Edge AI wake word and command in partnership with Sensory

Sensory's TrulyHandsfree-Micro on-device processing eliminates cloud dependency, saving energy, enhancing privacy & reducing server costs.

Edge AI wake word and command in partnership with Sensory

Application overview

Sensory's TrulyHandsfree-Micro paired with Sensory's VoiceHub and TI provided examples enables quick and easy integration of voice and wake-word commands on any general-purpose Cortex-M4F and Cortex-M33 processor. This application uses only CPU execution with no hardware acceleration required, achieving efficient performance with 40% and 27% CPU utilization for wake-word and command detection respectively at 96MHz.

This allows users to naturally interact with your product through hands-free voice control. Sensory's TrulyHandsfree-Micro executes entirely on the local device without transferring any voice data to the cloud for processing. When combined with wireless MCUs such as CC27xx/CC35xx, you can enable low-bandwidth and power-efficient wireless IoT connectivity through Bluetooth LE, Matter, Zigbee, or Wi-Fi for applications like hands-free appliances or electronic door locks all without requiring cloud connectivity.

Starting evaluation

Data collection

Use Sensory's VoiceHub to create custom audio wake-words and commands for your application.

Learn more

Build and train your model

With Sensory's TrulyHandsfree-Micro, training is not required. Sensory VoiceHub generates your wake and command word model.

Choosing the right device for you

Our wireless MCU family of CC27xx and CC35xx are powerful Cortex-M33 MCUs with Bluetooth LE, WiFi, Zigbee, Matter and Proprietary wireless connectivity

All the hardware, software and resources you’ll need to get started

Hardware

LP-EM-CC2745R10-Q1
CC2745R10-Q1 LaunchPad™ development kit for SimpleLink™ 2.4GHz wireless MCU.

Software & development tools

TI's GitHub demo
Learn  more about how to execute this AI task with TI and Sensory. 

Resource documents

Additional use cases

See all use cases