Speak builds a language-learning platform that focuses on speaking by generating audio conversations and listening to learners, using speech recognition, natural language processing and generative AI with no humans in the loop. The product is positioned as a complement to existing study methods, emphasizing drills, repetition and real-world context rather than grammar-first or test-prep approaches. Speak currently supports learning English from eight originating languages and plans to expand target languages starting with Spanish and French. The company reports its app has been downloaded more than 10 million times, average usage is roughly 10–20 minutes per day, and consumer pricing is typically $20 per month or $99 per year; Speak for Business has over 200 customers. Leadership says the firm may build tools to quantify fluency long term and could introduce more consumer-facing engagement mechanics, but it will prioritize learning efficacy over gamification. Speak is an AI-powered language-learning app that teaches users by having them learn speaking patterns and practice repetition rather than memorizing vocabulary and grammar. Since launching in its inaugural market of South Korea in 2019, Speak has grown to over 10 million users, with its user base doubling every year for the last five years and customers in more than 40 countries. The app began with English, added Spanish powered by a speech recognition model trained on in-house data, and plans to launch French next. Speak charges $20 per month or $99 per year for access to all features, including review materials and one-off courses. The company has about 75 employees across offices in San Francisco, Seoul, Tokyo and Ljubljana and is developing new models to deliver better real-time feedback on tone and pronunciation. The startup recently saw its valuation double to roughly $500 million following new capital, bringing total funding to $84 million. Speak provides mobile apps for Android and iOS that let users practice open-ended conversations with an AI tutor that gives feedback on pronunciation, grammar, and vocabulary. The company first honed its product in South Korea and is live in around 20 countries, including Japan, Taiwan, Germany, France, Brazil and Mexico. Speak claims well over 100,000 subscribers in South Korea and says it has helped roughly 3 million people—about 6% of Korea’s population—learn English. After an investment from the OpenAI Startup Fund, Speak adopted OpenAI technologies including GPT-4 and the Whisper API and launched a ChatGPT plugin. CEO Connor Zwick says the new funding will support launches in more markets, with a planned U.S. launch by the end of the year. Speak offers interactive speaking experiences that let learners hold open-ended conversations with an AI tutor while receiving feedback on pronunciation, grammar and vocabulary. The company combines OpenAI models with in-house speech, speech-generation and conversation models and is building a labeled dataset of second-language speaking examples to improve performance for accented speakers. Speak reports over 15 million lessons started annually, about 100,000 active subscribers and "double-digit million" annual recurring revenue. The product is offered via auto-renewing monthly and annual subscriptions granting access to courses, electives, review content and AI-guided practice. Founded in 2016 by Connor Zwick and Andrew Hsu, Speak has about 40 employees across offices in San Francisco (HQ), Seoul and Ljubljana. The company plans to expand to new languages and markets, including Japan, and invest in features that leverage text-generating models like OpenAI's GPT-3. Management says Speak has multiple years of runway and will use new funding to grow engineering, machine learning, product, marketing, content and operations.

Total raised

$141M

Funding rounds

4

Latest round

Series C

Latest activity

Dec 2024