Sarvam + LiveKit

Voice Infra

Real-time voice infrastructure for building production-grade voice AI agents. Telephony, SIP, and global scaling handled automatically.

Sarvam APIs Supported

saaras:v3

Speech-to-Text

Accurate transcription built for Indian accents, dialects, and real-world audio with speaker diarisation.

bulbul:v3

Text-to-Speech

Natural, expressive voices across 11+ Indian languages for conversational AI, dubbing, and accessibility.

sarvam-105b

LLM

Multilingual large language model with native proficiency across Indian languages for text generation and reasoning.

How it works

Install LiveKit Agents SDK

pip install "livekit-agents[sarvam,silero]" python-dotenv

Add Sarvam as provider

Create a .env file with LIVEKIT_URL, LIVEKIT_API_KEY, LIVEKIT_API_SECRET, and SARVAM_API_KEY

Deploy your agent

Run locally with the dev or console command, or deploy to LiveKit Cloud

Quick start

Install

Bash

pip install "livekit-agents[sarvam,silero]" python-dotenv

Code

Python

from livekit.agents import AgentSession
from livekit.plugins import sarvam

session = AgentSession(
    stt=sarvam.STT(
        language="unknown",
        model="saaras:v3",
        mode="transcribe",
        flush_signal=True,
    ),
    llm=sarvam.LLM(model="sarvam-105b"),
    tts=sarvam.TTS(
        target_language_code="en-IN",
        model="bulbul:v3",
        speaker="shubh",
    ),
    turn_detection="stt",
)

Run

Bash

python agent.py dev

Starts the agent in development mode and connects it to your LiveKit room. Use `python agent.py console` to test the agent locally in your terminal without a LiveKit Cloud project.

Use cases