Voxtral TTS is the text-to-speech engine developed by Mistral AI, accessible via the Platform console and the Audio Speech API. It generates natural voices from text in French, English and several other languages, with production-quality for voice-overs, conversational agents and real-time applications. Paired with Voxtral for transcription, it forms a coherent audio stack, hosted in Europe and compliant with the GDPR requirements of data-conscious enterprises.
What is Voxtral TTS?
The essentials
Voxtral TTS is Mistral AI’s text-to-speech offering, integrated into the Platform console and the Audio Speech API. It targets developers and product teams who want to integrate synthetic voice into their applications while staying within a European framework. The solution is part of a broader audio strategy: alongside Voxtral for transcription, chat models, agents and workflows in beta, it completes the Mistral ecosystem for building coherent voice experiences.
Key features
Voxtral TTS is used primarily via the Audio Speech API, which generates a voice from text according to several parameters (language, speed, selected voice). Integration into the Platform console makes testing the model simple: a text editor, a voice selector and a button to listen to the result. Audio quality is refined, with natural rendering in French and English and growing support for other European languages. The synergy with Voxtral for audio transcription makes it possible to build bidirectional use cases: transcribe a call to summarize it, then generate a voice response. Beta features of the Platform (Agents, Workflows, Observability) facilitate the construction of complete voice agents, capable of understanding, reasoning and responding. Pay-as-you-go pricing simplifies experimentation, with no minimum commitment, which aligns with the culture of most technical teams.
Use cases
Audio studios and podcasts use Voxtral TTS to produce quality French voice-overs without relying on a physical studio. Software publishers integrate voice synthesis into their applications to make content accessible (automatic playback, accessibility for visually impaired). Support teams build voice agents capable of responding 24/7 by combining Voxtral TTS with a Mistral LLM and agent logic. Public services and regulated actors in Europe find in Voxtral TTS a way to address their audio needs without transferring data outside the EU. Media use it to quickly generate audio versions of their written articles.
Advantages
The first benefit of Voxtral TTS is sovereignty: hosting voice data in Europe addresses a critical challenge for administrations, banks, insurers and regulated actors. The second is integration: for teams already Mistral customers, adding voice to the stack happens without changing providers. The third is audio quality in French, which rivals Anglo-Saxon leaders. The fourth is pricing flexibility: pay-as-you-go, with no commitment, which reduces experimentation risk.
Pricing
Voxtral TTS follows the pay-as-you-go logic of the Mistral API: no subscription, payment by consumption. Cost varies based on the volume of audio characters generated and the voice used. Mistral offers free credits to get started and the Platform console allows you to monitor consumption in real time. Large volumes can negotiate custom terms via enterprise contact.
Conclusion
Voxtral TTS marks Mistral’s entry into the voice synthesis market, with a central argument: European sovereignty combined with deep integration into the Platform ecosystem. For technical teams that want to build voice agents, audio content or accessible applications while respecting compliance constraints, it’s one of the most relevant options on the market in 2026.