DeepL unveils Aure, its most lifelike Voice AI translation model yet

Now available in DeepL Voice, the new model mirrors the nuances of human expression across languages, adapting in real time to changes in a speaker's tone, emotion and delivery

COLOGNE, Germany, Nov. 11, 2026 /PRNewswire/ -- DeepL, a global leader in Language AI, today unveiled Aure, its most dynamic and lifelike Voice AI model yet – introducing a major advancement in how human expression is translated across languages in real time. Powering DeepL Voice, the company's real-time AI voice translation product, Aure continuously adapts translated speech to reflect changes in how someone is speaking, from shifts in tone and emotion to changes in emphasis and pace. The result is voice translation that captures more than words being spoken, preserving more of the personality, nuance and expression that make human conversation natural and impactful.

"Our voice is fundamental to who we are and how we express emotion, collaborate and connect with others. And it's never static. We're constantly adjusting our tone, pace, and expression in ways that can completely change how our words are understood," said Jarek Kutylowski, CEO and Founder, DeepL. "For too long communicating across languages has meant compromising how we express ourselves. With DeepL Voice we are completely reversing that dynamic, adapting to the way people naturally speak rather than asking them to change. And now with Aure, we can capture even the subtlest shifts in expression as they happen and carry them into another language, in someone's own voice and in real time. This is a major step toward a world where language changes, but the person behind it doesn't."

Aure builds on DeepL's recent launch of voice preservation technology, which enables people to communicate across languages in their own recognizable voice rather than a generic synthetic one. While that advancement preserved the identity of a speaker's voice, Aure takes the next step: capturing how that voice changes and responds throughout a conversation, bringing more of the spontaneity and true expression of human speech into real-time translation.

A new approach to real-time speech translation

Achieving this requires a fundamentally different approach to how DeepL's Voice AI models process speech. Previously, DeepL Voice could generate a natural, expressive voice from a static sample, but this relied on the opening moments of a conversation to determine how a persons' translated voice should sound. That captured who was speaking, but not how – a fixed snapshot can't follow a change in emotion, emphasis or pace, or convey delivery that contradicts the words, such as sarcasm.

Aure moves beyond this approach, analyzing incoming speech continuously using both the translated words and source audio to shape how a sentence is delivered. This allows the translated voice to reflect not just what someone says, but how they say it.

This presented several technical challenges. Real conversations are noisy, so Aure needs to isolate the speaker's voice identity and prosody, while filtering out background noise, echo and audio distortions. Carrying a voice across languages is also far more complex than within a single language, particularly without exaggerating accents or distorting how someone sounds.

DeepL Voice also has limited real-world data of the same speaker talking across languages to learn from - the technology does not permanently store meeting transcripts or translations, and deletes this data from memory when a session ends. Aure must therefore adapt to each speaker in real time, before they have even finished a sentence; recognizing the characteristics that make someone's voice unique while accounting for how it naturally changes from one conversation to the next. Someone might sound tired one day, enthusiastic the next, or speak differently under stress. Aure can distinguish these variations and adapt the translated speech to how they sound in the moment.

To address these challenges, DeepL combined several training techniques that separate a speakers' voice identity, prosody and spoken content, allowing each to be carried into a new language independently. Reinforcement learning further refines the output, preserving translation accuracy and speech quality while producing natural-sounding speech, free of accent. Solving all three challenges together is what sets Aure apart.

Bringing more meaning to multilingual conversations

Voice is more than just a way to communicate words. It's how we express our personality, convey confidence and authority, build trust and connect with other people. In conversations that matter, how something is said can be just as important as the words themselves.

For DeepL Voice users, Aure makes it possible to carry more of that expression across languages. Someone might pause before making an important announcement, speak with emphasis and urgency during a critical negotiation, or shift from serious to enthusiastic when sharing good news with a global team. With Aure, these changes are reflected in the translated speech in real time, allowing people to communicate across languages with more of the spontaneity, personality and nuance they bring to conversations in their own language.

For global businesses, the implications extend beyond more natural-sounding translation. The ability to communicate with confidence, credibility and personality matters in everything from building customer relationships and navigating complex negotiations, to building international teams. By preserving more of what makes each person's voice their own, Aure brings us closer to a world where people can participate fully in any conversation, in any language.

Try Aure in DeepL Voice today

Learn more about Aure and try it out yourself here.

Aure's new capabilities are applied automatically, with no additional configuration required, across DeepL Voice for Online Meetings, DeepL Voice for In-Person Conversations and the DeepL API for Voice. The model initially supports English, German, Spanish, French, Italian, Japanese, Portuguese, Korean, with more languages to follow. DeepL Voice will continue to support other languages through its existing voice models.

Aure is the latest in a series of advancements to DeepL Voice as the technology gains traction among global enterprises, with leaders including Pioneer Corporation, Aramark and Avendra International, and Inetum are already using DeepL Voice to support multilingual communication and collaboration. DeepL Voice was also recently recognized in Fast Company's 2026 Next Big Things in Tech list in the Applied AI category.

About DeepL
DeepL is a global AI company building the language infrastructure that powers global business. More than 200,000 business teams and millions of individuals use DeepL's Language AI platform to communicate globally, collaborate and operate across languages in real time. By combining breakthrough AI models with enterprise-grade security and privacy, DeepL enables organizations to work seamlessly across markets and cultures. Founded in 2017 by CEO Jarek Kutylowski, DeepL now has more than 900 employees and is backed by leading investors including Benchmark, IVP and Index Ventures. Learn more at www.deepl.com.

SOURCE DeepL