Renesas Electronics Corporation and Syntiant Corp. recently announced the joint development of a voice-controlled multimodal AI solution that enables low-power, contactless image processing in vision AI-based IoT and edge systems such as self-checkout machines, security cameras, and video conferencing systems, as well as smart home appliances such as robotic vacuum cleaners.
The new solution combines Renesas RZ/V series vision AI microprocessors (MPUs) with the low-power multimodal, multi-functional Syntiant NDP120 Neural Decision Processor™ to deliver advanced speech and image processing capabilities. The solution features a constant-on function that can be quickly activated by voice from standby mode to perform vision tasks such as object recognition, facial recognition, and other critical functions for systems such as security cameras. For example, when a user-defined voice prompt activates and runs the system, the vision AI will recognize track operator behavior and control the operation or alert when suspicious behavior is detected.
The multimodal architecture makes it easier to create a contactless user experience for vision AI-based systems. Because software development can be done independently of the vision AI function, utilizing a dedicated, energy-efficient chip for voice recognition reduces standby power consumption and accelerates system development.
We expect demand for multimodal systems with multiple input information streams, including images and voice, to increase as an effective way to improve ease of use and security," said Keito Shinoda, senior vice president and head of the SoC Division, IoT and Infrastructure Business Unit, Renesas Electronics. The close collaboration between Renesas, with its advanced low-power image AI technology, and Syntiant, a superior provider of voice AI technology, will accelerate the adoption of low-power, ultra-small intelligent voice AI technology in embedded systems and bring new integrated solutions to customers worldwide."
Syntiant CEO Kurt Busch said, "Voice-based user interfaces give customers a new user experience, taking next-generation innovative ideas from concept to reality. We have shipped more than 15 million deep learning NDPs worldwide, enabling always-on voice functionality for various consumer and industrial IoT applications. Our collaboration with Renesas has created a powerful, low-power speech and imaging solution that we believe will accelerate customer adoption in a rich set of devices and cases around the world."
Renesas RZ/V series MPUs for vision AI integrate Renesas' unique DRP-AI (Dynamically Configurable Processor-AI) gas pedal and thoroughly combine high-precision AI inference with industry-leading power efficiency. The superior power efficiency eliminates the need for cooling measures such as heat sinks or cooling fans, thereby reducing bill of materials (BOM) costs and enabling the integration of vision AI into a wide range of embedded applications.
The Syntiant NDP120 chip integrates sophisticated AI functionality for numerous high-precision hands-free voice applications, including speaker recognition, keyword detection, multiple wake-up words, and local command recognition. Packaged with the Syntiant Core 2™ neural network inference engine, the NDP120 can also run multiple applications simultaneously and reduce power consumption to 1mW battery power.
The new voice-controlled multimodal AI solution is part of Renesas Electronics' Winning Combinations, a selection of mutually compatible devices from Renesas' broader product portfolio to build higher-end prototyping platforms for customers to accelerate time to market and reduce risk. Renesas' Winning Combinations includes a superior portfolio of analog, power, and embedded processing products to help customers accelerate their designs and reduce time to enter the market.
Renesas Electronics Corporation is a Japanese semiconductor manufacturer that delivers trusted embedded design innovation with complete semiconductor solutions that enable billions of connected, intelligent devices to enhance the way people work and live. Learn more at renesas.com.
Founded in 2017 and headquartered in Irvine, California, Syntiant Corp. is a leader in moving artificial intelligence and machine learning from the cloud to edge devices.
Syntiant's advanced chip solutions merge deep learning with semiconductor design to produce ultra-low-power, high-performance, deep neural network processors for always-on applications in battery-powered devices, such as smartphones, smart speakers, earbuds, hearing aids, and laptops. More information can be found by visiting www.syntiant.com.
本文转载自Asian Tech Press: Renesas Electronics and Syntiant Develop Voice-Controlled Multimodal AI Solution