Deep-tech company Shunya Labs on Monday launched a multilingual Voice AI platform that it says supports over 216 languages and dialects, with a focus on South Asian and Southeast Asian languages. The company said the platform is built to run on standard CPU-based servers rather than GPU infrastructure, and can be deployed on-premise, on the cloud, or at the edge.
Why It Was Launched
Shunya Labs said the platform is aimed at closing a gap in India's digital services. According to the company, more than 90% of internet users in India prefer content in their native language, while most enterprise voice systems are built on English-first architecture that struggles with Hinglish, code-switching, regional accents and low-resource languages. Reuters and other agencies have not independently verified this figure, and it should be treated as a company-cited statistic pending confirmation from a primary source such as government or industry survey data.
Also Read https://news4bharat.com/bharat-explainers/small-language-models-india-bfsi-saas-ai
What The Platform Does
The company said the platform combines three components: speech recognition trained on Indic and Asian language data, natural language understanding, and generative AI for conversational responses. Shunya Labs said the models were trained directly on Indic-language data rather than adapted from English-language systems.
CPU-Based Architecture
Shunya Labs said its CPU-first design allows enterprises to deploy voice AI on standard servers instead of costlier GPU-based cloud infrastructure, which it said supports lower latency, lower cost and on-premise data handling. The company said this is intended to appeal to sectors with strict data-privacy requirements, including banking and financial services, healthcare, government citizen services, and customer support.
Note: Claims regarding cost savings and latency have not been independently verified against third-party benchmarks and should be attributed to the company throughout.
The company listed the following potential applications for the platform:
Shunya Labs said the platform can be deployed across phone, app, web and IoT channels.
Benchmark Claim
Shunya Labs said its speech recognition models have achieved "world-record" benchmarks and outperform models from major global technology companies on Indic-language tasks.
This claim is sourced solely to the company; no independent or third-party benchmark data has been reviewed to verify it, and it should not be reported as an established fact without further verification from an academic or industry benchmarking source.
Also Read Google's Spam Update Just Changed The Rules, Here Is What Indian Banks Need To Know
Company Statement
Ritu Mehrotra, founder and CEO of Shunya Labs, said in a statement: "India's digital future will be shaped not just by technology, but by accessibility. At Shunya Labs, we believe that language should never be a barrier to participation in the digital economy. Our next-generation Multilingual Voice AI platform has been built with the vision of empowering every Indian to interact with technology naturally in their preferred language. As businesses and public institutions accelerate digital adoption, we are committed to creating AI solutions that are inclusive, intelligent, and designed for Bharat's unique linguistic diversity."
India's digital services push, including UPI, Aadhaar, DigiLocker and Jan Dhan Yojana, has extended financial and identity services to a large section of the population, though these services largely operate in English and Hindi.
Industry observers have said that future growth in digital adoption is expected to come from smaller towns and rural areas, where users are more likely to speak languages such as Bhojpuri, Marathi, Tamil, Kannada, Bengali, Odia and Assamese. Voice-based interfaces are seen by some industry players as one route to reaching these users, though the extent of this impact has not been independently assessed.
