Deepgram lands new cash to grow its enterprise voice-recognition business
Deepgram, a company developing voice-recognition tech for the enterprise, today raised $47 million in new funding led by Madrona Venture Group with participation from Citi Ventures and Alkeon. An extension of Deepgram’s Series B that kicked off in February 2021, led by Tiger Global, it brings the startup’s total raised to $86 million, which CEO Scott Stephenson says is being put toward R&D in areas like emotion detection, intent recognition, summarization, topic detection, translation and redaction.
” We are pleased that Deepgram achieved its highest ever pre- and post-money valuation despite the difficult market conditions,” Stephenson said to TechCrunch via email. (Unfortunately, he would not reveal the exact valuation. “We believe Deepgram is well-positioned to thrive in this more difficult macroeconomic environment. Deepgram’s speech AI technology is the core enabling technology behind many applications of our customers. As companies seek greater efficiency , the demand for speech understanding continues to grow.
Launched in 2015, Deepgram focuses on building custom voice-recognition solutions for customers such as Spotify, Auth0 and even NASA. Deepgram’s data scientists collect, label, evaluate, and label speech data. This allows them to create speech-recognition models that are able to understand brands and jargon, recognize accents and capture a variety of languages, and adapt to difficult audio environments. Deepgram, for NASA, created a model to transcribe communications between Mission Control (and the International Space Station)
“Audio is one of the largest untapped data resources in the world. Stephenson stated that audio data is difficult to use because it is an unstructured type of data and cannot be mined for insights. Deepgram is able to take unstructured audio data, and structure it as text and metadata at high speed and low costs for enterprise scale…. Companies can send all customer audio — hundreds or millions of hours — to Deepgram to be transcribed .”
Where does Deepgram get the audio data it uses to train its models? Stephenson was a little coy about this, but he did admit that Deepgram uses customer data for system improvement. He quickly pointed out that Deepgram complies with GDPR, and that users can request the deletion of their data at any time.
Stephenson stated that
“Deepgram models are primarily trained using data generated or collected by our data curation specialists, along with anonymized data submitted from our users. “Training models with real-world data is a cornerstone in our product’s quality. It’s what allows machine-learning systems like ours produce human-like results. If they choose .”
, they can opt out of having their anonymized data used in training.
Companies can integrate Deepgram’s API into their tech stacks in order to enable voice-based automations, and customer experiences. Deepgram’s API allows companies to integrate the platform into their tech stacks. This allows them to enable voice-based automations and customer experiences. Deepgram has been supported by In-Q-Tel (the CIA’s strategic investment arm) in the past. )
Deepgram — a Y Combinator graduate founded by Stephenson and Noah Shutty, a University of Michigan physics graduate — competes with a number of vendors in a speech-recognition market that could be worth $48.8 billion by 2030, according to one (optimistic?) source. Tech giants like Nuance, Cisco, Google, Microsoft and Amazon offer real-time voice transcription and captioning services, as do startups like Otter, Speechmatics, Voicera and Verbit.
Tech has many hurdles to overcome. According to a 2022 report by Speechmatics, 29% of execs have observed AI bias in voice technologies — specifically imbalances in the types of voices that are understood by speech recognition. The demand is strong enough to support the variety of vendors; Stephenson claims that Deepgram gross margins are comparable with top-performing software companies .”
This contrasts with the consumer voice-recognition industry, which has taken a downturn in recent years. Amazon’s Alexa division is reportedly on pace to lose $10 billion this year. And Google is rumored to be eyeing cuts to Google Assistant development in favor of more profitable projects.
In recent months, Stephenson stated that Deepgram’s focus was on on-the fly language translation, sentiment analysis, and split transcripts for multiway conversations. The company’s also scaling, now reaching over 300 customers and more than 15,000 users.
On the hunt for new business, Deepgram recently launched the Deepgram Startup Program, which offers $10 million in free speech-recognition credits on Deepgram’s platform to startups in education and corporate. Participating firms don’t have to pay any fees and can use the funds along with existing grant, seed and incubator benefits.
“Deepgram continues to grow quickly. Stephenson stated that Deepgram is a foundational AI infrastructure firm and has not seen a decrease in demand. “In fact, we have seen businesses look for ways of cutting costs and delegating repetitive, menial tasks, giving humans more time for more important, consequential work. This could be used to reduce large cloud compute costs, such as switching big cloud transcription to Deepgram’s transcription product or in new use cases such as drive-thru ordering. Also, triaging the first round .”
of customer service responses.
Deepgram currently has 146 employees distributed across offices in Ann Arbor and San Francisco. Stephenson declined to answer questions about the hiring plans for the remainder of the year. He was probably aware of the uncertainty of the current global economic climate and the difficulties of committing to a fixed number.
I’m a journalist who specializes in investigative reporting and writing. I have written for the New York Times and other publications.