DeepL Launches Voice-to-Voice Translation for 40 Languages at Spring Event

DeepL, the Cologne-based language AI company, today unveiled Voice-to-Voice, a real-time spoken translation suite that covers more than 40 languages. The announcement came at DeepL's Spring Launch virtual event and marks the company's most ambitious product expansion since it moved beyond its core text translation engine.

The suite ships in four flavours. Voice for Meetings plugs into Microsoft Teams and Zoom, letting participants hear or read translations as colleagues speak in their native language. Voice for Conversation handles one-on-one mobile and web dialogue. Group Conversation targets frontline workers and educational settings. And a Voice-to-Voice API opens the stack to enterprise developers who want to embed real-time interpretation in their own applications. The API became available to self-serve customers today, with the Meetings integrations entering an early-access waitlist.

Language coverage is broad: all 24 official EU languages plus Korean, Vietnamese, Thai, Arabic, Norwegian, Hebrew, Bengali, Tagalog, and others. The system currently works by converting speech to text, translating, and converting back, which introduces a one-to-two sentence delay depending on language pair complexity. DeepL says it plans to ship an end-to-end model that skips the text intermediary step entirely, and voice-preservation technology that maintains the speaker's original tone and timbre is expected before the end of 2026.

"In the global business environment, language is no longer simply a communication issue but a factor that affects organizational operations," said Gonzalo Gaiolas, DeepL's Chief Product Officer. "Companies can now handle global business in real time without hiring expensive interpreters."

The timing is deliberate. DeepL has spent the past year positioning itself as the specialist alternative to general-purpose AI assistants from OpenAI, Google, and Anthropic. A March 2026 benchmark study found DeepL Voice was preferred by 96% of professional linguists evaluating real-time meeting translation tools. The company is betting that domain focus beats breadth when the stakes are high, and that businesses will pay for a tool purpose-built for professional accuracy rather than rely on a chatbot's side feature.

DeepL's CEO, Jarek Kutylowski, has been blunt about the regulatory environment. In comments to Italian media earlier this month, he warned that excessive EU regulation could push companies to the United States, a statement that reads as part negotiating tactic, part genuine frustration. The company has not announced plans to relocate, and its engineering core remains firmly in Germany.

What matters more than the regulatory grumbling is what Voice-to-Voice says about the maturity of European AI companies. DeepL is not raising another mega-round or announcing an IPO. It is shipping product, expanding its surface area from text to speech, and doing so with the kind of focused execution that turns a translation tool into an enterprise communication platform. For European tech, that is the less glamorous but more durable kind of progress.