Navana.ai Launches Bodhi TTS in 10 Indian Languages

Navana.ai launches Bodhi TTS, an Indian voice AI model in 10 languages, priced far below ElevenLabs and Sarvam.

By Entrepreneur Staff | Sep 25, 2026
L-R -Jai Nanavati, CTO and Co-Founder, Raoul Nanavati, CEO and Co-Founder, Navana AI

You're reading Entrepreneur India, an international franchise of Entrepreneur Media.

Navana.ai, a voice AI company serving regulated enterprises, has launched Bodhi TTS, a text-to-speech model built for real-time Indian voice AI applications. It launches with over 50 natively Indian voices across 10 languages, including Hindi, Telugu, Tamil, Marathi, Bengali, Odia, Malayalam, Kannada, English and Gujarati, priced at INR 12 per 10,000 characters, roughly INR 0.72 per minute of generated speech.

Navana.ai trains and owns the entire Bodhi speech stack, with no data routed through third-party model providers, allowing customer audio and personal data to remain within an enterprise’s own security perimeter. The model is approximately five times smaller than comparable architectures, enabling it to run inside a bank’s own data centre. The company is ISO 27001 and SOC 2 Type II certified.

Bodhi TTS handles Indian text natively, reading figures, PAN numbers, phone numbers, dates and scheme names the way a bank would say them, and allows enterprises to upload dictionaries for custom pronunciation control. The model begins producing audio in under 100 milliseconds and supports zero-shot voice cloning from a few seconds of reference audio, added live via API without redeployment.

At INR 12 per 10,000 characters, Bodhi TTS undercuts ElevenLabs, which prices its Multilingual and v3 models at roughly INR 95 per 10,000 characters, and Sarvam AI’s Bulbul v3 at INR 30. Bodhi models already sit behind voice agents supporting more than INR 1,000 crore in monthly loan disbursals across Navana.ai’s banking clients, including Bajaj Finserv, Ujjivan Small Finance Bank and Jana Small Finance Bank. The company’s speech recognition model, Bodhi STT, covering 10 languages and over 40 dialects, opens by invitation the same day.

“India runs on voice, and most of it is not in English. We built Bodhi from the ground up to represent every accent, and priced it for the call volumes Indian businesses actually run, with everything in the box,” said Raoul Nanavati, Co-Founder and CEO, Navana.ai.

Navana.ai, a voice AI company serving regulated enterprises, has launched Bodhi TTS, a text-to-speech model built for real-time Indian voice AI applications. It launches with over 50 natively Indian voices across 10 languages, including Hindi, Telugu, Tamil, Marathi, Bengali, Odia, Malayalam, Kannada, English and Gujarati, priced at INR 12 per 10,000 characters, roughly INR 0.72 per minute of generated speech.

Navana.ai trains and owns the entire Bodhi speech stack, with no data routed through third-party model providers, allowing customer audio and personal data to remain within an enterprise’s own security perimeter. The model is approximately five times smaller than comparable architectures, enabling it to run inside a bank’s own data centre. The company is ISO 27001 and SOC 2 Type II certified.

Bodhi TTS handles Indian text natively, reading figures, PAN numbers, phone numbers, dates and scheme names the way a bank would say them, and allows enterprises to upload dictionaries for custom pronunciation control. The model begins producing audio in under 100 milliseconds and supports zero-shot voice cloning from a few seconds of reference audio, added live via API without redeployment.

Entrepreneur Staff • Editor

Entrepreneur Staff
For more than 30 years, Entrepreneur has set the course for success for millions of... Read more

Related Content