Mistral Releases Voxtral TTS: Open-Source Text-to-Speech That Runs on a Smartwatch, Supports 9 Languages

TechCrunch·high signal

Mistral launched Voxtral TTS on March 26, an open-source text-to-speech model based on Ministral 3B that achieves state-of-the-art performance while fitting on a smartwatch, smartphone, or laptop. The model clones any voice from a sub-5-second sample, handles seamless multilingual switching without losing voice characteristics, and delivers 90ms time-to-first-audio with a 6x real-time factor (10-second clip in ~1.6 seconds). Supports English, French, German, Spanish, Dutch, Portuguese, Italian, Hindi, and Arabic.