Startup Gnani to launch its voice foundation model at AI Impact Summit

IndiaAI-backed startup to unveil 14bn-parameter voice model in six languages, with plans to scale to all 22 Indian languages within 18 months

Gnani.AI, an IndiaAI Mission startup, will launch its voice foundation model at the India AI Impact Summit in New Delhi next week, its cofounder and chief technology officer Ananth Nagaraj said.

 

The pre-trained AI model is built on 14 billion parameters and capable of processing and generating complex, multilingual speech in real time. It is designed for low-latency, speech-to-speech communication in customer support, education, accessibility and public-facing systems.

 

“We are launching our voice-to-voice foundation model in six languages and target to expand it to all 22 languages in the next 18 months,” Nagaraj said. (The six languages are English, Hindi, Kannada, Telugu, Tamil and Gujarati.)

 

Gnani will also launch a multilingual text-to-speech model capable of cloning a voice “hyper-realistically”. Named Vachana STT, it has been trained on over one million hours of real-world voice data spanning more than 1,056 real-world domains.

 

“It will be a voice-plus avatar. If you look at the avatar, where we will use a customer service representative, it is a digital twin of him or her. In our voice-to-voice model or speech-to-text and the LLM [large language model] understanding layer, along with text-to-speech, all of them will be driving this avatar.”

 

Voice AI is expected to be the only practical interface for digital equality in India and the next big thing, according to Infosys chairman Nandan Nilekani. “Just as UPI made digital payments effortless for everyone, voice-driven interfaces can remove barriers to opportunity in sectors such as agriculture, education and others for every citizen. Literacy will no longer be a barrier,” he said last month, referring to Unified Payments Interface.

 

Nagaraj said he agreed with such views. “Human and machine interaction is going to be through voice. India being so diverse, there is going to be voice AI in people’s mother tongues. I think voice AI, when it unlocks human-machine interaction through native languages, will expose almost 1 billion users to the technology.”

 

Gnani is among a clutch of startups building India’s capabilities in advanced AI technology. Sarvam hopes to come out with its sovereign AI model, Vivek Raghavan, the company’s cofounder, said in November.

 

Soket is developing India's first open-source 120 billion parameter foundation model optimised for the country's linguistic diversity, targeting sectors such as defence, healthcare, and education.

 

Gnani, which counts Samsung Venture Investment among its investors, expects to end FY26 with a revenue of around Rs 160 crore, compared to Rs 56 crore it earned the previous year as it works with more financial services firms such as HDFC, Bank of Baroda and IDFC First.