Context
Researchers at IISc Bengaluru, with ARTPARK and Google, have developed ‘SraVaani’, an AI speech-recognition model trained on 65 Indian languages and dialects. It aims to improve access to AI and digital services for under-represented Indian languages, supporting India’s linguistic diversity.
About SraVaani
- Developers: Indian Institute of Science (IISc) SPIRE Lab, in collaboration with ARTPARK and Google.
- Objective: Extends automatic speech recognition to several regional and non-scheduled Indian languages that remain underserved by existing speech technology.
- Licensing: Freely and publicly available on Hugging Face under a Massachusetts Institute of Technology (MIT) licence.
- Total Coverage: Trained on 65 Indian languages and dialects.
- Breakdown: Covers 20 scheduled languages and 45 regional languages and dialects.
- Sample Languages: Supports languages such as Garo, Angika, Chakma, Kokborok, Tulu, Bundeli and Bajjika.
- Population Reach: Potentially opens speech AI access to around 25 crore people as per the 2011 Census, whose languages are not properly handled by current systems.
Performance Benchmark
- Word Error Rate: Recorded a 9.5% word error rate on Garo.
- Comparative Performance: Compares with a 69.4% error rate for the next-best system evaluated on the same language.
Why is this article important for UPSC?
- Understanding the SraVaani speech recognition model, which is an important Science & Technology development in Indian language AI.
- Understanding institutions such as the Indian Institute of Science (IISc) and ARTPARK, relevant to India’s AI research ecosystem.
- Understanding challenges in multilingual speech AI for underserved Indian languages, relevant to Science & Technology and linguistic diversity.
In this article, you will understand:
- Developers and objective of SraVaani
- Language coverage and licensing details
- Performance benchmark results
Q. With reference to SraVaani, an Indian multilingual speech technology initiative, consider the following statements:
1. SraVaani was developed by IISc’s Speech and Language Technology (SPIRE) Lab in collaboration with ARTPARK and Google.
2. SraVaani is trained on 65 Indian languages and dialects, covering 20 Scheduled Languages and 45 regional languages and dialects.
3. SraVaani is released under a proprietary licence restricted to research use only.
Which of the statements given above is/are correct?
(a) 1 and 2 only
(b) 2 and 3 only
(c) 1 and 3 only
(d) 1, 2 and 3
Answer: (a) 1 and 2 only
Explanation
• Statement 1 is Correct: SraVaani was developed by IISc’s SPIRE Lab, in collaboration with ARTPARK and Google, to advance speech technology for India’s diverse linguistic landscape.
• Statement 2 is Correct: SraVaani covers 65 Indian languages and dialects, comprising 20 Scheduled Languages and 45 regional languages and dialects.
• Statement 3 is Incorrect: SraVaani is openly available on Hugging Face under the MIT licence. Therefore, it is not a proprietary, research-use-only model.