Post-Cutoff
Search For AI agents

MicrosoftMAI-Voice familyPreviewSpeech and audioReleased 2 Jun 2026

MAI-Voice-2 / MAI-Voice-2-Flash

Microsoft’s in-house expressive TTS, used through standard Azure Speech SSML.

Model iden-US-Harper:MAI-Voice-2

Price$22 per 1M characters

Checked29 September 2026

Where it stands

54newer speech and audio models since its release

The 5 newest are listed. All speech and audio models

Before it from Microsoft
VibeVoice (ASR, ASR-Streaming, ASR-BitNet, Realtime-0.5B TTS)25 Aug 2025
After it from Microsoft
MAI-Transcribe-23 Sep 2026
Weights
Closed weights; access is hosted only.

How to call it

6 ways to use MAI-Voice-2 / MAI-Voice-2-Flash, with the model id each provider expects.

Providers, model ids, endpoints and docs for MAI-Voice-2 / MAI-Voice-2-Flash
ProviderModel idEndpointDocs
Azure Speech in Microsoft Foundry (SSML voice name) en-US-Harper:MAI-Voice-2 https://{region}.tts.speech.microsoft.com/cognitiveservices/v1 learn.microsoft.com
Azure Speech in Microsoft Foundry (SSML voice name) en-US-Harper:MAI-Voice-2-Flash learn.microsoft.com
Azure Voice Live (TTS output) MAI-Voice-2-Flash learn.microsoft.com
OpenRouter microsoft/mai-voice-2 https://openrouter.ai/api/v1/audio/speech
OpenRouter microsoft/mai-voice-2-flash
Web app (MAI Playground) playground.microsoft.ai

Pricing

Prices for MAI-Voice-2 / MAI-Voice-2-Flash
Per 1M characters$22
Per 1M characters, Flash$15

As published: “USD per 1M characters (MAI-Voice-2 / MAI-Voice-2-Flash, ‘starting at’)”.

Source: microsoft.ai, checked 29 September 2026.

Specs

Input
Text
Output
Audio
Context window
Not published
Max output
Not published
Open weights
No
Licence
Proprietary
Released
2 June 2026
Status
Preview
Type
Speech and audio

What stands out

What its maker marketed at launch, and what people found later, each with its source.

  • Gated instant voice cloning

    Matches a consented reference voice from a 5-60 s clip without training; only approved (Limited Access) licensed voices can be synthesized.

    Source: learn.microsoft.com

  • SSML emotion/style control

    mstts:express-as styles (angry, fearful, joyful, whispering, shouting, etc.) with styledegree, across 15 languages / 18 locales.

    Source: learn.microsoft.com

  • Low-latency Flash tier

    Found after launch

    MAI-Voice-2-Flash (public preview from 2026-07-23) targets voice agents/IVR; Microsoft quotes ~225 ms latency vs ~1 s for MAI-Voice-2 (for a 45 s clip).

    Source: microsoft.ai

Notes

Launched at Build 2026-06-02 (MAI-Voice-2); Flash followed 2026-07-23 (date per secondary sources). Both public preview in Azure Speech. Languages include en-US/AU, de, fr, es-ES/MX, pt-BR/PT, it, ko, zh-CN, tr, ru, th, nl, ro, hu, hi. Also used in Copilot (Audio Expressions). Predecessor MAI-Voice-1 no longer listed on the MAI-Voice docs page.

curl -X POST "https://$REGION.tts.speech.microsoft.com/cognitiveservices/v1" \
  -H "Ocp-Apim-Subscription-Key: $SPEECH_KEY" -H "Content-Type: application/ssml+xml" \
  -H "X-Microsoft-OutputFormat: audio-24khz-160kbitrate-mono-mp3" \
  --data '<speak version="1.0" xml:lang="en-US"><voice name="en-US-Harper:MAI-Voice-2">Hello from MAI Voice.</voice></speak>' -o out.mp3

Sources: https://learn.microsoft.com/en-us/azure/ai-services/speech-service/mai-voices · https://microsoft.ai/models/mai-voice-2/ · https://microsoft.ai/news/building-a-hillclimbing-machine-launching-seven-new-mai-models/

Not yet checked against an official source: Also on Fireworks and Baseten (ids not verified).