Microsoft's New Voice AI Speaks Hindi in 6 Voices — What It Costs
MAI-Voice-2.1 ships six Hindi voices and runs from Microsoft's Central India region. Its sister model transcribes speech live in 60 languages for $0.54 an hour. Both are previews — here is what that means.
यह लेख हिंदी में पढ़ें
What Microsoft released
| Model | What it does | Price |
|---|---|---|
| MAI-Voice-2.1 | Text to speech, 23 languages and 26 locales, tuned for long narration | $22 per 1M characters |
| MAI-Voice-2.1-Flash | The same voices, built for low latency — Microsoft says 150 ms end to end | $15 per 1M characters |
| MAI-Transcribe-2-Streaming | Live speech to text in 60 languages, with automatic language detection | $0.54 per hour of audio, through the end of 2026 |
The models are available through Microsoft Foundry, the MAI Playground, Vercel and OpenRouter, with LiveKit listed as coming soon.
The Hindi voices
Microsoft's documentation lists six Hindi (hi-IN) voices — Arjun, Dhruv, Grant, Harper, Kavya and Priya — plus two Indian-English (en-IN) voices, Dhruv and Priya. Four of the Hindi voices support a wide range of emotional styles (happy, sad, excited, whispering and more); Harper is set up for agent, call-centre, educational and narration styles, and Grant for audiobook and neutral delivery.
Hindi is the only Indian language with voices in the list as of the documentation's 1 October update. There are no Tamil, Telugu, Bengali or Marathi voices yet.
What it costs in practice
Text-to-speech is billed per character, so the arithmetic is simple. At the published rates, 1,000 characters cost $0.022 on MAI-Voice-2.1 and $0.015 on Flash. Live transcription at $0.54 an hour means a 30-minute call costs $0.27 to transcribe.
Microsoft says the streaming transcriber returns its first guesses "in just over 100ms", ranks first for accuracy on Artificial Analysis for both final and partial transcripts, and puts words on screen twice as fast as its closest competitor. Those are Microsoft's claims about its own model.
Read this before you build on it
- It is a preview. Microsoft's documentation says the voices are in public preview, "without a service-level agreement, and is not recommended for production workloads".
- Hinglish is not claimed. Microsoft confirms Hindi is supported for transcription. Its pages do not say the models handle code-switching — Hindi and English mixed in one sentence — so test that with your own audio before relying on it.
- Voice cloning is gated. Cloning a voice from a 5–60 second clip requires applying for limited access and uploading the speaker's consent. Microsoft says no unlicensed voice cloning is possible.
- Data location. Central India is one of the regions Microsoft serves these models from, which matters if your users' data should stay in the country.
If you need a Hindi voice for a production product today, compare it against the tools on our best AI voice tools for Indian languages page, which have been generally available for longer.
Sources
- Microsoft AI: Our first streaming transcription model (announcement and prices, 1 October 2026)
- Microsoft Learn: MAI-Voice-2.1 and MAI-Voice-2.1-Flash (voice list, regions, preview status, cloning)
- Microsoft AI: MAI-Transcribe-2 model page (supported languages, including Hindi)
Frequently asked questions
Does MAI-Voice-2.1 support Hindi?
Yes. Microsoft's documentation lists six Hindi voices — Arjun, Dhruv, Grant, Harper, Kavya and Priya — and two Indian-English voices. Hindi is the only Indian language with voices as of the 1 October 2026 update.
How much does MAI-Voice-2.1 cost?
$22 per million characters for MAI-Voice-2.1 and $15 per million for MAI-Voice-2.1-Flash. That is $0.022 and $0.015 per 1,000 characters. MAI-Transcribe-2-Streaming is $0.54 per hour of audio through the end of 2026.
Can I use MAI-Voice-2.1 in production?
Microsoft lists it as a public preview without a service-level agreement and says it is not recommended for production workloads. It is fine for prototypes and testing.
Save this summary as an image or share it.
AICreatorHub Team
The AICreatorHub editorial team is a group of hands-on AI practitioners, writers and developers based in India. We test AI tools and models ourselves, track official releases from OpenAI, Anthropic, Google, Meta and xAI, and translate them into simple, India-first guides in English and Hindi. Every article is written for real Indian use cases — pricing in rupees, free-tier tips and practical, tested steps — so you get accurate, up-to-date and genuinely useful AI information.