Skip to content
Sarvam AI (STT)

Sarvam speech-to-text for Bolna AI voice agents

Sarvam is a speech-to-text provider built for Indian languages that transcribes your callers in real time, including Hinglish and other code-mixed speech, using its Saaras models. The docs call it usually the strongest transcriber choice for agents serving Indian callers.

How Sarvam AI fits in the stack.

  • Best for

    Turning the caller's speech into text in real time, in their language.

  • Use this layer when

    You want to tune accuracy, latency, languages or vocabulary for your callers.

  • Connects to

    The phone call upstream, and the LLM that reads the text downstream.

The transcriber is the
listening layer.

Sarvam AI turns the caller's speech into text as they talk, so the LLM can understand it and reply.

  1. Telephony

    Phone network

    • Exotel
    • Plivo
    • Twilio
  2. STT · You are here

    Listener

    • Sarvam AI (STT)
  3. LLM

    Reasoning

    • OpenAI
    • Anthropic
    • Google Gemini
  4. TTS

    Voice

    • ElevenLabs
    • Sarvam AI (TTS)
    • Cartesia
  5. Tools

    Actions

    • Cal.com
    • n8n
    • AiSensy

How do I set up Sarvam with Bolna?

Which Sarvam models and languages does Bolna support?

ModelLanguages on BolnaNotes
saaras:v4English (India), Hindi, Bengali, Tamil, Telugu, Gujarati, Kannada, Malayalam, Marathi, Punjabi, Odia, AssameseRecommended. Automatic language detection, global English accents, reads keywords and context
saaras:v3Same 12 languagesSupported. Reads context, not keywords
saaras:v2.5, saarika:v2.5Not listedDeprecated by Sarvam. Existing agents keep working; move them to saaras:v4

For language codes and biasing limits, see the Sarvam setup guide in the Bolna docs.

Sarvam's other 11 Saaras languages (Urdu, Nepali, Konkani, Kashmiri, Sindhi, Sanskrit, Santali, Manipuri, Bodo, Maithili and Dogri) are not yet available through Bolna.

What can you build with Sarvam and Bolna?

  • Hinglish sales and support

    Callers who switch between English and Hindi mid-sentence are transcribed without the agent reconfiguring its language. See Hindi speech to text for the other Hindi transcribers on Bolna.

  • All-India customer support

    Give each language its own transcriber in one multilingual agent, with Sarvam on the Indian languages.

  • A fully Indic voice pipeline

    Pair Sarvam STT with Sarvam text-to-speech so both listening and speaking are tuned for Indian languages.

  • Order and account lookups

    Pass brand names and a description of the call so terms like PhonePe or order numbers are recognised. On Hindi agents, boosted English terms can come back in Devanagari script.

Other providers in this layer.

See all 10 speech-to-text providers
  • Smallest AI (Pulse)

    Speech-to-text

    Smallest AI's Pulse model transcribes callers for your Bolna voice agent with sub-100ms first-token latency, keyword weights and speaker labels.

  • Soniox

    Speech-to-text

    Soniox stt-rt-v5 transcribes callers for your Bolna voice agent with one multilingual model that auto-detects and code-switches, including Hinglish.

  • AssemblyAI

    Speech-to-text

    AssemblyAI Universal-3.6 Pro transcribes callers for your Bolna voice agent in 32 languages, with code-switching and keyword plus context biasing.

Keep exploring the voice stack.

See where Sarvam AI fits in your workflow.

Walk through provider choice, stack trade-offs and the exact workflow you want Bolna to automate.

Ready for enterprise.

Secure, compliant and dependable, from the first pilot to full scale.

Contact Sales
  • 99.9% uptime

    Built for your busiest hour. Even if the model fails, the call doesn’t

  • Support and SLAs

    A dedicated team, with service levels shaped around your business

  • Security & compliance

    Data residency, encryption, access controls, full audit trails and more

  • Deploy in your cloud

    Run Bolna in your own cloud or on-premise, your data remains yours

Certified by

  • ISO 27001
  • DPDP compliant

Frequently Asked Questions

Twelve today. English (India), Hindi, Bengali, Tamil, Telugu, Gujarati, Kannada, Malayalam, Marathi, Punjabi, Odia and Assamese, on both saaras:v4 and saaras:v3.