The technology can be implemented in a small footprint system-on-chip (SoC) for always-on battery powered applications such as smartphones and wireless headsets and, says the company, offers performance that up to now has not been achieved in standalone ultra-low-power ASIC implementations. The company is planning to build specialized deep learning chips for voice and audio processing and expand the framework to other applications in adjacent markets.
“Smartphone and wireless headset chipsets must meet strict and rigorous specifications,” says Mouna Elkhatib, CEO, AON. “Our team is well qualified to integrate the next generation of audio chips based on new AI technology in these systems since we have led voice and audio chip developments at leading semiconductor companies and we fully understand all the requirements. This is the reason OEMs are engaging with us at such an early stage.”
The company says it has demonstrated the alpha release of their software to leading OEMs and is currently prototyping the solution on the Xilinx Zynq-7000 FPGA family to take advantage of hardware programmability prior to SoC tape-out.
Founded in 2018, AONdevices specializes in low-power AI SoCs and algorithms. Its initial focus is on voice and audio recognition applications.
Voice control technology offers improved AI
AI-based speech recognition technology for voice assistants, IoT
SoC supports AI voice-driven smart home devices
Voice assistant integration top smart home trend at CES
Voice-powered IoT opportunities are many, says report