SraVaani — Multilingual Indian Speech Recognition Model

SraVaani — Multilingual Indian Speech Recognition Model
1. What is SraVaani?
•    SraVaani is a multilingual Indian automatic speech recognition (ASR) model developed by researchers at IISc’s SPIRE Lab in collaboration with ARTPARK and Google.
•    It is designed to convert spoken Indian languages and dialects into text, particularly those that remain poorly supported by existing speech-recognition systems.
•    The model has been trained on 65 Indian languages and dialects, including more than 40 languages that are not officially supported by many existing speech-recognition systems.
•    SraVaani is freely and publicly available on Hugging Face under an MIT licence, allowing wider use and further development.
Why is SraVaani significant for India?
•    India has a highly diverse linguistic landscape, but most existing speech-AI systems perform well primarily in a limited number of widely spoken languages.
•    Many regional and non-scheduled languages therefore remain underrepresented in speech-to-text technologies, creating a digital divide based on language.
•    SraVaani seeks to address this gap by extending speech-AI capabilities beyond the languages traditionally prioritised by commercial technology platforms.
•    The model could potentially benefit around 25 crore people, based on the 2011 Census population whose languages are not adequately handled by existing speech-recognition systems.
Language Coverage
•    SraVaani covers 20 scheduled languages and 45 regional languages and dialects, giving it a broad pan-Indian linguistic reach.
•    Its coverage includes 19 languages from Northeast India, 16 from eastern India, nine from western India, eight from northern India, six from southern India and five from central India, along with English and Sanskrit.
•    The model supports relatively underserved languages and dialects such as Garo, Angika, Chakma, Kokborok, Tulu, Bundeli and Bajjika.
•    By including such languages, SraVaani expands speech-based AI from mainstream Indian languages to regional, indigenous and non-scheduled linguistic communities.
Technical Significance
•    Automatic Speech Recognition (ASR) refers to technology that converts human speech into written text using computational models trained on spoken-language data.
•    SraVaani's significance lies not merely in the number of languages covered but in its attempt to provide usable speech recognition across India's linguistic diversity.
•    Its performance demonstrates that AI models can be developed to handle languages for which training data and existing technological support are comparatively limited.
•    The model is particularly notable for its performance in languages that are poorly represented in conventional speech-recognition benchmarks.