November 11, 2026 — DeepL unveiled Aure today, a new voice AI model that carries a speaker’s tone, emotion, emphasis and pace into real-time translated speech, not just their words. Why it matters: in a negotiation or a difficult conversation, how something is said often carries the meaning — and until now, machine voice translation flattened most of it.
What happened
Aure now powers DeepL Voice, the company’s real-time voice translation product. DeepL says the model continuously adapts translated speech as a conversation unfolds, reflecting shifts in tone and emotion, changes in emphasis, and changes in pace. It builds on the voice preservation technology DeepL launched in September, which let people speak across languages in their own recognizable voice rather than a generic synthetic one. Aure’s step is different: the earlier technology preserved who is speaking; Aure aims to preserve how they are speaking, moment to moment.
How it works
Previously, DeepL Voice generated an expressive translated voice from a static sample taken from the opening moments of a conversation — a fixed snapshot that could not follow a change in emotion or pace, or delivery that contradicts the words, such as sarcasm. Aure instead analyzes incoming speech continuously, using both the translated words and the original audio to shape how each sentence is delivered. DeepL says it trained the system to separate a speaker’s voice identity, prosody and spoken content so each can be carried into a new language independently, with reinforcement learning used to refine output while preserving translation accuracy and speech quality.
The company is candid about the constraints. Real conversations are noisy, so the model has to isolate a speaker’s voice and delivery while filtering background noise, echo and distortion. There is also very little real-world training data of the same speaker talking across languages — and DeepL says the technology does not permanently store meeting transcripts or translations, deleting session data from memory when a session ends, so Aure has to adapt to each speaker in real time, before they have finished a sentence.
Availability
Aure’s capabilities switch on automatically, with no extra configuration, across DeepL Voice for Online Meetings, DeepL Voice for In-Person Conversations and the DeepL API for Voice. At launch it supports eight languages — English, German, Spanish, French, Italian, Japanese, Portuguese and Korean — with more to follow; DeepL Voice will keep supporting other languages through its existing voice models. DeepL names Pioneer Corporation, Aramark and Avendra International, and Inetum among the enterprises already using DeepL Voice, and notes the product was recognized in Fast Company’s 2026 Next Big Things in Tech list in the Applied AI category. DeepL says more than 200,000 business teams and millions of individuals use its Language AI platform.
Why it matters
Text translation got good years ago; voice translation has lagged because a flat, monotone rendering strips out exactly the signals people use to judge urgency, sincerity and intent. A model that carries emphasis and pace across languages — in the speaker’s own voice, adapting live rather than from a frozen sample — is aimed squarely at the meetings where that nuance decides outcomes: negotiations, customer escalations, global team calls. The honest caveat is that this is DeepL’s own launch claim, with no independent benchmark of how faithfully expression survives translation published alongside it. The feature is live now, so enterprise users can judge it in their own calls rather than taking the announcement’s word for it.
Frequently asked questions
What is DeepL Aure?
Aure is DeepL’s new Voice AI model, unveiled November 11, 2026. It powers DeepL Voice and adapts translated speech in real time to reflect a speaker’s tone, emotion, emphasis and pace.
How is Aure different from DeepL’s voice preservation?
Voice preservation, launched in September, kept a speaker’s recognizable voice across languages. Aure adds the delivery: it continuously tracks how the speaker’s expression changes during a conversation and carries those changes into the translation.
Which languages and products support Aure?
Eight languages at launch — English, German, Spanish, French, Italian, Japanese, Portuguese and Korean — across DeepL Voice for Online Meetings, DeepL Voice for In-Person Conversations and the DeepL API for Voice, applied automatically with no setup.
Does DeepL store voice conversations to train Aure?
DeepL says the technology does not permanently store meeting transcripts or translations and deletes session data from memory when a session ends.
Sources: DeepL via PR Newswire (launch release), The Decoder via DEV Community.