DeepL Aure Learns Voices Mid-Conversation

🔥 What happened DeepL unveiled Aure, a new voice AI model that translates speech in real time while mirroring the speaker's emphasis, pace, and emotion. Unlike earlier models that analyzed a single static voice sample, Aure continuously reads the audio source throughout a conversation. It's live now in DeepL Voice and via API, starting with eight languages. 💡 Why it matters Previous voice models froze a voice at the start of a call, so mood shifts got lost in translation. Aure decouples voice identity, prosody, and content, and keeps learning mid-dialogue. That's the gap between a robotic announcer and an interpreter who actually feels the room. ⚡ Our take DeepL is now going head-to-head with Google and OpenAI's real-time interpreters, betting on something text models can't match: emotional continuity. Whether it holds up in noisy, real-world meetings is the open question.