Languages are usually taught by first practicing reading and writing, but native speakers always start learning a language by listening and speaking. Speak has built a language teaching platform that focuses on how native speakers learn. The startup uses AI to generate voice conversations and listen to user responses to improve language comprehension.
Now, Speak is announcing milestone funding that further strengthens its progress. A Series C of $78 million would bump the valuation to $1 billion.
The funding round is led by Accel, with participation from previous backers OpenAI (via Startup Fund), Khosla Ventures, and Y Combinator.
This funding is a major step forward for the startup. Speke confirmed a $20 million Series B extension at a $500 million valuation just six months ago.
Part of this investment activity can be attributed to the tremendous enthusiasm for Generate AI and one of Speak's key investors. As you can imagine, OpenAI is more than just a financial backer here. Speak leverages its own technology to power its platform, and the ChatGPT maker is also an early partner for the latest voice technology. Relatedly, Speak is proving to be one of the commercial opportunities for GenAI.
“Our investment in Speak in 2022 was driven by our shared vision of revolutionizing language learning with AI,” Ian Hathaway, Partner at OpenAI's Startup Fund, said in a statement. “We are excited to see their world-class AI talent and unique product vision create innovative learning experiences for our rapidly growing user base around the world.”
One of the main goals of this funding is to expand the number of languages users can learn with Speak, and therefore its customer base, starting with Spanish and French.
Eight-year-old Speak has so far focused on people studying just one language: English, the world's most popular learning language. Speak provides study and review materials designed around courses. The idea here is to supplement what the user may have learned elsewhere. The company currently lists eight native languages for learning English, based on the language groups that have been the most popular among English learners to date.
“Of the 1.5 billion people trying to learn English, the vast majority have spent more than 15 years studying it diligently. They know vocabulary and grammar better than any of us. But , the problem is they don't have the ability to talk about it,” said CEO Connor Zwick, who co-founded the company with Andrew Hsu (CTO). “For us, the value proposition has always been really about teaching people how to communicate in language.”
To be clear, this 1.5 billion number is the total addressable market for Speak, not the number of users. The actual number of active users is not disclosed. As a guide, Hsu says Speak's app has been downloaded more than 10 million times, averages about 10 to 20 minutes of use per day, costs $20 per month or $99 per year, and typically costs less than a human home. He said it's a fraction of the cost of hiring a teacher. Work on improving your conversation.
The company says its enterprise tier Speak for Business has more than 200 customers.
Zwick explains that Speak is both a learning method and a technology platform that works in a three-step process.
First, you are thrown into listening and speaking. It's an interesting approach considering Sue and Zwick met and began working together after being classmates as Teal Fellows. In theory, you could be thrown into building a company instead of spending years on it. First of all, learn.
“I'm not going to explain all the grammar rules,” Zwick said of the program's first steps.
You will then be asked to apply that new term or phrase over and over again. “Basically, you just practice saying it out loud in different other languages and you'll be able to pronounce it automatically without translation.”
Third, Speak “uses AI to present phrases in a real-world context, which is how we actually anchor them,” Zwick says.
Ironically, the goal is to enable learners to speak the new language with humans, but there are no humans in the loop. All of these are created using speech recognition, natural language processing, generative AI, and more to tailor learning to the learner.
But will it work? At this time, Speak does not have any integration with standardized language learning qualifications (if you believe in that particular metric). This is a path that other companies in the online language learning space are also taking, such as Duolingo, which provides thousands of English universities with English tests that international students can use to prove their English proficiency.
“We're clearly trying not to be a test prep solution because, frankly, unfortunately, all the tests to date are imperfect,” Zwick said. “What ends up happening is people are going to test the game, and they're going to try to be able to take that test, and they're going to be able to actually communicate in the real world, It's not like you're trying to be able to use the language or anything. The only way to really get a proper assessment is to have an expert consult you. But there are tests that do that for everyone. How can I apply it?”
But Hsu hinted that this could be part of the company's plans in the long term. “This system we are currently building to quantify fluency and effectiveness… [it] “This will be very helpful in launching things like true, actually accurate English fluency scores and tests,” he said.
Gamification is also not an area that Speke has studied in the past, and is outside of one of the big trends in online learning over the past few years (at least for now). But companies like Duolingo and Kahoot are leaning into it, turning the art of learning into something more like a game. New services like Eleven Labs' multilingual AI agent, launched in November, open up the field for more language learning services exploring ways to promote speaking that incorporate features like gamification. There is a possibility.
Some financings may also introduce such consumer-driven models.
Now that the company has more staff, “there is scope to introduce more of these behavioral mechanisms into the app to drive positive change for users,” Zwick said. But that doesn't come at the expense of actual learning. “When there is a tug-of-war between gamification, engagement, and effectiveness, we choose 100% effectiveness,” he said.
Ben Cuazzo, a partner at Accel who led the investment round, will join Speak's board of directors. “Speak has emerged as the preeminent player in consumer AI, demonstrating exceptional growth and market potential,” he said in a statement.