IIT Madras Launches Speech Tool for 26 Indian Languages

Researchers at IIT Madras, through its BodhanAI initiative in collaboration with AI4Bharat, have introduced *Indic-Transcribe*, a new automatic speech recognition system designed to serve India's vast linguistic landscape. The tool can transcribe speech in 26 Indian languages alongside English, positioning it as one of the more comprehensive multilingual speech technologies developed domestically.

What sets *Indic-Transcribe* apart is its attention to the practical realities of how Indians actually speak. The system is built to recognise diverse regional accents, work across multiple scripts, and, notably, handle code-mixed conversations where speakers blend languages such as Hindi and English or Tamil and English within the same sentence, a common feature of everyday Indian speech that many existing tools struggle to process accurately.

The tool will be made publicly accessible on the BodhanAI website starting September 5, allowing users and developers to test its capabilities firsthand. By tackling the challenge of linguistic diversity and mixed-language usage head-on, *Indic-Transcribe* reflects a growing effort within India's AI research community to build language technologies that are genuinely representative of the country's multilingual reality rather than optimised only for a single dominant language.