Tech

Speechmatics raises $62M for its inclusive approach to speech-to-text AI – TechCrunch

Last week I wrote about an AI startup developing technology that can change that in real time accent someone’s speech. But what if the AI ​​goal was instead to allow humans to speak as they do, to be understood as they are, while removing some of the biases inherent in many AI systems ? There is a great need for that too, and now a British startup called speechmatics — which developed AI to translate speech into text, regardless of accent or how the person speaks — announces $62 million in funding to grow its business.

US-based Susquehanna Growth Equity led the round with UK investors AlbionVC and IQ Capital is also there. This is Series B, a big step up for Speechmatics. The company was originally founded in 2006 by founder Dr. Tony Robinson spun off from AI research at Cambridge and had previously only raised around $10 million (Albion and IQ are among those earlier backers, along with CIA-backed In-Q-Tel and others).

In the meantime it has built up a customer base of around 170 – it only sells B2B to run consumer or business-oriented services – and while it doesn’t disclose the full list, some of the names include what3words, 3Play Media, Veritone, Deloitte UK and Vonage, who use the technology differently, not just to create transcriptions in the traditional sense; but to record spoken words to support other aspects of an app function, e.g. B. automatic subtitling, or to enable broader accessibility features.

Today its engine is capable of translating speech to text in 34 languages ​​and in addition to using the funding both to further improve accuracy there and for business development, it will also add more languages ​​and explore different use cases such as. for creating speech-to-text that can be used in the more challenging automotive environment (where engine noise and vibration affect the way AIs can pick up the sounds).

“We’ve collected millions of hours of data to fight AI bias. Our goal is to understand each and every voice in multiple languages,” said Katy Wigdahl, the startup’s CEO (a title she shared with Robinson, who recently stepped down from a leadership role).

This is manifested in both the product focus and the company’s mission, and it wants to expand on that as well.

“The way we look at language is global,” Wigdahl said. “Google will have a different package for each English version, but our single package will understand each.” It initially only made its technology available through a private API, which it sold to customers; To attract more users and potentially more paying users, it also now offers developers more open API tools to play with the technology and a drag-and-drop sampler on its website.

And indeed, if one of Speechmatics’ challenges is to train the AI ​​to be more human in its understanding of how humans speak, the other is to unite itself against other major speech-to-text technology vendors to make names.

According to Wigdahl, the company now competes with “big tech” — that is, big companies like Amazon, Google, and Microsoft (which now has Nuance) that have developed speech-recognition engines and offer the technology as a service to third parties.

But it says it consistently performs better than these on tests of understanding when languages ​​are spoken in the many ways they are. (One test quoted to me was Stanford’s “Racial Disparities in speech recognition study, where it recorded “an overall accuracy of 82.8% for African American votes compared to Google (68.6%) and Amazon (68.6)”. It said: “equivalent to a 45% reduction speech Recognition error – the equivalent of three words in an average sentence. It also provided TC with a “competitor-weighted average”:

However, there’s actually a tremendous opportunity here, given that between smaller developers and giant, outsized tech giants like Apple, Google, Microsoft, and Amazon, there are hundreds of giant companies that may not quite be on par (or interest) with Build in in-house AI for this purpose, but if you take a company like Spotify for example will definitely be interestedand definitely don’t want to be dependent on those big companies that are sometimes theirs too competitorsand sometimes hers real slides. (To be clear, Wigdahl didn’t tell me that Spotify is a customer, but did say that this is a typical example of the kind of size and situation that someone might come knocking on Speechmatic’s door.)

Again, that was partly why investors are so keen to fund this company. Susquehanna has a history of backing companies that seem like they could compete with the power players (it was an early and big supporter of Tik Tok).

“The Speechmatics team is undoubtedly a different pedigree of technologists,” Jonathan Klahr, MD of Susquehanna Growth Equity, said in a statement. “We started following Speechmatics when our portfolio companies told us that Speechmatics consistently outperformed all other options, including those of ‘big tech’ players, on accuracy. We stand ready to work with the team to ensure more companies can learn about and adopt this superior technology.” Klahr joins the board with this round.

In fact, as the technology becomes more mainstream and those who make it look for more ways to reduce any friction that there might be when using this technology, the voice has become a big point of opportunity, but also turned into a pain point. So a technology that works for “reading” and understanding all types of voices can potentially be applied to all possible types.

“In our view, voice is becoming the increasingly dominant human-machine interface and Speechmatics leads the category in applying deep learning to speech, with category-defining accuracy and understanding of all industry use cases and requirements,” added Robert Whitby-Smith , a partner, added at AlbionVC. “We have witnessed the impressive growth of the team and product over the past several years since our Series A investment in 2019, and as responsible investors we are pleased to support the company’s inclusive mission of understanding every voice worldwide.”

Speechmatics raises $62M for its inclusive approach to speech-to-text AI – TechCrunch Source link Speechmatics raises $62M for its inclusive approach to speech-to-text AI – TechCrunch

Related Articles

Back to top button