The WhisPro Software Development Kit (SDK) has been designed to allow the customisation to the customer's choice of wake words and voice commands, and runs entirely on-device, without the need for a network connection.
Voice User Interface (VUI) has rapidly become a primary interface for many IoT applications, providing an intuitive, hands-free interface for individuals to interact and control wearables and endpoint devices in the smart home, smart appliances, industrial/manufacturing, automotive, medical and more.
According to Mordor Intelligence, the voice recognition market is expected to grow at a CAGR of 16.8% from 2021 to 2026, reaching $27.2bn by 2026.
Through this collaboration, engineers using the CC3235x wireless MCU will be able to add speech recognition capabilities to IoT devices using the WhisPro SDK, such as wake words and speech commands.
WhisPro can support multiple wake words in parallel, including standard AI assistants (such as Alexa) and custom wake words, in various languages.
In terms of custom commands for voice control of the device, WhisPro can support up to 30 tailored commands to meet the unique needs of a product's use case, and the model can require as little as 50KB, determined by the compute resources and accuracy trade-offs.
The WhisPro engine available for TI's CC3235x is optimised for the Arm Cortex-M4 processor, offering extremely low latency and consuming absolute minimal power.
Commenting Naomi Heller, manager of Wi-Fi connectivity products at TI, said, "Across the entire IoT landscape, voice control is becoming a must-have feature for many devices and applications, from wearables to white goods. Our collaboration with CEVA ensures engineers using our proven and reliable SimpleLink Wi-Fi wireless MCUs have access to a voice user interface to address these exciting use cases."
Chad Lucien, Vice President and General Manager of the Sensing and Audio Business Unit at CEVA, stated, "We're delighted to announce our work with Texas Instruments, a powerhouse in the semiconductor industry, to make our WhisPro speech recognition software available for its wireless MCU family. With exceptional performance and the capability to customise the model to each customer's requirements, the hardware offering from TI when paired with our software will enable our mutual customers to deliver high performance, cost-efficient solutions for voice-enabled IoT endpoints and devices."
