OTP verifications, EMI reminders, delivery updates. When the message matters, voice in the customer's language gets through. Personalized, natural-sounding calls in 11 Indian languages, generated in real-time.
Conversational ・ Friendly
News ・ Authoritative
Conversational ・ Consistent
Entertainment ・ Dynamic
Want to use this API?
A PM Kisan beneficiary in rural Odisha receives an SMS about a payment credit. He can't read it. A Jan Dhan account holder gets a fraud alert SMS in English. She doesn't understand it. For transactional messages that MUST reach the recipient, voice in their language is the only reliable channel.
'Your OTP is 4-8-3-9' in a monotone robotic voice sounds like a scam call. Customers hang up. A natural voice saying 'Namaste Suresh ji, aapka OTP hai 4-8-3-9, kripya 5 minute mein use karein' sounds legitimate and gets acted on.
A payment app sending 500,000 EMI reminders per day across Hindi, Tamil, Telugu, Bengali, and Gujarati would need an army of call center agents. Dynamic TTS generates each call with the customer's name, amount, and due date in their language. Fully automated, fully personalized.
Define your message templates
Create templates with placeholders: {name}, {amount}, {due_date}, {order_id}, {otp}. One template per message type. 'Namaste {name} ji, aapki EMI Rs {amount} ki payment {due_date} ko due hai.'
Fill with live customer data
Your system populates templates with real customer data at call time. Name, amount, date, OTP, order number. Each call is unique to the recipient.
Generate audio in real-time
Send the filled template to Sarvam's API. Get natural audio back in under 250ms. Fast enough for real-time OTP calls where the customer is waiting.
Deliver via your telephony stack
Route the audio through Exotel, Ozonetel, Twilio, or any SIP/VoIP system. 8kHz telephony-optimized output. Sounds natural on actual phone lines.
from sarvamai import SarvamAI client = SarvamAI(api_subscription_key="YOUR_KEY") # Payment reminder in Telugu audio = client.text_to_speech.convert( text="Ramesh garu, meeku gedam chestunnam. Mee EMI Rs 5,000 ee nela 15 tarikhu varaku chellinchaali.", target_language_code="te-IN", model="bulbul:v3", speaker="meera", pace=1.0, enable_preprocessing=True ) with open("payment_reminder.wav", "wb") as f: f.write(audio.audios[0])
Fast enough for real-time OTP calls. Customer dials, API generates, voice plays. No perceptible delay.
Sounds natural on phone lines, not just in browser demos. Optimized for the bitrate callers actually hear.
Hindi, Tamil, Telugu, Bengali, Malayalam, Marathi, Gujarati, Kannada, Punjabi, Odia, Assamese. Route to the customer's preferred language.
API scales to handle enterprise call volumes. Batch process EMI reminders or real-time OTP verification.
Starting at ₹30 per 10K characters. View pricing
Higher is better
ElevenLabs Flash V2.5
ElevenLabs V3 Alpha
Cartesia Sonic-3
Two-factor authentication over voice. The OTP is spoken in the customer's language with clear number pronunciation. Particularly critical for customers who struggle to read SMS OTPs on small screens or in unfamiliar scripts.
Personalized voice reminders with the customer's name, EMI amount, and due date. In their language. A natural voice reminder is harder to ignore than an SMS and more personal than a push notification. Reduces missed payments without human call center agents.
Order confirmation, dispatch notification, delivery ETA, and COD amount. All spoken in the customer's language. Indian addresses and PIN codes pronounced correctly. 'Aapka order 7-7-2-1 kal subah tak deliver ho jayega' is clearer than an SMS the customer might not read.
Patient appointment confirmations, vaccination scheduling, medication timing reminders. For elderly patients or those with limited literacy, a voice call in their language ensures the message is understood and acted on.
India's SMS ecosystem is overwhelmed. The average Indian mobile user receives dozens of transactional and promotional messages daily. Important messages get buried. Spam filters catch legitimate alerts. And for the 200+ million adults with limited reading ability, SMS is fundamentally inaccessible. A voice call cuts through the noise. It demands attention, communicates in the recipient's language, and confirms comprehension in a way text never can.
The economics of voice notifications have changed. Traditional outbound voice campaigns required recorded prompts or live agents. Dynamic TTS eliminates both. Generate a personalized message for each customer in real-time, in their language, with their specific details inserted naturally. The cost per call drops to the telephony charge plus a few paise for audio generation. At this cost, every transactional SMS can have a voice fallback for recipients who don't read it.
Extend from one-way notifications to full voice interactions.
Natural, expressive voices for customer notifications in every major Indian language.
Start sending voice notifications Powered by Bulbul V3
Start sending voice notifications
Powered by Bulbul V3
