Arduino Assists with On-Device Voice Recognition
Arduino introduced a new development board called “Nicla Voice” that makes it possible to bring always-on voice and speech processing to IoT devices. The Nicla Voice is based on Syntiant’s ultra-low-power NDP120 AI accelerator.
It also contains a Nordic nRF52832 SoC with a single Arm Cortex-M4 CPU clocked at up to 64 MHz, 64 kB of SRAM, 512 kB of on-chip flash, plus 16 MB of SPI flash for additional storage and a Bluetooth LE radio for wireless connectivity. Implementing Syntiant’s Core 2 architecture, the NDP120 supports up to seven audio streams as well as I2S and TDM audio output support, and it’s supplemented with a Cortex-M0 CPU and 48 kB of on-chip SRAM.
Measuring approximately 22.8 × 22.8 mm, this is the latest dev kit in Arduino’s small-form-factor Nicla family, which launched with the Nicla Sense in 2021 and was followed by the Nicla Vision in 2022. Nicla Voice is equipped with a wide range of sensors, including a MEMS microphone, smart six-axis IMU motion sensor, and three-axis magnetometer. The sensor suite gives it the ability to not only listen for sounds or speech, but also to sense gestures, vibrations, and other movements.
With its headers and castellated pins, the Nicla Voice fit onto any custom carrier board when you’re ready to move to mass production. The I/O features includes 1x I2C bus (with ESLOV connector), 1x serial port, 1x SPI, 2x ADC, and a programmable I/O voltage from 1.8 to 3.3 V.