Why in news?
On 15 November 2026, the Ministry of Tribal Affairs highlighted the expansion of Adi Vaani. Its note reported text translation in nine tribal languages as of 15 October. A working group has also discussed developing it into a National Tribal Language Repository.
The platform
Adi Vaani supports translation, language documentation and learning. The ministry launched it on 1 September 2025. A consortium led by IIT Delhi developed it with other institutions and State Tribal Research Institutes (TRIs).
Native speakers contribute words, sentences and voice recordings. Language experts and TRIs check this material before it is used to train models. AdiKosh is the platform's word bank.
What the language tools do
Automatic speech recognition (ASR) converts spoken language into written text. Optical character recognition (OCR) extracts machine-readable text from an image, such as a scanned page.
Machine translation converts text from one language into another. Speech synthesis produces spoken output from text. A speech-to-speech translation system can link recognition, translation and speech synthesis.
Why language data matters
A corpus is a collection of language material used for research or training. A parallel corpus pairs text in one language with its corresponding translation in another. These pairs supply examples of how meaning is expressed across languages.
For example, a set of Hindi sentences paired with their Santali translations can support translation-model training. Recordings paired with written transcriptions can support speech-recognition training.
Constitutional provisions
Article 29(1) gives any section of citizens in India the right to conserve its distinct language, script or culture.
Article 350A concerns mother-tongue instruction for children from linguistic minority groups at the primary stage. States and local authorities should endeavour to provide adequate facilities. The President may direct a State to help secure these facilities.
Conclusion
The repository proposal would bring together corpora, primers, dictionaries and community stories. Primers are introductory resources for learning a language.
Sources
- Tribal Affairs Ministry: Adi Vaani progress and repository proposal, 15 November 2026
- IBM: speech recognition and audio-to-text processing
- IBM: OCR and machine-readable text
- IBM: machine translation and language-pair training data
- AI4Bharat: parallel sentence-pair corpora
- Legislative Department: Constitution, Articles 29(1) and 350A