Sarvam AI STT

The Sarvam AI STT provider enables your agent to use Sarvam AI's speech-to-text models for transcription. This provider uses Voice Activity Detection (VAD) to send audio chunks for transcription after a period of silence.

Installation

Install the Sarvam AI-enabled VideoSDK Agents package:

pip install "videosdk-plugins-sarvamai"

Importing

from videosdk.plugins.sarvamai import SarvamAISTT

Example Usage

from videosdk.plugins.sarvamai import SarvamAISTT
from videosdk.agents import CascadingPipeline

# Initialize the Sarvam AI STT model
stt = SarvamAISTT(
    # When SARVAMAI_API_KEY is set in .env - DON'T pass api_key parameter
    api_key="your-sarvam-ai-api-key",
    model="saarika:v2",
    language="en-IN"
)

#  Add stt to cascading pipeline
pipeline = CascadingPipeline(stt=stt)

note

When using .env file for credentials, don't pass them as arguments to model instances or context objects. The SDK automatically reads environment variables, so omit api_key and other credential parameters from your code.

Configuration Options

• api_key: (str) Your Sarvam AI API key. Can also be set via the SARVAMAI_API_KEY environment variable.
• model: (str) The Sarvam AI model to use (default: "saarika:v2").
• language: (str) Language code for transcription (default: "en-IN").
• input_sample_rate: (int) The sample rate of the audio from the source in Hz (default: 48000).
• output_sample_rate: (int) The sample rate to which the audio is resampled before sending for transcription (default: 16000).
• silence_threshold: (float) The normalized amplitude threshold for silence detection (default: 0.01).
• silence_duration: (float) The duration of silence in seconds that triggers the end of a speech segment for transcription (default: 0.8).