Specifications
Board Family
Arduino Nicla Voice (ABX00061), Arduino Nicla Vision and Arduino Nicla Sense ME (ABX00050)
Form Factor
22.86 x 22.86 mm miniature footprint with ESLOV connector and castellated edge pads for direct soldering
Operating Voltage
1.8 V board logic; single-cell Li-ion / Li-Po battery operation with on-board charging
Nicla Voice AI Engine
Syntiant NDP120 Neural Decision Processor: Syntiant Core 2 deep neural network inference engine, HiFi 3 audio DSP and an Arm Cortex-M0 at up to 48 MHz
NDP120 Memory and Power
48 KB dedicated SRAM; 1 µA typical / 3 µA maximum quiescent current with a dedicated low-dropout regulator
Nicla Voice MCU
ANNA-B112 module based on Nordic nRF52832 — 64 MHz Arm Cortex-M4F with 64 KB SRAM and 512 KB flash
Nicla Voice Microphone
IM69D130 MEMS microphone, 20 Hz to 20 kHz with 105 dB dynamic range and under 1% total harmonic distortion
Nicla Voice Motion Sensors
BMI270 6-axis IMU (±2/4/8/16 g accelerometer, ±125 to 2000 dps gyroscope) plus BMM150 3-axis magnetometer (±1300 µT x/y, ±2500 µT z)
Nicla Voice Wireless
Bluetooth 2.400-2.4835 GHz internal antenna — Bluetooth 5.0 via the Cordio stack, 4.2 via ArduinoBLE
Nicla Voice Interfaces
2x SPI and 2x I2C (one of each on the pin header) plus a 12-bit, 200 ksps ADC
Nicla Vision MCU
STM32H747AII6 dual-core — Arm Cortex-M7 up to 480 MHz plus Cortex-M4 up to 240 MHz
Nicla Vision Imaging
2 MP colour camera for on-board image analysis and TinyML vision models
Nicla Vision Sensors
6-axis smart motion sensor, integrated microphone and distance sensor
Nicla Sense ME AI Sensor
Bosch Sensortec BHI260AP self-learning AI smart sensor with 2 MB QSPI flash dedicated to the sensor
Nicla Sense ME Environment
BME688 environmental sensor for gas, pressure, humidity and temperature, plus a 3-axis magnetometer
Nicla Sense ME Memory
512 KB flash and 64 KB RAM with 2 MB SPI flash for local storage
AI Toolchain
Arduino IDE and Arduino Pro tooling with TensorFlow Lite Micro and Edge Impulse support for the TinyML workflow
Form Factor Enablers
On-board battery charger, USB debug interface and ESLOV connector for adding sensors and actuators
Target Workloads
Always-on keyword and speech recognition, vibration and acoustic predictive maintenance, gesture and contactless UI, asset tracking and object recognition
Overview
The Arduino Nicla family puts machine learning onto a 22.86 x 22.86 mm board. Three variants cover the three most common edge AI sensing problems. Nicla Voice pairs a Syntiant NDP120 Neural Decision Processor with an nRF52832 host MCU, so always-on speech recognition and audio event detection run at microwatt quiescent current while the host stays asleep — the NDP120 idles at roughly 1 µA typical and 3 µA maximum.
Nicla Vision takes a different route to the same goal: an STM32H747AII6 with a Cortex-M7 at 480 MHz and a Cortex-M4 at 240 MHz drives a 2 MP colour camera, a 6-axis motion sensor, a microphone and a distance sensor, running image classification and object recognition directly on the board. Nicla Sense ME focuses on machine health and environment, built around Bosch Sensortec's BHI260AP self-learning AI smart sensor with 2 MB of dedicated QSPI flash, alongside a BME688 gas, pressure, humidity and temperature sensor.
All three boards share the same 22.86 mm footprint, ESLOV connector, on-board battery charging and USB debug interface, and all three work with the standard Arduino toolchain plus TensorFlow Lite Micro and Edge Impulse. That shared platform means a design can prototype with one Nicla variant and switch sensing modality without re-tooling the enclosure or the software stack.
Key Benefits
Microwatt always-on inference: the NDP120's dedicated neural engine runs speech and audio models continuously at single-digit microamps, so battery-powered products do not need a wake-up duty cycle. Three sensing modalities, one footprint: audio, vision and environmental/AI-sensor variants share mechanicals, connectors and toolchain, cutting re-spin cost between product concepts. Ready TinyML pipeline: Arduino IDE plus TensorFlow Lite Micro and Edge Impulse support move a model from laptop training to on-board inference without custom embedded tooling.
Applications
Always-on voice interfaces and keyword spotting, contactless and gesture control panels, predictive maintenance from vibration and acoustic signatures, industrial asset tracking, machine vision for identification and sorting, and battery-powered smart sensor nodes.
Request a Quote — ARDUINO NICLA SERIES EDGE AI BOARDS — NICLA VOICE WITH SYNTIANT NDP120, NICLA VISION AND NICLA SENSE ME
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →