This event concluded on Mar 30, 2022. The content below has been archived for your convenience.
Join us in Burlingame to take part in the sharing, learning and celebrating tinyML. In this faction, we stand on the shoulders of giants. It is the incredibly open and collaborative nature of ML technology which allows this field to advance so quickly. From its inception in 2019, the tinyML community has grown tremendously and has benefited greatly by supporting one another.
In this session, we will look at a low-power real-time embedded mask-based beamformer for voice UI systems. Our solution is designed to improve wake word and voice commands trigger rates in real-life noisy scenarios and do not require any cloud interaction. The voice UI system is built with a denoising audio front-end, a wake word engine, and a voice command engine. Such a system is constrained by low-power and high-performance requirements. In particular, real-time processing and noise robustness are the most challenging issues. To meet the challenges, our solution is designed for embedded systems and is hybrid—a neural network is feeding a MWF-based multichannel processing algorithm. The 18k-parameter network is quantized in 16 bits and runs efficiently at 12MHz on anRT1060 MCU. In a 3-mics configuration, the complete speech enhancement solution is running on average at 160Mhz on the Arm Cortex-M7 device and leads to a 40% hit-rate improvement.
来NXP's booth #39 and hear how we are bringing machine learning and deep learning to the edge.
Empowering the edge everywhere requires machine learning tools and software for all. See the EdgeVerse processor and MCU portfolio empowered by the eIQ ML Software IDE deliver low cost-of-ownership and safer, more secure products.
