What is 'Bhashini'? India's AI-based Language Platform Which PM Modi Has Offered to SCO Countries Bhashini, an acronym for BHASa INterface for India, also known as Digital India Bhashini, was launched by PM Modi in July last year. It is India's Artificial Intelligence (AI)-led language translation platform.
You're reading Entrepreneur India, an international franchise of Entrepreneur Media.
"We would be happy to share India's AI-based language platform- Bhashini with everyone. This can become an example of digital technology for inclusive growth," shared PM Narendra Modi when addressing Shanghai Corporation Organization countries in a virtual summit on Tuesday.
The SCO countries currently consist of China, India, Iran, Kazakhstan, Kyrgyzstan, Pakistan, Russia, Tajikistan and Uzbekistan.
What is Bhashini?
Before ChatGPT by OpenAI became a household name worldwide in December last year, India had an epiphany just a few months before. The National Language Translation Mission (NLTM) was announced in Union Budget 2021-2022 by our Finance Minister Nirmala Sitharaman.
The initiative was launched as a result of a survey where over fifty per cent of respondents who did not have access to the internet shared that they would start using the web if the content and information were available in their native language.
NLTM's goal is to build a high-quality speech-to-speech machine translation system for major native languages, increase content in Indian languages on the internet in various domains, and create and nurture ecosystem players such as start-ups and government institutions to develop and deploy products and services in Indian languages.
Bhashini, an acronym for BHASa INterface for India, also known as Digital India Bhashini, was launched by PM Modi in July last year. It is India's Artificial Intelligence (AI)-led language translation platform. The project is also championed by the Ministry of Electronics and Information Technology.
<blockquote class="twitter-tweet"><p lang="en" dir="ltr">?️ <a href="https://twitter.com/hashtag/BHASHINI?src=hash&ref_src=twsrc%5Etfw">#BHASHINI</a> - To enable citizens to speak in their own language while talking to speakers of other Indian languages!<br><br>Visit <a href="https://t.co/U9y7T0sKL4">https://t.co/U9y7T0sKL4</a><a href="https://twitter.com/hashtag/AmritMahotsav?src=hash&ref_src=twsrc%5Etfw">#AmritMahotsav</a> <a href="https://twitter.com/hashtag/IndianLanguages?src=hash&ref_src=twsrc%5Etfw">#IndianLanguages</a> <a href="https://twitter.com/hashtag/BhashaDaan?src=hash&ref_src=twsrc%5Etfw">#BhashaDaan</a> <a href="https://twitter.com/hashtag/DiversityandInclusion?src=hash&ref_src=twsrc%5Etfw">#DiversityandInclusion</a> <a href="https://t.co/lN3PzIkC2o">pic.twitter.com/lN3PzIkC2o</a></p>— Ministry of Electronics & IT (@GoI_MeitY) <a href="https://twitter.com/GoI_MeitY/status/1564910887223275520?ref_src=twsrc%5Etfw">August 31, 2022</a></blockquote> <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>
The AI platform's purpose lies in acting as an orchestrator to unify and align "a large diverse network across government, industry, academia, research groups and start-ups to bring all their contributions into an open repository." Bhashini also aims to encourage non-English citizens to use the internet in their own language.
Bhashini is built on a dataset through a crowd-funding initiative called Bhasha Dan.
What is Bhasha Dan?
An initiative to crowdsource language inputs, Bhasha Dan calls upon the public to help build an open repository of data. The aim of this initiative is to create a large dataset for Indian languages, which will help to train AI models that will further help different stakeholders to create products or services.
A citizen can contribute to the dataset through four categories- Suno India, Bolo India, Likho India, and Dekho India.
Suno India- In this category, users are required to type what they hear in audio, and then they'll be needed to validate the text provided by other participants. The total participation in this segment is 1479.
The contributions made to it are in 12 languages, with 22 hours and 29 minutes of duration transcribed, and 3 hours and 51 minutes of duration verified. About 22 per cent of Suno India's target has been achieved with only 26 hours and 20 minutes being done out of 120 hours.
Bolo India- In this category, users are required to input sentences in their voice, and then they'll be needed to validate the audio recordings provided by other participants. The total participation in this segment is 1501.
The contributions made to it are in 12 languages, with 44 hours and 47 minutes of audio recorded and 24 hours and 26 minutes of audio recordings verified. About over 50 per cent of Bolo India's target has been achieved.
Likho India- In this category, users are required to translate the prompted text into their native language. Additionally, they'll have to validate the translations provided by other participants. The total participation in this segment is 572.
The contributions made to it are in 39 language pairs, with 3096 of translations done, and 869 translations validated. Only 0.3 per cent of Likho India's target has been achieved. Only 3965 of 1200000 sentences have been achieved.
Dekho India- In this category, users are required to label the pictures provided, and they'll have to verify the image labels provided by other participants. The total participation in this segment is 597.
The contributions made to it are in 12 languages, with 4110 labelled images, and 444 validated images. Only 0.04 per cent of Dekho India's target has been achieved.
Recently, it was reported that Bhashini may soon power automatic transcription and translation of Kerala High Court proceedings in English or Malayalam.
