Jensen's New "Arm" & How to Advance Voice Processing with Classic DSP

July 27, 2021

Jensen's New "Arm" & How to Advance Voice Processing with Classic DSP

In this week’s Embedded Insiders, Brandon and Rich try to decide if data sheets specs are reliable, or if industry benchmarks are the only reasonably- accurate measure of component performance without actually testing them yourself.

Later, NVIDIA CEO Jensen Huang discusses how the company’s pending acquisition of Arm is progressing, and what benefits the combined company will bring to each other and the market as a whole if the deal passes regulatory checks. Jensen also weighs in on the semiconductor shortage and how his company plans to keep pricing in check amidst supply and demand volatility.

Finally, Tiera Oliver investigates the world of voice processing at the edge, where companies are adding more AI but not necessarily more processing. Engineers from Knowles Intelligent Audio and Cadence Design Systems weigh in on how classical DSP is being used to make your smart virtual assistant as clever as possible at battery-scale power consumption. And, to bring things full circle, Tiera introduces MLPerf Tiny Inferencing benchmark that can help voice recognition designers determine how fast, accurate, and low power their edge AI stacks can be before building out complete systems.

Tune in.