Posted by New Delhi: United We Care, a mental health tech company leveraging AI, has launched Shunya Labs, a new deeptech venture aimed at transforming the core infrastructure of voice AI. Designed from scratch and optimised for India’s multilingual population, Shunya Labs claims to have developed an Automatic Speech Recognition (ASR) system that outperforms major global players across accuracy, speed, and language capabilities and crucially, runs efficiently on standard CPUs.
With a record-low average Word Error Rate (WER) of 3.37%, Shunya Labs’ ASR platform supports real-time transcription across 32 Indic languages, including Hindi, Marathi, Assamese, and Maithili, with 7 more under development. The system also enables on-premises deployments, designed to meet privacy and latency requirements for mission-critical applications in sectors such as healthcare, finance, government, and defence.
“We didn’t set out to beat the benchmarks — we set out to invent what didn’t exist,” said Ritu Mehrotra, Founder of United We Care. “And when we built it, we realised we’d created something the world didn’t know it needed: AI voice infrastructure that listens like a human, runs like a machine, and respects the sanctity of privacy. That’s Shunya Labs.”
Shunya Labs emerged from technology originally developed for Stella, United We Care’s AI wellness platform. The voice AI engine evolved to incorporate a 230 million-node Clinical Knowledge Graph and a proprietary Spatio-Temporal Graph Attention Network (STGAT), enabling it to handle real-time emotional and contextual understanding, a rare feat in existing voice systems.
According to the company, the platform is optimised for both CPU and GPU environments, reducing enterprise cloud computing costs by up to 20 times, while maintaining military-grade privacy standards.
“Shunya isn’t just a name. It’s our origin point — a nod to the Indian discovery of zero that changed mathematics forever,” said Sourav Banerjee, Founder and CTO. “At Shunya Labs, we start from first principles, engineer with surgical precision, and build for the future — not features. In a world drowning in noise, we build the intelligence layer that actually listens.”
With production deployments planned across multiple global markets and an upcoming release on Hugging Face, Shunya Labs aims to position itself not as a voice API provider, but as the bedrock of next-generation AI voice systems.