SraVaani

UPSC Reference Category: GS Paper III – Science & Technology | Artificial Intelligence | Indian Languages

What is it?

SraVaani 1.0 is an open-source automatic speech recognition (ASR) model developed under Project Vaani.

Key Features

  • Recognises speech across 65 Indian languages and dialects.
  • Based on FastConformer architecture.
  • Approximately 430 million parameters.
  • Uses a hybrid TDT–CTC decoder.
  • Released under the MIT licence.
  • Developed using India's Vaani speech corpus.

Importance

It can help address the digital divide created by the dominance of English and a few major Indian languages in AI systems.

Applications:

  • Voice assistants
  • Education
  • Digital governance
  • Accessibility
  • Translation
  • Speech-to-text.

Key Pointer

SraVaani → AI → Speech Recognition → 65 Indian languages/dialects → Project Vaani