Ultra-low-power on-device AI processor startup Atlazo has announced the first edge AI system-on-chip (SoC) for tiny devices. The AZ-N1, says the company, is the industry's first fully integrated AI SoC featuring a highly power efficient artificial intelligence (AI) and machine learning (ML) processor, the Axon I, for audio, voice, health monitoring, and sensors. The chip also integrates a low-power CODEC supporting up to four microphones, battery charger, DC/DC regulators, and Bluetooth connectivity.
The AZ-N1 AI SoC, says the company, can handle the most challenging tasks required to process audio, sound, biometrics, and other sensor signals, and classify activities in a fraction of the power budget used by comparable solutions on the market today.
"Atlazo's mission is to enable smart tiny edge devices with a new level of always-on AI capabilities and analytics in an SoC form to unleash new features and applications," says Dr. Karim Arabi, co-founder and CEO of Atlazo, Inc. "We provide comprehensive support to our customers from system design to algorithm development enabling them to add new features to expand their addressable market."
A single Axon I processor supports a variety of AI/ML networks including DNN, LSTM, and GRNN and popular feature extraction techniques such as MFCC. When computing an audio DNN network for example, says the company, the Axon I performs more than 130 inferences per second. Customer applications currently being developed with the AZ-N1 AI SoC include smart TWS earbuds, hearing aids, Personal Sound Amplification Products (PSAPs), and fitness and health monitoring devices.
AZ-N1 AI SoC is sampling to select customers now with a complete hardware and software development kit and key AI algorithms such as keyword spotting and speaker verification.