DeepL unveils Aure, a highly realistic voice AI translation model capable of capturing speaker nuance

DeepL has launched Aure, its latest voice AI translation model, which the company describes as its most dynamic and lifelike Voice AI to date. This new model is designed to go beyond literal translation by preserving the unique characteristics, subtle emotions, and expressive qualities inherent in human conversation, such as tone variation, emotional shifts, emphasis, and speaking pace. Aure builds upon DeepL's existing voice preservation technology, which allows users to communicate across languages using their recognizable original voice timbre, moving past monotonous synthetic sounds. While the model is currently unavailable for Chinese, it is integrated across DeepL Voice's web conferencing, face-to-face conversation modes, and its voice API. The initial language support for Aure includes English, German, Spanish, French, Italian, Japanese, Portuguese, and Korean, with plans for future expansion. For any language not yet supported by Aure, DeepL Voice will automatically revert to its previous voice model.