Deepgram Inc., maker of artificial intelligence-based automated transcription technology it delivers as a service, today announced that it has raised an additional $47 million to complete a previously announced Series B venture round of $72 million, bringing its total funding to $86 million.

San Francisco-based Deepgram’s voice-recognition engine covers more than 30 languages and uses deep learning software complemented by human workers to process interviews, videos, speeches, meetings and other spoken audio into text. Grand View Research Inc. has valued the global voice and speech recognition valued at $14.42 billion in 2021 with a 15% annual growth rate.

Raising the money in the current financial climate posed a lot more challenges than would have been met even six months ago, said Chief Executive Scott Stephenson. “There is no way of knowing how much money could have been raised had the fruitful market environment from previous quarters carried into 2022,” he said, citing what he said is the combination of the company’s rapid growth and strong profit margins.

‘Position of strength’

Stephenson said Deepgram will use the funding to invest in research and engineering “to address today’s most difficult speech understanding challenges, including emotion detection, intent recognition, summarization, topic detection, translation, redaction, entity detection and more.

“Deepgram anticipates being able to make it to the other side of even a protracted economic downturn in a position of strength,” he said.

The funding was led by Madrona Venture Group LLC, which has been an active investor in AI and machine learning for over a decade, Stephenson said. “Their roots in Seattle’s tech ecosystem will be an advantage as Deepgram explores collaboration opportunities with large cloud computing providers,” he said in an apparent reference to Amazon Web Services Inc. and Microsoft Corp.

The task of understanding and transcribing speech is increasingly “a solved problem,” Stephenson said. The next phases of development are about improved legibility and formatting options followed by conversational analysis and contextualizing by such factors as tone of voice, syntax and sentiment.

Deepgram makes its platform available via an application program interface with a tiered pricing schedule that it publishes on its website. While most customers use Deepgram’s hosted application programming interface program, the company also has an on-premises deployment option for users in highly regulated industries that require local data processing for security or privacy reasons.

Image: geralt/Pixabay