THE WORLD IS NOT STANDING STILL.RSS
BIG CHANGE.
BIG CHANGE, one tap away
BIG CHANGE, one tap away

Make room for the big picture.

Add BIG CHANGE to your home screen or desktop. It opens in its own window, with the latest stories when you’re online.

iPhone & iPad
Open bigchange.ai in Safari. Tap Share, then Add to Home Screen. Turn on Open as Web App if shown, then tap Add.
Android
Open bigchange.ai in Chrome. In the browser menu, choose Install app or Add to Home screen.
Desktop
In Chrome or Edge, use the install icon in the address bar or the browser’s app menu. In Safari on Mac, choose File → Add to Dock.

Markdown edition

# DeepL introduces Aure for expressive voice translation

> A new model aims to carry changes in delivery across languages.

By BIGCHANGE.AI Editorial

Published: 2026-11-12T07:43:41.010Z
Updated: 2026-11-12T07:43:41.010Z
Canonical: https://bigchange.ai/blog/deepl-aure-live-voice-translation

![Two people converse.](https://bigchange.ai/api/media/file/conversation-concept.png)
AI-generated conceptual illustration for BIG CHANGE

DeepL introduced Aure on November 11, aiming to make live translated speech follow changes in a speaker's delivery.

## The big change

- Aure uses incoming audio and translated words to adjust delivery continuously, DeepL says. Its expression and accuracy claims remain unverified here.
- Listening and reading have different meeting requirements. The desktop deployment guide says each translated-audio listener needs a license and app, while other participants can read captions without either.
- Ending a meeting does not necessarily erase its text everywhere. DeepL describes temporary service memory alongside participant-device storage, so organizers need to check retention settings as well as access.

## Voice identity and changing delivery

In its [September announcement](https://www.deepl.com/en/blog/deepl-voice-preserves-your-voice-across-languages), DeepL described voice preservation: translated speech would sound recognizable as the person speaking, instead of using the same synthetic voice for everyone. The company also claimed it would carry emphasis and urgency without retaining voice samples or creating profiles for later use.

Consider an illustrative conversation about a deadline. Hearing the words identifies the date; hearing a change in emphasis may affect how urgent the listener thinks it is. This explains why delivery matters alongside the words. It is not evidence that a synthetic rendition correctly identifies the speaker's intent.

The [November release](https://www.deepl.com/en/press-release/aure-voice-ai-model) contrasts Aure's continuous adaptation with an opening voice sample that fixed delivery earlier. It supplies no independent Aure evaluation. We did not test the model.

DeepL says Aure applies automatically to its meetings, in-person and Voice API products in English, German, Spanish, French, Italian, Japanese, Portuguese and Korean. Other languages retain existing models; no extra configuration is required.

## What meeting participants need

According to the [meeting guide](https://support.deepl.com/hc/en-us/articles/16810100955676-About-DeepL-Voice-for-Online-Meetings), a licensed user connects a supported Teams, Zoom or Google Meet call. The organization must allow the DeepL bot to join and enable chat, where it posts the translation-access link. The bot streams meeting audio for translation. Merely installing software is therefore not the same as allowing the meeting connection.

The [deployment guide](https://support.deepl.com/hc/en-us/articles/30158330516508-Deploy-the-DeepL-Voice-desktop-app-in-your-organization) distinguishes the translation manager from the audience: only the manager needs a license for live captions, which others can read on the website. Spoken translation requires each listener's own license and desktop app. Check which output attendees need before assuming one organizer's subscription covers the whole call.

The same guide says users are responsible for necessary consents and for informing everyone invited that audio will be transcribed, translated and output. This is the provider's guidance, not a finding about which law governs a particular meeting. The [Conversations guide](https://support.deepl.com/hc/en-us/articles/16813214348316-About-DeepL-Voice-for-Conversations) directs customers to sales for plan details and pricing.

## Service memory is not every copy

DeepL's meeting guide says recent transcription and translation messages remain temporarily in memory until the call ends. Text accessed by participants can persist on their devices for up to 30 days, under administrator-controlled retention. It also says customer data does not train DeepL's language models. These are provider descriptions, not an independent audit of a meeting.

Check local retention before assuming a call's end removes every copy. Retention choices also belong in the explanation given to participants before sensitive discussions.

For important instructions or decisions, confirm the substance directly. Fluent delivery alone cannot establish that every word and nuance was translated correctly.

## Sources

- [DeepL Aure announcement](https://www.deepl.com/en/press-release/aure-voice-ai-model) — Model and rollout claims.
- [DeepL voice preservation](https://www.deepl.com/en/blog/deepl-voice-preserves-your-voice-across-languages) — Earlier voice feature.
- [DeepL meetings guide](https://support.deepl.com/hc/en-us/articles/16810100955676-About-DeepL-Voice-for-Online-Meetings) — Access and data handling.
- [DeepL desktop deployment guide](https://support.deepl.com/hc/en-us/articles/30158330516508-Deploy-the-DeepL-Voice-desktop-app-in-your-organization) — Licensing and participant notice.
- [DeepL conversations guide](https://support.deepl.com/hc/en-us/articles/16813214348316-About-DeepL-Voice-for-Conversations) — Plans and pricing contact.