Microsoft MAI voice models arrive in LiveKit for TTS agents

Microsoft AI released an official text-to-speech plugin for LiveKit Agents, giving developers access to expressive MAI voices including MAI-Voice-2-Flash. The documented example uses the en-US-Harper voice at 24 kHz and requires an Azure Speech resource key and region.

• The microsoft_ai.TTS plugin ships in the livekit-agents[microsoft-ai]~=1.8 package
• The example uses the MAI-Voice-2-Flash model with the en-US-Harper voice at 24 kHz
• Deployment requires MICROSOFT_AI_TTS_API_KEY and MICROSOFT_AI_TTS_REGION variables
• Pricing, concurrency limits and language coverage for MAI via LiveKit are not disclosed