sarvam-ai

name	sarvam-ai
description	Indian language AI via Sarvam AI APIs. Text-to-Speech (TTS) with 38 voices in 22+ Indian languages using Bulbul v3, Speech-to-Text (STT) with Saaras v3. Use when the user asks to generate speech in Indian languages, transcribe Indian language audio, or needs Indic TTS/STT.
metadata.openclaw	{"emoji":"🇮🇳","requires":{"env":["SARVAM_API_KEY"]}}

Sarvam AI — Indian Language TTS & STT

Sarvam AI provides high-quality Text-to-Speech and Speech-to-Text for 22+ Indian languages.

Authentication

• API key must be set in $SARVAM_API_KEY (gateway env)
• Header: API-Subscription-Key: $SARVAM_API_KEY
• Base URL: https://api.sarvam.ai

Quick Start (Helper Scripts)

# Text-to-Speech
scripts/sarvam-tts.sh "नमस्ते दुनिया" --lang hi-IN --speaker aditya --output hello.mp3

# Speech-to-Text
scripts/sarvam-stt.sh audio.wav --lang hi-IN

Text-to-Speech (TTS)

Batch TTS

curl -X POST "https://api.sarvam.ai/text-to-speech" \
  -H "API-Subscription-Key: $SARVAM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "inputs": ["Your text here"],
    "target_language_code": "hi-IN",
    "speaker": "aditya",
    "model": "bulbul:v3"
  }'

Response: {"request_id": "...", "audios": ["<base64-wav>"]}

Streaming TTS (preferred for longer text)

curl -X POST "https://api.sarvam.ai/text-to-speech/stream" \
  -H "api-subscription-key: $SARVAM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Your text here",
    "target_language_code": "hi-IN",
    "speaker": "aditya",
    "model": "bulbul:v3",
    "pace": 1.0,
    "speech_sample_rate": 22050,
    "output_audio_codec": "mp3",
    "enable_preprocessing": true
  }' -o output.mp3

TTS Parameters

Param	Values	Default
model	bulbul:v3 (best), bulbul:v2	bulbul:v3
target_language_code	See language codes below	required
speaker	See speakers below	required
pace	0.5 - 2.0	1.0
speech_sample_rate	8000, 16000, 22050, 44100	22050
output_audio_codec	mp3, wav, aac, opus, flac, pcm, mulaw, alaw	mp3
temperature	0.0 - 1.0	0.6

Bulbul v3 Speakers (38 voices)

aditya, ritu, ashutosh, priya, neha, rahul, pooja, rohan, simran, kavya, amit, dev, ishita, shreya, ratan, varun, manan, sumit, roopa, kabir, aayan, shubh, advait, amelia, sophia, anand, tanya, tarun, sunny, mani, gokul, vijay, shruti, suhani, mohit, kavitha, rehan, soham, rupali

Speech-to-Text (STT)

Batch STT

curl -X POST "https://api.sarvam.ai/speech-to-text" \
  -H "API-Subscription-Key: $SARVAM_API_KEY" \
  -F "file=@audio.wav" \
  -F "model=saaras:v3" \
  -F "language_code=hi-IN"

Response: {"transcript": "...", "language_code": "hi-IN"}

STT Models

Model	Best for
saaras:v3	Best quality, all languages
saarika:v2.5	Good quality, faster
saarika:flash	Fastest, lower quality

Language Codes

Language	Code
Hindi	hi-IN
Bengali	bn-IN
Telugu	te-IN
Marathi	mr-IN
Tamil	ta-IN
Gujarati	gu-IN
Kannada	kn-IN
Malayalam	ml-IN
Odia	od-IN
Punjabi	pa-IN
English (Indian)	en-IN
Auto-detect	unknown (STT only)