Sarvam AI TTS

The SarvamTTSService provides high-quality speech synthesis for Indian languages. It supports multiple regional languages and offers both WebSocket-based streaming and standard HTTP interfaces.

Installation

To use Sarvam AI, install the required dependencies:

pip install "piopiy-ai[sarvam]"

Prerequisites

• A Sarvam AI API key (Get yours here).
• Set your API key in your environment:

  export SARVAM_API_KEY="your_api_key_here"

Configuration

SarvamTTSService (WebSocket) Parameters

Parameter	Type	Default	Description
api_key	str	Required	Your Sarvam AI API key.
voice_id	str	"anushka"	Speaker voice ID (see below).
model	str	"bulbul:v2"	Model ID (bulbul:v2, bulbul:v3).
params	InputParams	None	Configuration parameters (see below).

InputParams Options

Option	Type	Default	Description
language	Language	Language.EN	Target language (e.g., Language.HI).
pitch	float	0.0	Voice pitch adjustment (-0.75 to 0.75).
pace	float	1.0	Speech pace multiplier (0.3 to 3.0).
loudness	float	1.0	Volume multiplier (0.1 to 3.0).
temperature	float	0.6	Randomness for v3 models (0.01 to 1.0).

Usage

Basic Streaming Setup

import os
from piopiy.services.sarvam.tts import SarvamTTSService
from piopiy.transcriptions.language import Language

tts = SarvamTTSService(
    api_key=os.getenv("SARVAM_API_KEY"),
    voice_id="anushka",
    params=SarvamTTSService.InputParams(
        language=Language.HI # Hindi
    )
)

Available Voices

Gender	Voice IDs
Female	anushka, manisha, vidya, arya
Male	abhilash, karun, hitesh

Notes

• Language Support: Sarvam supports Bengali, English (India), Gujarati, Hindi, Kannada, Malayalam, Marathi, Odia, Punjabi, Tamil, and Telugu.
• Model Versions: bulbul:v2 is the stable standard, while bulbul:v3 offers enhanced naturalness and lower latency.