DSP Group's DBM10 AI SoC Is Designed for Battery-Operated Devices
This low-power chip features a digital signal processor and neural network engine for voice and sensor processing.
The DSP Group, known for wireless and voice processing solutions, has announced a new SoC in its dual-core DSP processor family — the DBM10, designed for battery-powered devices like wearables, tablets, and even smart home gadgets.
“Edge applications for AI are many and diverse, but almost all require the ultimate in terms of low power, small form factor, cost effectiveness, and fast time-to-market, so we are very excited about what the DBM10 brings to current and new customers and partners,” said DSP Group CEO Ofer Elyakim.
This DBM10 is a high-performance, ultra-low-power AI SoC based on a digital signal processor and neural network engine. The 4x4mm chip runs the Hello Edge 30-word detection model at 1MHz and allows porting of large models. Its compact footprint makes it suitable for wearable devices.
The SoC is equipped with the company's nNet Lite NN hardware accelerator that enables its AI capabilities to speed up the execution of NN inferences. The DBM10 follows an open platform approach with a cross-platform toolchain that's compatible with all AI/ML training frameworks. The software framework also supports RTOS and provides a Linux/Android driver interface along with dynamic power management capabilities.
The DBM10 features embedded memory, as well as serial and audio interfaces for communication with other devices in the system, such as an application processor (AP), codecs, microphones, and sensors.
Currently, no details on volume production and shipment are available. However, you can find more information on the DBM10 in DSP Group's press release and product page.
