On February 18, Bengaluru-based conversational AI startup Gnani.ai introduced Vachana TTS, a new text-to-speech system that can produce human-like audio and clone voices in 12 Indian languages. The launch took place at the India AI Impact Summit in Delhi and is the company’s second offering under its Inya VoiceOS platform, developed as part of the India AI Mission.
According to the company, Vachana TTS has achieved a Mean Opinion Score of 4.23 and a character error rate of below 0.6 per cent, making it suitable for production-scale deployment.
Supports 12 Indian languages
Vachana TTS supports Hindi, Bengali, Tamil, Telugu, Kannada, Malayalam, Gujarati, Marathi, Punjabi, Odia, Assamese and Indian English.
The company said the model captures natural rhythm, pronunciation and tone suited to each language. “Independent evaluations confirm Vachana TTS outperforms existing global TTS providers on Indic language naturalness and pronunciation accuracy, at a substantially lower cost per character - making high-quality voice synthesis economically viable for government and enterprise deployments at population scale,” Moneycontrol cited the company as saying in a release.
Also Read | What is Gemini 3.1 Pro? Google’s latest AI update explained
How Vachana works
Vachana can recreate a person’s voice using less than 10 seconds of audio. With a short speech sample, the system preserves traits such as pitch, speaking speed and overall tone, allowing the generated output to sound close to the original speaker.
The same voice can then speak in multiple Indian languages while maintaining consistency. This feature is designed to support enterprises and government departments operating across different linguistic regions.
Built for scale in India
Vachana TTS is designed to function in low internet conditions and can handle multiple users simultaneously. It supports real-time voice generation for chatbots and conversational systems, as well as bulk audio content creation.
Also Read | Think you can spot a scam? AI-powered deepfakes and phishing are making it harder than ever
Gnani.ai said the model is built, trained and deployed entirely within India, with voice data and models hosted in Indian data centres to support data localisation and privacy needs.
Speaking at the summit, Ganesh Gopalan, Co-Founder & CEO, Gnani.ai, said, ""We are bringing genuine emotion into synthesised speech - warmth, urgency, empathy - delivered with significantly better accuracy than anything built for Indian languages before, and at a price point that makes it accessible to every enterprise and government body in the country."
