Syntiant was formed in 2017 to bring deep learning down to the power consumption level required by wearables such as earphones and smartphones. The company claims its products are suitable for always-on voice, sensor and image tasks.
The NDP architecture is intended to “eliminate” data movement penalties, Syntiant claims and is focused on always-on functions for battery-powered equipment, such as keyword spotting, speaker identification, wake word, event detection, image recognition and sensor synthesis. Hence the good match with Infineon’s microphone technology.
“Analog neural networks and deep learning are a match made in heaven, delivering more than 50 times improvement in efficiency versus traditional digital stored-program architectures,” said Kurt Busch, CEO of Syntiant, in a statement.
“Through our collaboration with Syntiant, we are rapidly evolving edge capabilities and can achieve breakthrough AI and sensor fusion – continuing to deliver industry leading solutions to our customers,” said Adrian Mikolajczak, head of Infineon’s Silicon Valley Innovation Center.
The collaboration is likely to impact another Infineon audio partner – XMOS Semiconductor Ltd. (Bristol, England) – in which Infineon led a $15 million investment round (see Infineon backs XMOS to work where audio meets AI).
“For machine learning to be deployed in edge devices, it has to become much more power efficient. We believe ultra-low-power, analog neural networks like the ones Syntiant provides could dramatically boost the adoption of distributed AI,” said Wendell Brooks, president of Intel Capital.
Intel Capital recently closed a Series A round of equity finance with venture capital firms Seraph Group, Danhua Capital and Embark Ventures also participating. The size of the round was not disclosed but it was said that the money would be used for the further development and commercialization of Syntiant’s analog neural network technology.
Related links and articles: