Achronix’s FPGAs Automatically Recognize Spoken Word

By Chad Cox

Production Editor

Embedded Computing Design

October 13, 2023


Image Credit: Achronix

Santa Clara, California. Achronix Semiconductor Corporation collaborated with for an accelerated automatic speech recognition (ASR) platform designed around the Speedster7t FPGA. Spoken language is transformed into text in over 1,000 concurrent real-time streams and a 20x performance boost over other qualifying products.

"One of the key aspects of the accelerated ASR solution built on Achronix Speedster7t FPGAs is its ability to reduce both OpEx and CapEx while maintaining top-tier performance significantly," noted Bill Jenkins, the Director of AI Product Marketing at Achronix.

The VectorPath accelerator card showcases a Speedster7t FPGA running’s Achronix-FPGA-optimized ASR IP that serves ultra-low latency speech-to-text competences. Up to 20 CPU-only-based servers or 15 GPU cards can be replaced with one card. When 1,000 streams are not required, the AI is flexible and allows a tradeoff between accuracy and performance.

Having a word-error in the 99th percentile, the solution also provides latency of 54 ms end-to-end. It can be adhered to customer specifications or keep the vertical-specific or custom data sets in standard machine learning (ML) frameworks.

“The architecture of the Achronix Speedster7t FPGA with its 2D network on chip (NoC) and ML processor (MLP) arrays gave us the building blocks required to create an ASR product that is significantly more optimized than anything available on the market today,” said Peter Baldwin, CEO of

 For more information, visit and

Chad Cox. Production Editor, Embedded Computing Design, has responsibilities that include handling the news cycle, newsletters, social media, and advertising. Chad graduated from the University of Cincinnati with a B.A. in Cultural and Analytical Literature.

More from Chad