Deepgram is an automatic speech recognition startup that uses end-to-end deep learning technology to train speech models to learn under complex, real-world conditions—regardless of customers' unique vocabulary, accents, product names and acoustic environments. It's an entirely new approach compared to the other players in the space, like Google, Amazon and Nuance.

CEO Scott Stephenson, came up with the idea for Deepgram while researching the creation of dark matter in a bunker two miles underground as a Ph.D. student. In the few hours that weren't devoted to his research, Scott recorded audio from his life, 24,7, and realized there wasn’t a tool available that would help process his recordings and pinpoint valuable timestamps.

The company has since raised $13.9m to date and intends to use the funds to expand operations and its business reach. I invited Scott onto the podcast to learn more about his tech startup story. Scott shares how they are helping enterprises unlock the potential of their audio data with custom trained speech recognition built for accuracy and scale.

Scott also reveals how businesses are using Speech Recognition APIs to empower their business and what makes Deepgram stand out in a crowded market place dominated by tech giants.