OREANDA-NEWS. April 02, 2018. As the AI ecosystem continues to expand, NVIDIA revealed today that speech analytics startup Deepgram is the newest member of the GPU Ventures portfolio.

Founded in 2015, San Francisco-based Deepgram is the first end-to-end deep learning speech recognition system in production that uses NVIDIA GPUs for inferencing and training.

“Using GPUs made our inferencing 100 times more efficient than when using CPUs,” said Scott Stephenson, CEO and co-founder of Deepgram. “Our technology coupled with NVIDIA’s expertise in AI allows us to make a greater impact in speech analysis.”

Deepgram is also part of our Inception virtual accelerator program, which includes more than 2,800 startups. Deepgram competed in the Inception pitch day competition last May. In this year’s Inception awards finale, six finalists will compete for a combined cash prize of $1 million at GTC in San Jose, on March 27.

Deepgram’s automatic transcription tool, Deepgram Brain, searches for keywords in transcripts by both sound and text. It also helps businesses analyze customer calls to improve their service. The company already serves over 5,000 clients, from financial institutions to journalists.

“While many companies already implement accelerated speech recognition, true speech analytics has until recently been largely untouched by deep learning,” said Jeff Herbst, vice president of business development at NVIDIA. “Deepgram has done an amazing job introducing deep learning to this field, and we look forward to working closely with them to advance deep learning-driven speech analytics to the next level.”