Wispr Flow has raised $280 million in Series B funding at a $2 billion valuation as the voice technology company looks to accelerate development of its AI-powered speech platform. The round was led by longtime investor Menlo Ventures, with participation from existing backers Notable Capital, NEA, Neo Ventures, 8VC, and MVP Ventures, alongside new investors including Acrew, Forerunner, Goodwater, Peak XV, Together Fund, and PLUS Capital. The financing brings Wispr Flow’s total capital raised to $361 million and comes as voice interaction becomes an increasingly central part of how its customers communicate and work.
Funding Supports Voice AI Expansion
Wispr Flow said the new capital will primarily support research aimed at improving speech recognition accuracy and reducing the need for users to correct dictated text. The company measures performance through what it calls its “zero edit rate,” which tracks how much dictated content can be used exactly as generated without further changes. Improving that metric has become a central development priority as customers increasingly use voice for professional messages, documents, and other tasks where transcription errors can disrupt productivity.
Canto Targets Real-World Speech Conditions
Alongside the funding announcement, Wispr Flow unveiled a preview of Canto, its first proprietary speech model designed to perform under challenging real-world conditions. The company said Canto can reduce word error rates from more than 30% to approximately 5% to 10% in environments involving background noise, wind, music, strong accents, or nearby conversations. Across more typical usage scenarios, Wispr Flow expects the model to reduce the number of dictations requiring edits by approximately 30% to 35%.
Multilingual Speech Becomes a Priority
Canto is also designed to handle users who switch between languages or speaking styles during conversations, an area where traditional speech recognition systems can struggle. Wispr Flow highlighted Hinglish, which combines Hindi and English, as an example where accurate transcription requires understanding not only spoken words but also how users expect those words to appear in written form. The model can additionally draw on personalized information such as dictionaries and names to improve its ability to recognize vocabulary specific to individual users.
Building Beyond Dictation
Wispr Flow is positioning the technology as part of a broader effort to develop AI systems that understand user intent, preserve context, and eventually translate spoken instructions into completed actions. The company has expanded beyond its core Flow dictation product with Notetaker, a recently launched tool that captures information from meetings, while its Wispr Advanced Interfaces Lab is researching new approaches to human-computer interaction. Chief Scientist Ariya Rastrow, previously a founding member of the team behind Amazon Alexa, leads the lab as Wispr Flow explores voice and other modalities as interfaces for artificial intelligence.
Growing Adoption Across Enterprises
The company said users have now produced more than 60 billion words through Flow, indicating growing adoption of voice as an alternative to traditional keyboard input. Wispr Flow also reports that its technology is used by employees at nearly all Fortune 500 companies and across more than 10,000 enterprises. The growing user base has increased the importance of reliability as the platform moves from experimental voice interaction toward more frequent use in professional workflows.
Wispr Flow’s $280 million Series B gives the company substantial new capital to improve speech recognition while extending its voice technology into broader AI interfaces. The launch of Canto reflects its effort to address accuracy, multilingual speech, and noisy environments as voice becomes more widely used for everyday work. With a $2 billion valuation and expanding enterprise adoption, the company is betting that increasingly reliable voice interaction can become a core way users communicate with computers and AI systems.