Mistral releases a new open-source model for speech generation

French AI company Mistral launched an open-source text-to-speech model called Voxtral TTS, aimed at voice AI applications in customer support and sales. Supporting nine languages, it adapts custom voices with minimal samples, delivering real-time performance and state-of-the-art features at a low cost. This move intensifies competition with firms like ElevenLabs and OpenAI.

Key Points

• Mistral released Voxtral TTS, an open-source text-to-speech model.
• The model enables enterprises to create voice agents for customer support and sales.
• It supports nine languages, allowing customization with samples under five seconds.
• Features include subtle accent capture and quick response time (90ms TTFA).
• The RTF is 6x, indicating efficient real-time audio rendering.
• Mistral aims to create a comprehensive voice solution for enterprises.

Relevance

• The increase in demand for natural-sounding voice AI solutions aligns with the 2025 trend of advanced speech technologies.
• Mistral's open-source approach taps into the growing preference for customizable enterprise AI solutions.
• Competition within the text-to-speech space mirrors the broader AI landscape where open-source tools are gaining traction.

Mistral's Voxtral TTS positions the company as a strong competitor in the voice AI market, catering to enterprises with advanced features and a focus on customization that may reshape user engagement and customer support strategies.