AI BriefingMicrosoft Launches MAI-Voice-2
·2026.06.05 06:38
Key point
Microsoft has launched MAI-Voice-2, a TTS model supporting 15 languages and voice cloning.
Details
Microsoft's latest high-performance TTS (Text-to-Speech) model, MAI-Voice-2, has been unveiled. This model supports the building of more natural and expressive voice agents.
Key Features:
- Voice Cloning: Enables sophisticated voice cloning through short voice samples.
- Emotion Control: Provides rich expressiveness through fine-grained emotional adjustment.
- Multilingual Support: Maintains a consistent voice identity across 15 languages.
Deployment and Pricing Information:
- Available immediately on Azure AI Foundry, with integration set to expand to VSCode, Dynamics 365 Contact Center, Teams, and more.
- Priced at $22 per 1 million characters, positioning it as an economical alternative to the OpenAI Realtime API.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.