As voice interfaces expand across customer support and enterprise automation, Gnani.ai is pushing advanced speech-to-speech AI models that move beyond basic text-to-speech systems. Speaking at the AI Impact Summit 2026 at Bharat Mandapam in New Delhi, Ganesh Gopalan, CEO and co-founder of Gnani.ai, said low latency and emotional intelligence are essential for building reliable enterprise-grade voice AI solutions.
At the summit, Gnani.ai unveiled a 5 billion-parameter closed speech model under research preview as part of its sovereign AI stack, Inya VoiceOS. The company, selected under the Centre’s Rs 10,372-crore IndiaAI Mission, also introduced Vachana TTS, a voice cloning model supporting 12 Indian languages, including Hindi, Bengali, Tamil, Telugu, Kannada, Malayalam, Gujarati, Marathi, Punjabi, Odia, Assamese and Indian English.
“Voice is the most natural form of communication in the world,” Gopalan said, adding that multilingual capability is a basic necessity for voice AI systems in India. He argued that most startups rely on text-based processing before converting it back to speech, which misses tone and emotion. “If you are having a telephony conversation with a voice AI system, if that system even hesitates for a second, you’ll slam the phone,” he said, stressing that low latency and emotional context are essential. The company is working to reduce hallucinations by limiting layers in its AI models.
Gnani.ai began collecting proprietary voice datasets in 2017 and claims to have built the largest annotated Indian language voice dataset. “We were never satisfied until we had covered every district of India,” Gopalan said. The company used a mix of proprietary, public, AIKosh, and synthetic datasets for training. Backed by Samsung and InfoEdge, the Bengaluru-based startup is also in talks for fresh funding amid rising demand for enterprise voice automation. Gopalan emphasised owning every layer of the AI pipeline instead of relying on global APIs, warning that wrapper models may not survive at scale due to pricing, latency and accuracy pressures.
On compute access, he said Nvidia H100 GPUs are only now becoming available in India but praised the IndiaAI Mission for affordable pricing, reportedly a little over $1 per hour for H100, compared to Rs 500–600 per hour by some private providers. Addressing voice cloning risks, he said Gnani.ai runs parallel systems — Inya Assist for cloning and Inya Shield for detection and biometrics. Older systems relied on a single passphrase like “My voice is my password”, but newer banking solutions use dynamic passphrases and behavioural analysis to prevent fraud.
Also read: Viksit Workforce for a Viksit Bharat
Do Follow: The Mainstream LinkedIn | The Mainstream Facebook | The Mainstream Youtube | The Mainstream Twitter
About us:
The Mainstream is a premier platform delivering the latest updates and informed perspectives across the technology business and cyber landscape. Built on research-driven, thought leadership and original intellectual property, The Mainstream also curates summits & conferences that convene decision makers to explore how technology reshapes industries and leadership. With a growing presence in India and globally across the Middle East, Africa, ASEAN, the USA, the UK and Australia, The Mainstream carries a vision to bring the latest happenings and insights to 8.2 billion people and to place technology at the centre of conversation for leaders navigating the future.
