Achronix Semiconductor Corporation, in collaboration with Myrtle.ai, has introduced an accelerated automatic speech recognition (ASR) solution revolutionizing the industry. The innovation, based on the Speedster7t FPGA, processes spoken language into text across 1,000 concurrent real-time streams with remarkable accuracy and rapid response times. Significantly outperforming its competitors, the solution boasts up to a 20 times improvement in performance.
The ASR solution, which will be showcased at the upcoming SC23 conference in Denver from November 12-17, 2023, at booth 2019, is powered by a VectorPath accelerator card featuring a Speedster7t FPGA running Myrtle.ai’s Achronix-FPGA-optimized ASR IP. This technology offers unparalleled real-time, ultra-low latency speech-to-text capabilities. Notably, a single card can replace up to 20 CPU-based servers or 15 GPU cards, making it highly efficient and cost-effective.
The Achronix-Myrtle.ai accelerated ASR solution sets new standards with a word-error rate and 99th percentile latency of 54 ms end-to-end, including pre- and post-processing, along with data movement to and from the CPU. Furthermore, its adaptability shines through; it can be easily customized or retrained using standard machine learning frameworks, accommodating vertical-specific or custom data sets.
Bill Jenkins, Director of AI Product Marketing at Achronix, emphasized the solution’s cost-efficiency, stating, “This solution powered by a Speedster7t FPGA can reduce costs by up to 90% compared to traditional CPU/GPU-based server solutions, whether enterprise or in the cloud. This capability translates to tangible business savings while providing exceptional real-time speech-to-text capabilities.”
Peter Baldwin, CEO of Myrtle.ai, lauded the unique architecture of Achronix Speedster7t FPGA, attributing it to the product’s unmatched optimization. “The extremely low latency inherent in these FPGAs makes them ideal for real-time workloads. We are excited to provide users the ability to scale up their ASR services at lower cost and faster than ever before,” he said.
Achronix’s ASR solution is set to transform industries reliant on swift and accurate speech-to-text conversion. Compatible with major deep learning frameworks such as PyTorch and easily re-trainable for multiple languages or specialties, the solution is currently being deployed with early-access customers and is now available to the general market.
