Voxtral 3 TTS

Voxtral 3 TTS, also referred to as Voxtral Speech 3 or Voxtral Mini TTS 3, is a text-to-speech model introduced by Mistral AI as part of the Voxtral 3 audio suite. Designed for real-time voice agents and natural speech synthesis, the model delivers expressive speech generation across numerous languages, dialects, and emotional ranges with low latency.

The architecture supports advanced capabilities such as zero-shot voice cloning, multi-format streaming output, and fine-grained control over speech pace and prosody. It integrates smoothly into conversational pipelines alongside speech-to-text transcription models and reasoning engines to power responsive voice-based applications.

Voxtral 3 TTS is available through Mistral's managed cloud APIs, including EU-hosted endpoints, or for deployment on private infrastructure, enabling organizations to maintain data sovereignty and control over audio processing workflows.