DeepL Unveils Aure Voice Model for Translation

The model carries a speaker's tone, emotion and pace into real-time speech translation across eight languages at launch.

Updated on Nov. 11, 2026 in Language Learning

DeepL Unveils Aure Voice Model for Translation

DeepL has unveiled Aure, a voice AI model that carries a speaker's tone, emotion, emphasis and pace into real-time speech translation. Aure powers DeepL Voice and is DeepL's most expressive voice model to date.

Why it matters

The earlier system could not reflect shifts in mood or delivery that contradict the words spoken, such as sarcasm, so Aure aims to preserve more of the meaning carried by how something is said. Translated speech that sounds more natural could make cross-language conversations easier to follow.

Aure supports eight languages at launch: English, German, Spanish, French, Italian, Japanese, Portuguese and Korean, with all other languages still using DeepL's existing voice models. In blind tests, 96 percent of linguists preferred DeepL Voice, though the timeline for expanding beyond the initial languages has not been disclosed.

The players

DeepL

DeepL is a language technology company headquartered in Cologne, Germany, known for its machine translation products.

The details

Aure isolates a speaker's vocal identity and prosody while filtering out background noise, echoes and distortion. It analyzes incoming speech continuously using both the source audio and the words already translated, and is applied automatically in DeepL Voice for online meetings, in-person conversations and the DeepL API for voice.

Timeline

  1. September: DeepL introduced a feature that translates users in their own recognizable voice.

  2. April: DeepL launched voice-to-voice translation.

  3. May: DeepL announced about 250 job cuts, roughly a quarter of its workforce.

Market Landscape

Aure builds on DeepL Voice's earlier system, which determined translated voice delivery only from the first moments of a conversation. The update positions DeepL to compete in real-time speech translation by focusing on expressive delivery rather than only word accuracy.

Users of DeepL Voice for online meetings, in-person conversations and the DeepL API get Aure automatically with no extra setup. Conversations in the eight supported languages may sound more natural, while other languages continue on existing voice models.

The takeaway

Real-time translation is moving beyond words to capture how something is said, which can help conversations feel more natural across languages. DeepL Voice does not permanently store transcripts or translations and deletes them from memory once a session ends.

Further reading

Explore more on Language Learning to see how translation tools are reshaping how people communicate across languages.

Source note: This article includes information reported by Trending Topics.