Osegi
AI·1 min read

DeepL unveils Aure, an AI model for expressive synthetic voices

DeepL has introduced Aure, a new AI model designed to deliver human-like translations with expressive, generated voices that convey emotion.

Illustration of a sound wave merging with translated text, representing DeepL's expressive AI voice model Aure.
Image: heise online

DeepL has presented a new AI model called Aure. According to the company, it is meant to enable translations that sound more human, using generated voices that carry expression and emotion rather than a flat, robotic delivery.

The announcement marks a step beyond text. DeepL is best known for written translation, and Aure points to the spoken word: translated content that is not only accurate but also sounds natural and conveys the tone of the original speaker.

Why it matters

Synthetic voices have long been criticised for sounding monotone, which limits their use in settings where tone matters, such as meetings, customer conversations or media content. A model that reproduces emotion could make machine-translated speech more usable in those situations.

The company says Aure relies on specific techniques to achieve its expressive output. The details of those methods were not available in the material we reviewed, so it remains to be seen how the model works in practice and how it compares with competing voice systems.

What comes next

Open questions include which languages Aure will support, how and when it will be offered to customers, and what it will cost. Independent tests of voice quality will show whether the emotional range lives up to the promise.

Apps worth a look

SUMO: the number puzzle

Tap the right tiles, add them up, hit the target. Easy to learn, hard to master. Free, for all ages.

Tags:deeplaureai voicetranslationspeech synthesis

More from this day

See all →

More in: AI

See all →