Audio
156d agoMistral Launches Voxtral, 3‑Second Voice Clone
Mistral’s new Voxtral model lets users clone a voice from just three seconds of audio in nine languages, pushing text‑to‑speech toward near‑real‑time personalization. The open‑weight release invites developers worldwide to build multilingual voice assistants, audiobooks, and accessibility tools without costly proprietary licenses.
The Signal
It signals a shift toward democratized, low‑latency voice synthesis.