By Xiang-Yang Lu
Arduino announces launch of the Arduino Pro series
At the 2020 Consumer Electronics Show (CES), Arduino announced the official launch of the Arduino Pro series that are more compatible to practical applications. The new series brought together different boards with more integrate applications, while there are also hardware, software, and solutions, such as LoRa gateways, micro-programmable controllers, and cloud services.
Among the various series of Arduino Pro board, the Nicla series come in the smallest size. It is apparently used on sensor nodes for the Internet of Things (IoT), and it required a small piece of covering shell and battery for assembly to be used in various occasions. The collected data is processed before it can be transmitted to other devices through Bluetooth or Wi-Fi.
For the Nicla series, Arduino has launched two boards in the past, Nicla Vision and Nicla Sense ME. The former features applications for processing image and visual sensing, while the latter for motion and environmental sensing. Both boards feature AI inference for data processing, basically the purpose of developing TinyML.
Nicla Voice by Arduino
At the most recently CES 2023, Arduino launched the third Nicla series board- Nicla Voice, which is designed for sound and voice sensing.
The first feature of Nicla Voice is the use of the NDP120 chip by Syntiant, a U.S. startup chip manufacturer. Nicla Voice adopts a special circuit design architecture called PIM to implement AI inference through the PIM architecture that is faster and more power-saving than the use of traditional approaches. The power consumption can be as low as hundreds of microwatts, making it suitable for various applications with little battery power.
NDP120 is targeted for applications in speech recognition, while still retaining computing. In other words, it can process data collected by sensors at the same time. In short, in addition to being a microphone designed for the voice sensing, Nicla also features a 6-axis inertial sensor and a geomagnetic sensor.
Since Nicla Voice supports voice sensing using AI, it also co-branded with Edge Impulse, a well-known online TinyML software tool provider. In this way, Nicla Voice can support more AI development tools, especially that Nicla Voice has online tuner and compiler that can optimize AI models and algorithms. Trained models, after the weight of the device is greatly reduced, are put into the chip for inference, which allows NDP120 to save even more power.
In addition to performing sound-related AI inference and recognition on the scene, Nicla Voice can also transmit data through Bluetooth or an I2C wire interface. Although Nicla Voice supports transmissions of data to various hardware devices that support Bluetooth, Arduino currently prefers the use of boards from the Portenta series and MKR series.
Conclusion
Many people may find that applications in voice recognition more boring than those in image recognition, but there are still many potential applications to be tapped into, in my humble opinion.
