Mistral launches Voxtral TTS open source multilingual text-to-speech model

Mistral AI unveiled its first text-to-speech model as an expansion of the Voxtral family. The Paris-based startup released the 4 billion parameter system with open weights on Thursday. The model supports nine languages including English, French, German, Spanish, Dutch, Portuguese, Italian, Hindi and Arabic. The system targets enterprise deployment in voice assistants, customer support and sales engagement tools. Organizations can run it on their own infrastructure rather than relying on third-party APIs. Mistral said the model is lightweight enough to operate on consumer hardware including laptops, smartphones and edge devices while maintaining frontier-quality performance. The model replicates a speaker's voice from a few seconds of reference audio, capturing tone, accent, intonation and emotion. It also performs cross-language voice control, such as generating English speech with a French accent from a short prompt. Mistral said in human evaluations the model matched or outperformed competing systems in naturalness. It exceeded models from ElevenLabs that have lower latency while achieving parity with more advanced offerings in lifelike interaction. The company wrote in a blog post that the model excels at both contextual understanding and speaker modeling. Voxtral TTS provides full control and customization for enterprises looking to own their voice AI stack due to its compact size, low cost and latency and easy adaptability. The launch builds on the company's earlier release of speech-to-text models.