Picovoice Promises 10-Minute Wake Word Detection on an Arduino Nano 33 BLE in New Tutorial
Based on the company's Porcupine engine, this short guide gets you started quickly — and at zero cost bar the dev board.
Picovoice, a specialist in lightweight machine learning for voice recognition, has written a guide to getting up and running with wake word detection on the Arduino Nano 33 BLE — at no cost bar the development board itself.
"Learn how to train and deploy voice recognition models into Arduino Nano 33 BLE with Picovoice Porcupine Wake Word Engine," the company writes of its recently-published guide. "This tutorial will take less than 10 minutes from the start to a working demo!"
It's a bold claim, but the short document is undeniably easy to follow. The tutorial begins with Picovoice Console, for which free account registration is required, to gain access to the company's Porcupine wake word detection system. Rather than relying on a pre-set wake word, the console allows the user to train recognition of any arbitrary word or phrase — "hey Nano" in the company's example, but trivially customizable.
The tutorial then switches to deployment of the model on the Arduino Nano 33 BLE board, using an example sketch template which can be customized with the trained model — distributed as a hexadecimal constant which can be quickly pasted into place. When flashed and run, the sketch listens for the wake word using the Arduino Nano 33 BLE's on-board microphone — and writes to the serial port when it's recognized.
"You can train and deploy several models with Porcupine and recognize them concurrently," Picovoice notes in the tutorial's conclusion. "The increase in CPU and RAM usage is negligible. If you want to understand complex voice commands with multiple slots and entities […] you can also use Rhino Speech-to-Intent to accomplish this."
The full, short tutorial is now available on the Picovoice website — along with links to register for a free account.