Reinforcing its position as a leading solution provider of deep learning solutions for edge AI applications, the latest Syntiant NDP115 Neural Decision Processor brings multi-modal artificial intelligence to edge devices in all industries. The new NDP115 is particularly suited to deal with the new audio development challenges, since its built-in DNN and DSP capabilities support both machine learning and traditional audio and voice algorithms.
The Syntiant NDP115 is a special purpose processor for deep learning and is ideal for always-on applications in battery-powered devices. The NDP115 applies neural processing to run multiple applications simultaneously with minimal power consumption.
Built using the Syntiant Core 2 programmable deep learning architecture, the NDP115 is designed to natively run multiple deep neural networks on a variety of architectures, such as CNNs, RNNs and fully connected networks. To bring some perspective, the new NDP115 brings a level of ML performance that delivers 25x the tensor throughput than the Syntiant Core 1 - found in the production Syntiant NDP100 - and is able to run speech inferences at a mere 280 microwatts. A programmable Tensilica HiFi 3 DSP is included for flexible feature extraction and classical audio processing.
Available in a 2.1 mm x 2.1 mm 25-ball WLBGA package (0.4mm ball pitch), the NDP115 is small, yet powerful enough to deliver highly accurate, cloud-free audio and sensor AI processing for a wide variety of edge products, ranging from industrial to consumer electronics, being particularly attractive for hearables and wearables, all the way to the smart home, such as for remote controls.
"The NDP115 offers the multi-modal functionality of our Core 2 inference engine in a compact, cost- and power-efficient solution for ultra-power and size constrained applications," says Kurt Busch, CEO of Syntiant. "Combined with our machine learning software models, the purpose-built NDP115 enables developers to easily deploy full audio and sensor processing solutions that address all kinds of consumer and commercial use cases, from home security to industrial IoT."
Ideal for close-talk, far-field, keyword speech and audio event classification applications, the NDP115 also supports I2C and pulse density modulation (PDM) interfaces for sensor fusion, multi-axis acceleration, tilt, magnetic field and pressure. The embedded user programmable Tensilica HiFi-3 DSP with 144KB 64-bit data RAM, and 64KB 64-bit instruction RAM, combined with an embedded Arm Cortex M0 microcontroller, are ready to deal with the direct input of PCM audio and support up to five audio streams.
The Syntiant NDP115 is now shipping in production volumes. Pricing for 10Ku quantities is $3.25 per unit.
www.syntiant.com
