Parlando di Voxtral

Mistral AI has launched Voxtral TTS, a 4B parameter text-to-speech model for generating realistic, multilingual voice audio. The model supports nine languages, emphasizing low latency and the ability to capture emotional expression and natural speech patterns for use in voice agents. It features zero-shot cross-lingual voice adaptation and can be easily customized with just a few seconds of a reference voice prompt. Built on a transformer and flow-matching architecture, Voxtral TTS is designed for scalable enterprise applications and is available via API.