Ceva, Novatek collaborate to enhance smart camera IP system-on-chip
Ceva, a wireless connectivity and smart sensing solutions provider, has extended its almost decade-long partnership with Novatek, a Taiwanese fabless chip design company. Novatek’s new smart camera IP system-on-chip NT98530 will integrate Ceva’s SensPro2 DSP for advanced computer vision and edge AI workload applications. The Novatek NT98530 smart camera SoC targets various applications, including the surveillance, retail, smart city and transportation sectors.
Using Ceva SensPro 500 digital signal processor from the SensPro2 Sensor Hub DSP architecture family, Novatek NT98530 can execute complex computer vision algorithms and neural network inference. Ceva SP500 DSP is up to six times more powerful in computer vision, twice as fast in AI inference tasks, and consumes 20% less energy than its predecessor.
Novatek NT98530 smart camera IP SoC provides high image quality, low bitrate and an enhanced cost-to-performance ratio. The SoC supports 640 megapixels/second, facilitating video performance of up to 8 megapixels at a speed of 60 frames per second. Further, these results can be accomplished while conserving energy during complex AI inference tasks.
“Ceva SP500 DSP delivers superior real-time performance for computer vision, AI-based analytics and multi-sensing sensor fusion at the edge, ensuring our customers have a powerful, flexible edge AI camera solution that they can customize to their requirements,” said Jimmy Su, AVP at Novatek.
Ceva joins Intel Pathfinder for RISC-V ecosystem
Ceva announced that the BX1 and BX2 sound DSPs and its audio front-end software stack would be commercially available through the Intel Pathfinder for the RISC-V platform. RISC-V, an open standard instruction set architecture, has been around for over a decade. Ceva is a strategic member of RISC-V International, supporting its adoption across the smart sensing industry.
Ceva BX1 DSP is designed for battery-operated devices and offers high performance for voice and audio applications. Moreover, this low-power DSP features voice control with always-on speech recognition. The Ceva BX2 DSP offers superior far-field noise reduction to precisely target voice assistants and natural language processing (NLP) at the edge. Additionally, it features object-based audio rendering plus 3D audio processing for intensive auditory applications.
Ceva also offers the ClearVox and WhisPro audio front-end software packages that integrate advanced algorithms to optimize audio and voice processing. WhisPro is a neural network-based keyword-spotting software package that allows users to add wake words and commands to voice-enabled IoT appliances. That said, ClearVox is designed to enhance speech clarity and is ideal for Alexa-like devices.
“We are thrilled to make our Ceva BX audio DSPs and audio front-end software IP available for customers developing with Intel Pathfinder for RISC-V,” said Moshe Sheier, the vice president of marketing at Ceva.
Comments