The IndiaAI Mission-backed startup's new text-to-speech model supports over 50 voices, multilingual conversations and voice cloning for customer service, education and enterprise applications.
Indian artificial intelligence startup Chariot has launched Chariot-1-TTS-Multilingual, a text-to-speech (TTS) model designed to generate speech across 11 Indian languages and English. The model introduces capabilities including multilingual code-mixing, contextual pronunciation and voice cloning, targeting applications across customer service, education and enterprise communications.
Announced on November 12, 2026, the model supports more than 50 voices and allows users to switch between two supported languages within the same sentence. It can also generate synthetic speech using a few seconds of reference audio and process up to 1,000 characters in a single request.
Unlike conventional text-to-speech systems that depend on predefined text normalisation rules, Chariot said its model uses contextual reasoning to determine how written expressions should be pronounced.
For instance, the system can distinguish between the expression "B+" when referring to a blood group and when used as an academic grade. It is also designed to interpret currency amounts, dates, abbreviations and technical terminology according to their context.
According to the company, the model recorded a character error rate of 7.60% for Hindi and 6.47% across nine other Indian languages in its internal evaluations.
On an English verbalisation benchmark covering more than 1,000 sentences across 40 categories, Chariot reported an accuracy rate of 93.78%.
The company also claimed that its model ranked ahead of selected offerings from ElevenLabs, Cartesia, Microsoft, OpenAI, Amazon Web Services and Sarvam across several evaluation measures. However, the benchmark coverage varied between competing models, and the results have not been independently verified.
Parth Sarthi, Co-founder of Chariot, said the model was developed to address everyday speech patterns involving numbers, symbols, abbreviations and transitions between languages.
The company is positioning the technology for deployment in AI voice agents, customer support systems, educational platforms and multilingual public helplines.
Developers can select from the existing voice catalogue or create customised voices using reference recordings.
Chariot, previously known as Gan.ai, is headquartered in Delhi and was founded by Sarthi and Suvrat Bhooshan, who have backgrounds at Google DeepMind and Meta's Fundamental AI Research laboratory.
The company has previously worked on AI-powered voice applications involving brands and organisations including Zomato, Coca-Cola, Cadbury, HDFC Bank and Doordarshan.
Its earlier projects include personalised voice experiences and multilingual content generation for marketing and public communication campaigns.
The launch follows Chariot's introduction of Chariot-1-ASR Realtime, a speech recognition model designed for Indian-language transcription.
Chariot is also among the organisations selected under the IndiaAI Mission's Foundation Models initiative, which supports the development of indigenous artificial intelligence capabilities.
The latest release expands Chariot's speech AI portfolio as developers and enterprises explore multilingual voice interfaces for customer engagement, accessibility and automated communication.
Disclaimer: This article may include information derived from interviews, press releases, public statements, research, company communications and other publicly available or third-party sources. Such material may be summarised, paraphrased or contextualised for journalistic and editorial purposes. All rights in third-party content remain with their respective owners.