Skip to content
CallMissed
Book a demo

Explore CallMissed

Back to all models
Text to Speechindian-languages
Sarvam AI — bulbul:v4-flash logo

Bulbul V4 Flash

by Sarvam AI · Released 2026

Sarvam AI's fast text-to-speech model. 222 persona voices across 9 Indian languages plus Hinglish, at a lower price than Bulbul v3. Speed control from 0.5 to 2.0 and 8 to 24 kHz output.

Sarvam AI — bulbul:v4-flash logo
Text to Speech

Bulbul V4 Flash

Powered by Sarvam AI · Proprietary TTS model

Context Window

N/A

Parameters

Undisclosed

Max Output

2,500 characters per request

Category

Text to Speech

Overview

Bulbul V4 Flash is Sarvam AI's fast, low-cost text-to-speech model. It offers 222 persona voices, each named `<name>_<language>_<style>` such as `simran_en_customer` or `aparna_hi_customer`, so a voice already carries its language and delivery style: conversational, customer-support, ads, narration and more.

It covers nine Indian language codes (hi-IN, en-IN, bn-IN, gu-IN, kn-IN, mr-IN, pa-IN, ta-IN, te-IN) plus code-mixed Hinglish personas. Malayalam (ml-IN) and Odia (od-IN) are not supported on this model; use Bulbul v3 for those. The default voice is `shubh_enhi_ads`, and Bulbul v3 short names such as `shubh` or `priya` are not valid here.

Each request takes up to 2,500 characters. `speed` runs from 0.5 to 2.0 and `temperature` from 0.01 to 1.0. Output sample rates are 8000, 16000, 22050 and 24000 Hz, with 22050 Hz as the default; formats match Bulbul v3. It works in live voice-agent calls as well as the REST endpoint.

Pricing

MetricPrice
Price /10K chars₹20.0000

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

Key Highlights

  • 222 persona voices across 9 Indian languages plus Hinglish
  • Lower price than Bulbul v3
  • Speed control from 0.5 to 2.0
  • Works in live voice-agent calls

Benchmarks

BenchmarkScore
Voices222
Languages9
Max input2,500

Technical Details

  • 222 persona voices; default voice shubh_enhi_ads
  • Language codes: hi-IN, en-IN, bn-IN, gu-IN, kn-IN, mr-IN, pa-IN, ta-IN, te-IN, plus Hinglish personas
  • No Malayalam (ml-IN) or Odia (od-IN): use Bulbul v3 for those
  • Speed 0.5 to 2.0; temperature 0.01 to 1.0 on /v1/audio/speech
  • Sample rates 8000, 16000, 22050, 24000 Hz (default 22050)
  • Up to 2,500 characters per request

Strengths

  • 222 persona voices, each tuned to a language and style
  • Lower price per character than Bulbul v3
  • Works in live voice-agent calls
  • Code-mixed Hinglish personas

Limitations

  • No Malayalam or Odia: use Bulbul v3 for those languages
  • Up to 2,500 characters per request
  • Output sample rates stop at 24 kHz
  • Voice names are not interchangeable with Bulbul v3 short names

Use Cases

Voice agentsIVR systemsCustomer-support callsAd and promo voiceovers

API Example

curl https://api.callmissed.com/v1/audio/speech \
  -H "Authorization: Bearer cm_YOUR_KEY" \
  -d '{"model": "bulbul:v4-flash", "input": "Namaste, aapka order confirm ho gaya hai.", "voice": "shubh_enhi_ads"}' \
  --output speech.mp3

Endpoint: POST /v1/audio/speech · Model ID: bulbul:v4-flash

Try Bulbul V4 Flash now

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

Start free Book a DemoRead docs