Author: vitorcherulli

Thousands of anonymous listeners cast their votes on Voice Arena, comparing Hindi voice samples without knowing which company produced them. The result: Dheemanth Reddy’s Maya 2 claimed the top spot for Hindi, surpassing models from ElevenLabs and Cartesia with over 15,000 blind votes. This public leaderboard gave Reddy an independent measure of the naturalness he had been working to achieve.

Reddy builds voice AI models designed to sound natural in everyday languages. He believes listeners can detect nuances that no product description can capture. A voice may pronounce every word clearly yet still feel foreign because its rhythm doesn’t match how people actually speak. Blind voting placed that judgment directly with native speakers.

“You do not have to explain natural speech to someone who has spoken the language their whole life,” Reddy says. “They hear the voice and know whether it feels right.”

Voice Arena removed brand recognition from the comparison, letting each sample stand on what listeners heard. Voters didn’t know which model Reddy had created, so his name and company couldn’t influence their choices. Each response contributed to a record built by people outside his team.

The ranking also addressed a core question: What does it mean for a machine to sound native? For Reddy, basic intelligibility is just the starting point. He considers how a sentence flows, how delivery affects meaning, and whether the voice sounds like it belongs in the conversation. These qualities matter most when a model is expected to interact, not just read a script.

“Native listeners can hear when the words are right but the voice still feels unfamiliar,” he says. “That difference matters.”

Reddy believes the voice AI field has focused too much on English demonstrations that perform well under limited conditions. Success in one language doesn’t automatically transfer to another. A model built around English patterns may struggle elsewhere because each language has its own cadence and expressive habits. Maya 2 was developed with the expectation that Hindi listeners deserve the same level of care.

This focus separates Reddy’s work from treating additional languages as afterthoughts. His goal is speech that reflects how people communicate within the language itself. That requires close attention to voice sound and the model’s ability to understand meaning. A system can’t sustain a convincing exchange when the voice feels unnatural or the conversation loses direction.

“The language cannot feel like an added feature,” Reddy says. “It has to feel like the model was built to speak with that person from the beginning.”

Voice Arena gave Reddy a public test for that standard. Listeners didn’t respond to the model’s size or the organization’s reputation. They heard samples and selected their preferences. More than 15,000 blind votes contributed to Maya 2’s Hindi ranking, giving the result a basis beyond Reddy’s own assessment.

He doesn’t present the ranking as proof that every multilingual voice problem is solved. It identifies one area where the model performed strongly under public comparison. It also supports his belief that listening tests should play a key role in evaluating naturalness. People who speak the language bring knowledge that can’t be replaced by a company’s description of its own model.

“Voice is experienced by hearing it,” he says. “The people listening should have a real role in deciding whether the model works.”

Maya 2 followed Reddy’s earlier work on Maya 1, an open-weights speech model available for developers. The later model faced a different form of outside judgment through blind listener preference. It had to earn votes without depending on Reddy’s explanation of what the audience was supposed to hear.

Reddy earned a master’s degree in Computing, Entrepreneurship, and Innovation from New York University. He now serves as co-founder and CEO of Maya Research, where he builds models and directs technical work. He was also awarded an Emergent Ventures grant for developing voice AI that sounds native.

As co-founder and CEO, Reddy is building that work within a company backed by South Park Commons. His earlier Maya 1 model is one of the world’s top open-weights speech models and the only model from India on the Artificial Analysis / Speech Arena leaderboard. His technical experience has also led to roles outside Maya Research, including serving as a judge at the Cerebral Valley AI hackathon.

The significance of Maya 2’s Hindi ranking lies in the standard it applies. Reddy wants multilingual models to be evaluated within the languages they claim to speak well. A successful English demonstration says little about whether a Hindi voice will sound familiar to someone who hears the language at home. Each language deserves direct testing by listeners who can recognize its details.

Reddy expects people to use speech for more tasks that currently require typing or navigating menus. Those interactions will depend on whether the system understands the speaker and responds in a voice that feels appropriate. Poor native-language quality can create distance before the conversation begins. A model that sounds natural helps users focus on the task rather than the interface’s limitations.

Reddy sees the Hindi leaderboard as one public checkpoint in a much longer technical effort. It showed that Maya 2 could compete through blind listening and placed native speakers at the center of evaluation. For him, that is where judgment of a voice model belongs.

“People should be able to hear the voice and trust their own judgment,” Reddy says. “That is the standard the model has to meet.”