The Ministry of Electronics and Information Technology announced on 16 October 2026 that the Digital India BHASHINI Division, with the Indian Institute of Science (IISc), Bengaluru, has developed Text-to-Speech (TTS) models for the Sourashtra language. These AI models can generate speech from text in both Tamil and Sourashtra scripts using male and female voices. The project aims to improve digital access and preservation for the Sourashtra-speaking community, which numbered around 2.48 lakh people in the 2011 Census. The associated speech corpus is now available on the AIKOSH platform for researchers and developers.
Key points
- New AI models generate Sourashtra speech from both Tamil and Sourashtra scripts
- Includes male and female voices trained on 10,000 and 11,400 utterances respectively
- The SPICOR TTS 3.0 Sourashtra Corpus (28+ hours of audio) is available on AIKOSH
- Supports a community of approximately 2.48 lakh people (2011 Census)
When it applies
The document does not state when it takes effect.
Who it affects
- Sourashtra-speaking community, students, language learners, researchers, and developers