CEVA has announced that it will make its CEVA-BX1 and CEVA-BX2 audio DSPs and audio front-end software stack available for commercial developers through the Intel® Pathfinder for RISC-V* ecosystem. A Strategic Member of RISC-V International, CEVA is making its industry leading audio DSPs and software stack accessible for prototyping and production design through the Intel® FPGA based pre-silicon development platform.
“DSP is an essential technology for processing complex audio and voice workloads and this initiative makes it easier for the RISC-V ecosystem to research, evaluate and eventually productize SoCs incorporating our best-in-class IPs,” said Moshe Sheier, Vice President of Marketing at CEVA.
“The availability of CEVA’s audio DSPs and software IP for the Intel® Pathfinder for RISC-V program is another step in our commitment towards simplifying and accelerating the development of RISC-V platforms,” said Vijay Krishnan, General Manager, RISC-V Ventures from Intel. “Complementary IP plays an important role in the RISC-V ecosystem.”
The CEVA-BX1 processor combines efficient DSP compute capability with high-level programming and compact code size requirements of embedded applications targeting hearables, wearables and other low power smart audio use cases. Using an 11-stage pipeline and 4-way VLIW micro-architecture, it offers parallel processing with a Single Instruction Multiple Data (SIMD) ISA, widely used in neural networks inference, noise reduction and echo cancellation, as well as high accuracy sensor fusion algorithms.
The CEVA-BX2 is targeted for high performance audio devices such as Smart TVs and conferencing devices, smart speakers, soundbars, and car infotainment systems, using quad 32×32-bit MACs and octal 16×16-bit MACs, with enhanced capability for supporting 16×8-bit and 8×8-bit MAC operations. Its 11-stage pipeline and 5-way VLIW micro-architecture, support parallel processing with dual scalar compute engines, load/store and program control.
The CEVA-BX Instruction Set Architecture (ISA) incorporates support for Single Instruction Multiple Data (SIMD) as well as optional floating point units for high accuracy algorithms. The CEVA-BX is accompanied by a comprehensive software development tool chain, including an advanced LLVM compiler, eclipse-based debugger, DSP and neural network compute libraries, neural network frameworks support and Real Time Operating Systems (RTOS).
Developed in-house and leveraging CEVA’s vast expertise in audio and voice processing, CEVA offers its ClearVox and WhisPro audio front-end software packages. The AVS certified ClearVox audio front end incorporates advanced algorithms that cope with different acoustic scenarios and microphone configurations, including optimized software for speaker direction of arrival, multi-mic beamforming, noise suppression and acoustic echo cancellation, as well as the related firmware and driver software.
WhisPro is an AVS certified neural network-based keyword spotting (KWS) software package, allowing customers to add wake-words and commands to voice-enabled IoT devices. As a voice user interface, WhisPro is targeting the rapidly growing use of voice as a primary human interface for intelligent cloud-based services and edge devices. WhisPro provides a customizable (any language, any keyword) wake-word and keyword voice interface. This allows users of headsets, wearables, smart TVs, smart speakers, smartphones, infotainment systems, and other voice-enabled IoT devices to interact with cloud-based voice assistant services, as well as enabling voice activation on edge devices through its small footprint and processing needs, without cloud connectivity.