2 hrs ago
IIT Madras Launches Four Open AI Models for Indian Languages
Bodhan AI has created four computer models that work with Indian languages.
The models can recognize speech, create speech, translate text and read words from images.
They support many Indian languages, including regional accents and dialects.
The speech-recognition model works with 27 languages.
The translation model supports 22 languages, while the speech-generation and OCR models support 23 languages each.
The models are being shared openly so schools, companies and developers can use them.
They are hosted on digital infrastructure designed to remain under Indian control.
IIT Madras says the project can help build better AI tools for multilingual education.
Bodhan AI, NVIDIA and AI4Bharat plan to continue improving these models together.
Bodhan AI, incubated at IIT Madras, launched four foundational AI models for Indian languages.
The models provide speech recognition, speech generation, machine translation and optical character recognition.
They are available with open weights and hosted APIs on sovereign digital infrastructure.
The speech recognition model supports 27 languages, while the other models cover 22 or 23 languages.
Bodhan AI, NVIDIA and AI4Bharat will collaborate on datasets, training methods and evaluations.
- Who
- Bodhan AI, the Indian Institute of Technology Madras, AI4Bharat and NVIDIA are involved.
- What
- Four foundational AI models for Indian-language speech, translation and optical character recognition were launched.
- Where
- The launch was by Bodhan AI at the Indian Institute of Technology Madras in Chennai.
- When
- Why
- The models are intended to support India's multilingual education ecosystem through sovereign digital public infrastructure.
Key facts
- Models launched
- Speech recognition, speech generation, machine translation and optical character recognition
- Speech recognition coverage
- 27 Indian languages
- OCR coverage
- 23 Indian languages
- Translation coverage
- 22 Indian languages
- Text-to-speech coverage
- 23 Indian languages
- Availability
- Open weights and hosted APIs on sovereign digital infrastructure
- Education initiative
- Part of the Bharat EduAI Stack
Sources
Related news
Gnani.ai launches sovereign AI stack for Indian enterprises
Gnani Launches Artha Sovereign AI Stack for Indian Institutions
India and Nepal Advance Inclusive Multilingual Language AI Collaboration
India’s Vice-President Launches Gnani Artha Sovereign AI Stack
Meta Launches Muse Voice Transcribe with Indian Language Support
BharatGen CEO Rishi Bal Details India’s Sovereign AI Strategy
