Researchers at the Indian Institute of Science (IISc) in Bengaluru have released SraVaani, an open-source voice AI model that recognises speech in 65 Indian languages and dialects. Developed by IISc’s SPIRE Lab with ARTPARK and support from Google, it converts spoken words into text across 10 scripts and automatically identifies the language without user selection. SraVaani supports 20 scheduled Indian languages and 45 regional languages and dialects, including Garo, Angika, Chakma, Kokborok, Tulu, Bundeli and Bajjika. IISc said several covered languages are not officially supported by existing speech-recognition systems, and together these languages are spoken by about 25 crore people, based on the 2011 Census. The model is freely available on Hugging Face under an MIT licence. IISc said SraVaani performed on par with leading Indian systems on commonly supported languages, with a bigger advantage in underserved languages; on Garo, it recorded a word error rate of 9.5%, compared with 69.4% for the next-best system tested. Prasanta Kumar Ghosh, professor at IISc and principal investigator of Project Vaani, said the aim was for voice AI to work for every Indian. SraVaani was trained using data from Project Vaani, which recorded more than 31,000 hours of speech from 156,000 people across 165 districts in 28 states, with participants