Navana.ai launches Bodhi TTS in 10 Indian languages

MUMBAI: Navana.ai has put Bodhi TTS into general availability for enterprises that need text-to-speech in Indian languages, not only English call-centre scripts. The model is live on the company’s platform from 23 September 2026 and is pitched at banks, insurers and lenders that run high-volume voice bots.

Bodhi TTS covers Hindi, Telugu, Tamil, Marathi, Bengali, Odia, Malayalam, Kannada, English and Gujarati. Navana.ai says it ships with more than 50 voices and starts streaming audio in under 100 milliseconds, a latency band meant for live phone and app dialogues rather than offline dubbing.

What Bodhi TTS offers enterprises

Indian business copy is messy in ways global voice stacks often mishandle. The company says Bodhi TTS is tuned for currency amounts, PAN numbers, phone numbers, dates, EMI schedules, scheme names and account numbers that appear in banking and insurance scripts. Teams can upload pronunciation dictionaries for product names, branch names and customer names without retraining the model.

Zero-shot voice cloning is part of the package. A few seconds of reference audio can be sent over the API to create a custom voice; Navana.ai says the reference is cached so later turns do not add delay. The firm also claims the model is about five times smaller than the nearest comparable architecture, which would let some buyers keep inference inside their own data centres when compliance teams refuse to send customer audio outside.

Raoul Nanavati, co-founder and chief executive of Navana.ai, framed the launch around accent and cost. “India runs on voice, and most of it is not in English,” he said. “Until now, a bank that wanted a voice agent in Marathi or Tamil had to choose between a global model that mispronounced everyday words and a price that only worked for a pilot.” He said Bodhi was built for regional accents and priced for the call volumes Indian firms actually run.

Pricing, credits and early access

List pricing is Rs 12 per 10,000 characters. Navana.ai puts that at roughly Rs 0.72 per minute of generated speech. New sign-ups get Rs 1,000 in free credits. Banks, insurers and lenders can apply for early access to Bodhi TTS and Bodhi STT; the first 100 approved teams are offered 100 hours of usage credits valid for 90 days, according to the company.

The product pages on navana.ai already market India-ready Bodhi TTS APIs and a playground. That matters for buyers who want to test pronunciation on their own scripts before tying the model into an IVR or WhatsApp voice flow. Navana.ai also stresses on-prem options and alignment with data-localisation expectations for regulated sectors.

What is still unclear from the public brief is how Bodhi TTS will be benchmarked against other Indian speech vendors on noisy, code-mixed audio at scale. For now, the concrete offer is a dated launch window, a fixed per-character price, a ten-language matrix and enterprise controls that do not require a full retrain every time a brand name changes.

For contact-centre and collections teams, the immediate test is practical: feed a Marathi EMI reminder and a Tamil KYC prompt through the playground, listen for number and date errors, then decide whether the latency and cloning workflow fit the production stack.