Microsoft MAI-Transcribe-2-Streaming + MAI-Voice-2.1
Microsoft launched three in-house voice models: MAI-Transcribe-2-Streaming (real-time transcription, 60 languages, first hypotheses in just over 100ms) and MAI-Voice-2.1 / 2.1-Flash text-to-speech, available via Microsoft Foundry.
Why it matters
Gives Microsoft a priced, first-party stack for both sides of a voice-agent conversation — streaming ASR plus expressive TTS — competing directly with Deepgram and ElevenLabs.
What you could build with it
A contact-center SaaS team could build a real-time agent-assist product that transcribes customer calls as they happen and surfaces suggested replies mid-sentence, using the streaming partials to cut agent handle time.
Does it hold up?
Too early to judge — launched Oct 1; Microsoft claims #1 on Artificial Analysis for streaming transcription accuracy, but no independent production deployments have reported yet.
Built with Microsoft MAI-Transcribe-2-Streaming +…
- SiliconANGLE: Microsoft targets ultra-realistic voice agentsSiliconANGLE · Independent coverage of the three-model launch, pricing ($0.54/audio hour) and the 320ms partials claim.
- PYMNTS: Microsoft targets contact center market with real-time AI voice modelsPYMNTS · Covers the contact-center angle and quotes Microsoft Foundry's Naomi Moneypenny on acting mid-sentence.
- Unite.AI: Microsoft launches MAI-Transcribe-2-Streaming and two MAI-Voice modelsUnite.AI · Third writeup confirming availability through Microsoft Foundry on launch day.
Learn more
First spotted on article: source.
More AI models
ElevenLabs Eleven v4 + v4 Turbo: new emotive TTS architecture with a 100ms real-time variantElevenLabs released Eleven v4 and Eleven v4 Turbo on September 28: a new text-to-speech architecture with inline…models · JEV 0.72VoiceStudio: open-source, fully-local ElevenLabs alternative with voice cloning, dubbing and dictation in 646 languagesVoiceStudio is an open-source, fully-local voice AI studio: voice cloning from a 3-second sample, voice design, video…models · JEV 0.71DeepSeek V4-Flash official API: public beta with upgraded agent capabilities, Responses API and Codex supportDeepSeek's official V4-Flash is now a public beta API with massively upgraded agent capabilities - benchmark scores…models · JEV 0.69Mistral Forge: enterprise platform for training and continuously improving proprietary modelsMistral AI launched Forge (Sep 28) — a full-lifecycle model training platform (pre-training, SFT, DPO/ODPO, RL…models · JEV 0.69kev: open, trainable Jev-like family of small decision models on Qwen3.5/3.8A family of small decision models (0.8B to 27B) built on Qwen3.5/3.8 that reproduces Jev's prefill-only typed-decision…models · JEV 0.68GPT-Synopsys: OpenAI and Synopsys multi-year deal to build an AI model that operates EDA toolsOpenAI and Synopsys signed a multi-year agreement on Sep 30 to jointly develop GPT-Synopsys, a specialized model that…models · JEV 0.66
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.