb/s)

• Sensitivity: ±1dB high precision

• Supply Voltage: 1.6V – 3.6V

Typical Applications

Ideal for TWS earbuds, smartwatches, AR/VR glasses, smartphones, and intelligent IoT home devices.

Technical Documentation

For detailed electrical characteristics, mechanical drawings, and application notes, please download the full series datasheet:

• 👉 Download Full Series Datasheet (PDF): en   cn