The joint efforts pair ST’s STM32 hardware and software with Sensory’s voice-control technologies, including the new VoiceHub online portal that supports seamless creation of embedded speech-recognition models using custom wake words, voice-control command sets, and large natural-language grammars in almost twenty languages and dialects.

The solution is based on an STM32Cube software extension package and runs on a high-performance STM32H7 MCU, taking advantage of its architecture, internal Flash, SRAM, and high CPU speed. This combination plays a key role in increasing voice-control accuracy and minimizing command-recognition time. Hosting the voice application and speech models in the generous on-chip memory of the high-performance STM32 MCUs further boosts the system integration and ease of use, as well as lowers cost of ownership.

“This collaboration sets to jump-start the development of embedded-voice user interfaces, adding friction-free command control and custom wake word to any device, from wearables to smart-home appliances,” said Ricardo De Sa Earp, Executive Vice President, General-Purpose Microcontroller Sub-Group Vice President, STMicroelectronics. “The unique combination of ST and Sensory technologies will enable the STM32 user community to deploy ‘Voice AI on the edge’ without any programming, data-science, or machine-learning expertise, for free in prototypes and with favorable licensing terms in production.”

“Sensory designed our VoiceHub so developers could quickly and painlessly create custom speech-recognition models. However, after creating a custom model, integrating the model onto hardware, and moving to licensing terms were the next hurdles that needed to be cleared,” said Todd Mozer, CEO, Sensory. “This world-class collaboration with ST creates a complete software, hardware, and licensing package for embedded speech recognition across the STM32 family and makes adding Voice UIs, simple.”