Skip to content
CallMissed
Book a demo

Explore CallMissed

Back to all models
Text to Speechpreviewvoice-agentsmultilinguallow-latency

MAI-Voice-2.1-Flash

by Microsoft · Released Preview

Preview. Low-latency Microsoft AI voice model built for real-time voice agents, the recommended pick for live calls. 97 voices across 23 languages including Hindi and Indian English. Plain text, adjustable speed. $0.1563 per 10K characters.

Text to Speech

MAI-Voice-2.1-Flash

Powered by Microsoft · Neural TTS (Microsoft AI)

Context Window

N/A

Parameters

Undisclosed

Max Output

4096 characters per request

Category

Text to Speech

Overview

MAI-Voice-2.1-Flash is a preview text-to-speech model from Microsoft AI, built for low latency and real-time voice agents, and the recommended pick for live calls. It works on POST /v1/audio/speech and inside voice agents and voice sessions.

There are 97 voices across 23 languages. Voice handles are lowercase locale-name pairs such as hi-in-kavya, and a voice speaks only its own locale. Hindi voices are hi-in-arjun, hi-in-dhruv, hi-in-grant, hi-in-harper, hi-in-kavya and hi-in-priya; Indian English voices are en-in-dhruv and en-in-priya. With no voice set, the first voice for the request language is used (hi-in-arjun for Hindi), otherwise en-in-priya.

It takes plain text only, with no SSML or tags. Preview: Microsoft ships this model without an SLA and may change or retire it at short notice. If it is retired, the ID keeps working and is answered by a named successor, so an agent that uses it may sound different. Use a generally available model for production-critical agents. It is not processed in India, is paid-plan only and is never a default.

Pricing

MetricPrice
Price /10K chars₹15.0000

1 credit = ₹1 ≈ US$0.0104 (US$1 = ₹96). Transparent per-call pricing — you pay only for what you call, with no per-seat charge or minimums.

Key Highlights

  • Preview model from Microsoft AI
  • 97 voices across 23 languages, including Hindi and Indian English
  • Low latency — the recommended pick for live calls
  • Available on the speech endpoint and in voice agents

Benchmarks

BenchmarkScore
Voices97
Languages23
Price$0.1563
StatusPreview

Technical Details

  • Plain text only — no SSML or tags
  • Speed 0.5 to 2.0 (speed on the speech endpoint, tts_speed in voice agents)
  • Formats: mp3 (default, 24 kHz), wav, pcm, opus, webm, mulaw, alaw — flac and aac return 400
  • Up to 4096 characters per request
  • Not processed in India — agents hosted in India cannot use it
  • Cannot be used in tts_fallbacks

Strengths

  • 97 voices across 23 languages
  • Hindi and Indian English voices
  • Low latency for live conversation

Limitations

  • Preview — no SLA; may change or be retired at short notice
  • Plain text only — no SSML or tags
  • Not processed in India
  • Paid plans only; not usable as a fallback voice

Use Cases

Real-time voice agents and live callsHindi and Indian English voice experiencesMultilingual customer-facing audio

API Example

curl https://api.callmissed.com/v1/audio/speech \
  -H "Authorization: Bearer cm_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "mai-voice-2.1-flash",
    "voice": "hi-in-kavya",
    "input": "Namaste, aapka order aaj shaam tak pahunch jayega."
  }' \
  --output speech.mp3

Endpoint: POST /v1/audio/speech · Model ID: mai-voice-2.1-flash

Try MAI-Voice-2.1-Flash now

Get 1000 free API credits on signup. No credit card required.

Start free Book a DemoRead docs