Shunya Labs Launches Va· AI Translation for 55 Indian Languages

"VaÄ· is accompanied by the world's number 1-ranked open-weight speech recognition model and a neural text-to-speech engine, all released simultaneously and all free to deploy," the company said in a statement.

The company claimed that VaÄ· is a real-time "Any-to-Any Translation system that converts speech in any of India's 55 languages into any of the other 54," instantly, with end-to-end latency under 1.5 seconds.

Shunya Labs said the system also features zero‑shot voice cloning, which ensures no prior training data is required. Vaķ's neutral TTS delivers natural-sounding output across all 55 languages with streaming capability, custom voice creation, zero-shot voice cloning, and prosody and emotion control, the statement said.

The system also preserves the speaker's voice and emotional register, it added.

The Pingala ASR ranked first "on the Hugging Face OpenASR leaderboard with the lowest ever recorded 3.10 per cent word‑error rate."

It features a CPU-first architecture delivering sub-250ms latency for edge and offline deployment. The full model weights are publicly available for local deployment, enabling organisations to run the models on‑premises without sending voice data to foreign servers.

It also aids sovereign deployment for government agencies, hospitals reaching rural patients, and courts delivering justice in local languages.

VaÄ· made cross-linguistic citizen services, healthcare outreach, judicial access, and educational delivery technically and economically viable for the first time at a national scale, the release said.

"This is a sovereign, self-funded innovation built within the Nasscom ecosystem. Our vision is simple: every developer can build, every government can deploy, and every Indian can be heard in their own language," said Sourav Bandyopadhyay, Founder & Chief Scientist, Shunya Labs.

VaÄ· covers 43 Indoâ€'Aryan languages including Hindi, Bengali, Marathi, Gujarati, Urdu, Bhojpuri, and Rajasthani, 7 Dravidian languages, 3 Sinoâ€'Tibetan and 1 Austroasiatic language, plus Indian English, collectively reaching over 1.17 billion native speakers in aggregate, the startup said.

- IANS